News of Timm
FSE 2010 accepts two papers
In the Future of SE workshop:
- "Software is Data Too" with Andrian Marcus;
-
"On the Shoulders of Giants" with
Earl Barr, Christian Bird, Eric Hyatt, and Gregorio Robles.
- Lounge Room
- DVD player remote (2*AAA)
- Camera (4 * AA)
- WII remotes (2*2*AA)
Video remote (2*AAA) - Bedrooms:
Alarm clocks (2*AAA) - Study
- Desk clock (1*AA)
- Remote mouse (1*AAA)
- Other
- Camera
- Air-Con remotes (3*2*AAA)
- Fire Alarms (4*wide 6V)
- Study
- Two years ago, a radiator froze over and ruptured, sending gallons of water down into the bedroom roof. Took us weeks to get dry.
- We are pulling up the kitchen roof this year- to fix a pesky leak from the plumbing under the guest bathroom.
- And last night, after 2 weeks of snow, the drains froze and burst and now we have drip-drip-drip inside the front door.
- Google search
- Google analytics
- code highlighting
- and a comment system.
- IEEE ASE 2009 (give one paper, one poster, one keynote at the doctoral symposium, present one tutorial);
- Consult with some Sydney academics;
- Meet with the Chinese Academy of Science,
- Work on an NSF panel,
- Then go to Ithaca to meet with some colleagues on a new STTR project.
- 2/10's of a mile past Limestone General Store turn on to Limestone Hill Palace Road.
- Stay on the paved road with the double yellow center line.
- Go 3.7 miles to the Palace of Gold.
- Morgan Run Road: beautiful lake-side park.
- Kentuck Knob amazing architecture and sculptures.
- Moundsville Prison : not for the faint hearted.
- Palace of Gold : can anyone say "Almost Heaven"?
- Friendship Hill : a 200 year-old mansion, 20 minutes from M'town.
- The PROMISE'10 web site;
- Data mining via scripting;
- Agent-oriented programming;
- Awk programming;
- The WVU CBR project;
- the WVU AI lab.
- The colors of the grass were beautiful... exquisite shades of gray green and gold strolling into each other.
- The road rolled like a river.
- The iPhone played music- and such music. Never heard those little tinsel sounds before in the background of ColdPlay's "when I ruled the world". Pause. There are little tinsel sounds in the background of "when I ruled the world", right?
- Stopped at Brigeville's Eat'n Park and I fell in love with everyone there. It was so bright and cheerful and munchy- everyone scoffing the Sunday buffet. After rapidly inhaling two plate fulls of roast turkey, scrambled eggs, and bacon, I read my Kindle and sipped Decaf coffee. The coffee was so gooood that I took a cup of take out. Funny thing is that once I got the to car, I forgot all about it.
- Meet the usual gang of 40 people smart people I see at every conference. And we said smart and smart-ass things to each other.
- Saw some amazing things like 1980s AI (a PROLOG-based QSIM) being applied to software process engineering. Yeah AI!
- The conference series I've been building over the last few years (with Gary Boetticher, Tom Ostrand, Guntheur Ruhe) went very well. Both days were good but day one was GREAT- most of the talks revisit old results in our experiments repository. We had new blood arrive to offer us lots of exciting and challenging ideas. My talks went well.
- At the dinner for that conference, I sat with Barry Boehm (he invented something called "software engineering" back in the 1960s). Between us, we filled up the paper tablecloth with enough bubbles and arcs to fill a textbook. At the end, I folded it all up and Barry said "now, just send that to the National Science Foundation and it'll be your grant".
- Several times, certain people starting tenure track came up to me and asked, quite pointedly, to work with me. It seems that some of the senior bulls in the herd are advising people "go work with Tim, he'll get you published".
- A senior NSF guy heard some of my ideas and told us that they could work as a special kind of NSF grant. So I spent the rest of the conference stitching up a deal with Microsoft/Turkey/USA/UK to explore generality in defect detectors.
- I didn't like Vancouver. I'd moved there in 2000, all browned-off from the NASA experience and disappointed++ with the whole North American experiment. We stayed 18 months, camped high above the city in a lonely little concrete turret. Helen got nowhere with her thesis and the winter really got us down. When NASA rang and said "come back as a research chair", we could not move fast enough to get outta there. So this trip was hardly a nostalgic visit.
- My professional association with Gary came to an end. He's off making stock market money now so our interests have now diverged. We make sure we ran away from everyone else one night and went to a private dinner, just the two of us. Afterwards, outside the restaurant, there was a poignant moment: I turned right to go back to the hotel and Gary said he was going the other way. Silence. Then manly hug and farewell.
- I don't think my talk went well. Strange to say, I'd given almost the same talk seven days earlier and it went fine. But now, it was last thing last day and me and the audience were tired++. At the end, there was "any questions?" and everyone sat dumb. Then my friends, bless 'em, jumped up and asked animated questions so I didn't look like a dumb ass.
- to be pursued by the ultimate hunter killer;
- then caught;
- then held close (but not too close) by this guy who might never, ever, go away.
- Real world data collection is more like ambulance chasing that bus driving. The old DoD model of rigorous process control just breaks down in the modern era of distributed software development. Rather than lament lack of formal process, we should adapt our learning methods to handle the idiosyncrasies of our data.
- Static code attributes are a wide and shallow well- easy to get to the bottom, very hard to get much further. Our learners may have learned all they can learn from these attributes.
- The only way up is sideways. My data miners have struck a performance ceiling and the only way up is to change the performance target.
- The performance ceiling is very close- we can exploit that. Rather than large-scale automatic methods, it may be more productive to explore human-in-the-loop interactive learning strategies.
- 4 ESE journal submissions;
- 3 TSE journal submissions;
- 1 SPIP journal submission;
- 1 IJTAI journal submission;
- 1 ICSE paper submission
- In 2003, the world had 35 trillion dollars of savings.
- Then the financial expansion of India and China hit. By 2007, that number had grown to 70 trillion.
- Sometime this century, the world population will peak and, after that, we'll move from the era of economic growth to economic sustainability.
- And when that happens, we'll have to match growth rates with the size of the safe havens.
- Unlimited growth will no longer be an option cause there'll be no safe place to park the products of that growth.
- Start at Morgantown.
- Go to Spruce knob, the highest part of the state. If you want to see the wild WV, stand there and gaze.
- Then on to the Greenbank radio telescope. Acres of radio technology magic, Listening to the skies
- After that, the next stop is The Greenbrier. For centuries, rich politicians fled the summer heat of Washington for an elaborate mountain retreat. And i do mean elaborate. $4000/night. But visitors can wonder the grounds and walk thru stores and museums dating back to the 1780s
- Then go to Skyline drive: way up in the hills, tacked on to the mountain tops, great great views.
- Have a look at Harper's Ferry. Its wonderful: a stone village on the banks of two mighty fine rives. Lots of cool bed and breakfast places.
- On to, Gettyburg's the place where the tide turned in the american civil war. 3 days, mega-causalities. I wasn't that interested in going there but once i got there- a very humbling and magnificent place.
- Back to Morgantown.
- Place the camera's flat square battery in a little cradle and plug it into wall socket #1;
- Plug notebook #1 into wall socket #2; then take two USB cables and plug the video recorder and iPod #1 into notebook #1;
- Plug the notebook #2 into wall socket #3;
- Take another USB cable and connect iPod #2 into notebook #2;
- Charge cell phones #1 and #2 with wall sockets #4 and #5.
- Connect our wireless router into the (a) the motel Ethernet and (b) wall socket #6.
- Everywhere you go, find human bones.
- Then take any friend or relative and have them drawn into the lethal web of the bad guys.
- Climax when our heroine arrives at some crucial moment when everyone's lives are on the line. At this point, she does something plucky but usually not very effective.
- End with crappy detective acknowledging that our heroine is needed to rid the city of crime.
- Repeat in the next book.
- Tips (for Beginners)
- Tricks (for Programmers)
- Tools (for Data Mining)
- Tasks (for Supporting the Business)
- Tests (of Different Tools)
- Traps (for the unwary)
- To do (What's Next?)
- Team (About Us)
- Theory (Only if you Care)
- Pre-soviet: ornate, festive.
- Soviet era: clean, crisp, sterile, no decadent Western decoration.
Post-soviet:
proud buildings glad to be buildings. The centerpiece of our conference center was a five stories of mezzanines
around a glass elevator shaft with glass elevator cabs. The elevator motors, top of the shaft, where left
exposed and panted yellow just to draw your attention to them.
Sat May 24 16:15:45 PDT 2008
Indiana Jones, sucks
How can you screw up an Indiana Jones movie? Inject adrenaline into the veins of the audience, hang them out over the edge, give them a crescendo at the end, etc etc.
For shame Mr Spielberg. And what a waste of a perfectly good Cate Blanchet.
Sat May 24 16:12:19 PDT 2008
Begin 2 days of denial and anger
TSE just rejected a journal paper submission. And so it begins. Denial, anger, bargaining, depression, acceptance. Oh, then death. God I hate that cycle.Sat May 24 16:06:50 PDT 2008
NSF accepts research proposal
Three years funding: Automated Quality Prediction-- Exploiting Knowledge of the Business CaseSat May 24 16:04:05 PDT 2008
Paper accepted to ICSM'08
Automated Severity Assessment of Software Defect ReportsSat May 24 16:02:30 PDT 2008
Regularities in Learning Defect Predictors
Paper submitted to ASE'08. Collecting large consistent data sets for real world software projects is problematic. Therefore, we explore how little data are required before the predictor performance plateaus; i.e. further data do not improve the performance score. In this case study, we explore three embedded controller software, two versions of an open source anti-virus software (Clam AV) and a subset of bugs in two versions of GNU gcc compiler, to mark the regularities in learning predictors for different domains. We show that only a small number of bug reports, around 30, is required to learn stable defect predictors. Further, we show that these bug re- ports need not necessarily come from the local projects. We demonstrate that using imported data from different sites can make it suitable for learning defects at the local site. Our conclusion is that software construction is a surprisingly uniform endeavor with simple and repeated patterns that can be discovered in local or imported data using just a handful of examples.Mon Apr 7 19:02:30 PDT 2008
Accidental gay activist
Once upon a time in 1979, I accidently lead the Gay Mardi
Gras. At the time, it seemed kinda fun. Bit of a lark,
really. But, man, I soooo missed the point.
The year before, the same march became the defining moment in the Sydney Gay Pride movement. A night time march went terribly wrong. The cops attacked the gays, systematically bashing them. Dozens were dragged away to a nearby police station and beaten, all night long. Their friends gathered outside, passing the hat, trying to raise ball for those they could hear, screaming inside.
Of course, I knew none of this. I was just at a club, dancing and prancing the night away with my mates. Four straight fag-hags, just here for the party. Someone said "the Gay Mardi Gras is outside" so we went to look. It wasn't the carnival that it is today- just a block of blokes wearing black leather, carrying a sign or two- protests in remembrance of the police abuse from the year before. We wondered through the crowd and when it moved off, we sort of got carried along.
Then the T.V. camera crew showed up and all the gays boys screamed "what if my mum sees me!?". The crowd parted like Moses at the Red Sea, opening up to reveal me and three straight friends in the middle of the crowd.
With everyone else gone, we stepped into the gap. Holding hands, we walked at the front of the mob, leading the chant, guiding them for a mile or two through the center of Sydney. It was all great fun, cause we did not know that we risked a beating with each step.
I'd like to think that if I knew the risks I would have done it anyway. But I didn't. So while I'm proud of the service I gave to the gay community, that service was so uniformed that I earned almost negative karma points that night.
Sat Apr 5 19:21:34 PDT 2008
Creation is a predictable, controllable process that can be engineered
Based on the success of the software engineering, I want argue that creativity can be engineered and, routinely, is done so in a commercial setting. To do so I must demythologize "creation". Rather than focus obsessively on that "a ha!" moment when the world spins on a dime and the new paradigm falls like an apple in Newton's garden, we should look at the conditions that preceded creation and increases the probability of finding shiny new apples.
To any individual, creation may feel like a special process but, as Hari Seldon will one day observe, given a large enough community, then the outputs of that communities creativity and individuality can be predicted.
Sat Apr 5 12:31:29 PDT 2008
Why I think about Turing Machines
CS and UI is the art of building glass walls to hide users from the computational cliff. Hopefully, most users are blind to how narrow is their treacherous path. But without the language of the Turing machine, you are blind to certain restrictions to writing, unable to explain why your system is broken, unable to avoid that problem in the future.
Sat Apr 5 12:28:02 PDT 2008
Automated Severity Assessment of Software Defect Reports
New paper, submission to ICSM'08Fri Apr 4 20:34:34 PDT 2008
2CEE, A Twenty First Century Effort Estimation Methodology:
New conference paper: International Society of Parametric AnalysisWed Apr 2 18:37:51 PDT 2008
Ok, now I get it
Just say a virtualized LINUX installed as a windows app inside a virtualization box. Their package
manager was linked to the CSEE package manager and this virtualized app
goes looking for updates to the package base every hour. So this stuff auto-updates on
a regular basis.
What made the demo very impressive was that the last 2 weeks I had to commission 10 new computers. Take what the vendor gives, change that, alter that, etc etc etc.
The same task with virtualization technology would have been a 10 minute install on each machine. Easy.
So now I get it.
Thu Mar 20 19:41:36 PDT 2008
Optimizing Requirements Decisions With KEYS
New paper, PROMISE 2008Wed Mar 19 21:08:56 PDT 2008
Implications of Ceiling Effects in Defect Predictors
New paper, PROMISE 2008Wed Mar 19 21:06:00 PDT 2008
Comparing Design and Code Metrics for Software Quality Prediction
New paper, PROMISE 2008 workshopWed Mar 19 21:04:20 PDT 2008
New talk
Learning Near Optimum Inspection PoliciesSat Feb 9 20:13:48 PST 2008
NSF workshop: Software Engineering & Creativity
Creation is a predictable, controllable process that can be engineered
(a paper for the 2008 NSF workshop on creative writing and SE)
by Tim Menzies
Feb'08Based on the success of the software engineering, I want argue that creativity can be engineered and, routinely, is done so in a commercial setting. To do so I must demythologize "creation". Rather than focus obsessively on that "a ha!" moment when the world spins on a dime and the new paradigm falls like an apple in Newton's garden, we should look at the conditions that preceded creation and increases the probability of finding shiny new apples.
To any individual, creation may feel like a special process but, as Hari Seldon will one day observe, given a large enough community, then the outputs of that communities creativity and individuality can be predicted. As commented by the Australian Band, the Whitlams in the their song "One in a Million":
Some say, "love it comes once in a lifetime," That's enough for me She was one, one in a million There's five more in New South Wales.
- 1/3 in management and planning
- 1/6 in development
- 1/4 in unit test (testing all parts in isolation)
- 1/4 in system testing (testing all parts, in combination)
- The average wait time for the muse is M days
- The number of required poems is P poems of Quality Q per day
- And better and better quality poems are less and less likely (and some exponentially decay constant k)
- Then the number of starving artists you need to go shiver in the forest is...
- Brooks (1995)
- F. Brooks, "The Mythical Man Month", Addison Wesley, 1995.
- Florida (2005):
- R. Florida, "The Flight of the Creative Class: The New Global Competition for Talent". Collins Books.
- Glass (2007):
- R.L. Glass, "Software Creativity 2.0" developer.
- Menzies (2006)
- T, Menzies and Z. Chen and J. Hihn and K. Lum, "Selecting Best Practices for Effort Estimation", IEEE Transactions on Software Engineering, Nov, 2006, Available from http://menzies.us/pdf/06coseekmo.pdf
- Menzies (2007):
- T. Menzies and O. Elwaras and J. Hihn and M, Feather and B. Boehm and R. Madachy, "The Business Case for Automated Software Engineering", IEEE ASE, 2007, Available from http://menzies.us/pdf/07casease-v0.pdf
- Raymond (2001):
- E. S. Raymond and B. Young, "The Cathedral and the Bazaar: Musings on Linux and Open Source by an Accidental Revolutionary", O'Reilly & Associates, 2001. A short form is available from http://catb.org/~esr/writings/cathedral-bazaar/cathedral-bazaar/
- AI- a Modern Approach
- Graham's On LISP book
- Seibel 's Practical COMMON LISP
- Test case assessment: assess candidate tests by their distance to the PTs. This wold not suffer from the N^2 clustering cost; rather M*N where M is small and just |PT|.
- Classification: assess each PT by their classification certainty; i.e. after moving all old tests to their nearest member of PT, then look for members of PT with much higher frequencies of one class over the rest. At least for one data set I experimented with, there were under a dozen frequency PTs and only 3 certain PTs.
- Test case generation: generate new tests from PTs that have few/no candidate tests near them
- Test case prioritization: rank candidate tests by their distance to the frequently visited certain members of PT
- Anomaly detection: at runtime, recognize anomalies by new observations that are far away from frequently visited PTs. This would be fast, assuming very few certain frequent PTs.
- Danger case detection: a special case of anomaly detection where we raise an alert if a new observation is not close to a frequency certain PT
- Repair: to avoid anomalies or danger cases, impose the delta between the new observation and frequent certain PTs. Rank those repairs by their effort (smaller efforts are better).
- The (human engineers) time it takes to construct models of increasing breadth and depth
- The (computers) time it takes to evaluate those more elaborate models
- The (human decision makers) time it takes to understand the results of those evaluations
- Discover, define, and apply new scientific principles, engineering processes and methods This work augments and extends software systems engineering with AI methods.
- Bridge and transcend CISE disciplines by encouraging collaborations that might include information science and engineering. Our proposal draws from the disciplines of systems engineering, AI, computational analysis, and process/product modeling and simulation.
- Science and engineering of software for real-world systems This work will use real-world software systems models under development at JPL.
- Model-based design lies at the juncture of human involvement (the capacity to provide the insights, guidance, preferences, etc) and computation (the capacity to store and explore voluminous amounts of detail). JPL spacecraft, and their embedded software, provide an exemplary proving ground for demonstrating and deploying this proposals novel techniques in order to head off a looming set of limits that threaten to stall the entire model-based engineering paradigm. The success of the proposed work in this highly visible context will pave the way to widespread adoption of these advances.
- Also, empirical SE needs real worlds cases studies: this work would create a corpus of models that a broad community of researchers could use as the basis of experimentation and exploration.
- Finally, the projects educational activities will include involvement of undergraduate and graduate students in research, the development of new and the enhancement of existing courses. We have a good track record and remain strongly committed to bringing research and educational opportunities to students from the under-represented groups.
- 10 pages: conference paper
- 8 + 6 pages: two workshop papers
- 10 pages: technical report for NASA
- a 20+ pages: NSF proposal
- 100% said "The instructor showed enthusiasms when teaching";
- 9 of the 11 respondents answered "Learn?" with "quite a lot"
- Very few missed classes (10/11 missed 3 classes of less), even though there were no exams in the class.
- my new Chumby;
- the laptop, to configure a bedtime Chumby channel;
- the iPod, to try out on the Chumby speakers (it sounded great),
- and the new bluetooth ear piece for the PDA (very good pick up).
- All books take zero seconds to read.
- No new science fiction, since all new ideas are now completely copyrighted;
- After the Chinese take over, all English was banned.
So English professors had nothing to teach,
- Exception: Composition 101 was mandatory for everyone so it was the only subject taught at university. And all the faculty had to teach it, all the time.
- Computer Science fared no better:
- Once the Nasquanians landed and gave out their BioTech, all old-fashioned digital geeks became unemployed.
- All Republicans became refugees and moved to Australia. There, they took over the government, sold the hospitals to Hong Kong, and spend their time threatening New Zealand with nuclear destruction.
- Middle Earth was discovered, not in New Zealand, but in a suburb of the Australian town of Wagga Wagga. New Zealand's economy was now in tatters since all tourist now skip the land of the long white cloud.
- All snow machines were broken (so no more snow, ever).
- We live in Australia, with all the spiders and snakes.
- No more chocolate.
- No more squirrels.
- No more coffee.
- No more beaches.
- We have kids.
- soccer injury: one
- Cath: nil
- Data Discretization Simplified: Randomized Binary Search Trees for Preprocessing
- Monday Dec 10, Brown Conference Room, 1pm
- Thesis: 14.4 MB (pdf)
- Slides from defense: 2.2 MB (pdf)
- Date:
- December 3rd, 2007 (Monday)
- Time:
- 2PM
- Venue:
- ESB G64 (Ground floor ESB)
- Abstract:
-
It is important to have a good cost estimate in order to budget a new project. Unfortunately, software effort estimation methods are often inaccurate. Molokken and Jorgensen report that 60-80% of the time a software project will overrun its estimate by an average of 30% [98]. Furthermore, most estimates do not describe the uncertainty of the estimate [52,54,118]. In addition, each source of uncertainty, as described by Kitchenham and Linkman [68], has yet to be represented.
Previously, Jorgenson has argued that most effort estimation is done manually [49, 52]. However, manual methods have difficulty sampling the space of uncertainty [54]. In this thesis, the design principles of an effort estimation tool currently being deployed at NASA's Jet Propulsion Laboratory are discussed. This tool, called 2CEE, represents an approach to effort estimation that provides greater interaction of the cost analyst with the estimation model. The approach places an emphasis on representing estimation uncertainty. This thesis describes how 11 of Jorgensen's 12 expert judgment best practices may be automated in a model, 7 of which are demonstrated in 2CEE. Thus, the distinction between manual and automated methods for effort estimation is questioned. Instead, an alternate ideology is proposed where neither manual nor automatic estimation methods dominate, but rather each augments the other.
- denial:
- "say what???!!??";
- anger:
- "reviewers are idiots!";
- bargaining:
- "maybe I can write to the editor?";
- depression:
- "it was a crap paper, they are all out to get me";
- acceptance:
- "oh well, maybe I can rewrite it".
- Elvis
- Boatshow in Monaco
- Mercedes Benz Muzeum
- Grappige Borden (Funny Signs)
- Ruby on Rails 101 - Presentation Slides for a Five Day Introductory Course
- The original 1973 survey listed their numbers as 2,300.
- A 1984 report lists their numbers as 160
- Subsequent reports are less detailed, listing their population as "rare", "rare", and "rare" in 1988, 1990, and 1996.
- One book published in 2001 reported numbers between 100 and 500 and added that a pipeline was being built dangerously close to their tiny habitat. Also, burning of forest areas near caves is thought to be the dominant, long-term threat to the species.
- A 2006 article in Cambridge University Press reports a 2002 finding of bumblebees in caves 250 miles away from where the bats were first found. This new group, plus the old, would bring the total population to 1500.
- I can't find anything more recent than that 2002 survey.
- Firstly, the worst person you can marry is another academic. That can only end one way- buried deep in the two-bodied problem.
- Secondly, and this is important, the best person you can marry is another academic. They'll be the only ones who understand the pressure of publication and tenure.
- At day 3 lunch, every seat was still full.
- There was coffee right outside the sessions- no need to sneak off . Enough comfy chairs to sit around and shoot the breeze.
- Poster session- it was humming. Dozens of people circle round and round for hours and hours.
- During the talks- no one in the corridors shooting the bull. Everyone in the sessions.
- In the sessions, nearly all seats full.
- Talking to lots of folks about new ideas- and them offering new insights and tools on what you can do to handle problem X,Y,Z a little better.
- Noon:
- Gave my talk. Well rehearsed. Lots of positive comments afterwards.
- 2pm:
- Exit, stage left. Simple train ride to airport
- 4pm:
- Upgrade to business class for $40? You bet!
- 6pm:
- Touch down, Pittsburgh. Strange time in car park walking down long lines of cars looking for mine. Happiness when, in the distance, you see your car's lights winking back when you hit the clicker.
- 8pm:
- Morgantown. Eat'n park milkshake. Daggy bliss.
- 9pm:
- Getting embraced from drunken neighbor on porch. Calling out for help to her husband. "You're on your own", he called back as he sucked back on his cigar.
- 10pm:
- Playing with cat who hasn't seen me in 10 days. Loving and cuddlely at first, then she went into pissy-scratchy cat mode (you never write, you never call).
- 11pm:
- Lots of zzzzz
- Manual methods such as "read the biggest thing first/last"
- Traditional data mining methods such as those advocated by author and those deployed in NASA-funded inspection tools.
- A new data miner called "WHICH"
-
Saturday: - A domestic day. Shopping. IKEA. Assembling IKEA furniture (once more with the allen key). Dinner with friends.
- Sunday:
- More with the allen keys (fixing up the chair that I'd built wrong last night). Wrote 2 papers: a text mining report and the TSE submission with the Turks
-
Monday - Morning- trying to call the NSF about possible projects. Afternoon- driving back to M'towm. Evening- meeting with Brian W. at a bar to talk about the NASA stuff
- Tuesday:
- Morning with the dean (talking about NASA).
Afternoon talking to the faculty (about NASA). Then teaching.
Then much snoozing in the evening.
- Wednesday:
- Morning- talking with the WVHTC and Galaxy Global about research ideas. Afternoon- talking to students. Evening- writing up grant ideas from GG.
- Thursday:
- Morning- trying to do NSF stuff. Afternoon- students, teaching. Evening- much snoozing. And Helen arrived from Baltimore 'bout 10pm. Yaaah!
-
Friday:
-
8:30am meeting on the national archives stuff. Rest of the day trying
to do an NSF grant (got 3 pages out- like crapping bricks). Lunch at Blue Moose,
walked back through a gray overcast autumn day between trees of gold and orange.
- Saturday:
Slept in! Hooray! Spend the morning doing not much. Went swimming. Then
pumpkin carving.
Wrote some blogs- Provides an integrated set of COCOMO models
- Can be used to systematically analyze strategic and tactical policy decisions
- Can be tuned/calibrated with constraint sets instead of traditional historical records.
- BEFORE: model shelling to file1, learner reading file, learning shelling to file2, simulator reading file2, adjusting inputs to model, repeat
- NOW: model and learner compiled into the one executable. all data passed in RAM
- BEFORE: tar3
-
NOW: restrict to binary values, add a counter to each attribute range.
After each run of the model, add the score from that run to a counter
for that range. every N=100 runs,
- resort the ranges by those counters
- "fix" the top N/100 ranges.
- let the other attributes float free
- repeat
- BEFORE: model was a Visual Basic tree thing that worked out the relative importance of each node by calling up the tree
- NOW:: partially evaluate and cache the importance of each leaf. cross-compile to a c function. call make to combine it with the locally guided search
- Over- and under- sampling wins nothing
- Supposedly better Bayes reasoners do no better, and sometimes a little worse.
- Some discretization adds a little, but not much
- Some marginally improvement from boosting, but it takes much much longer to run
- Dozens of other learners have a performance that is statistically insignificantly different to my older methods
Mon Aug 23 19:19:52 PDT 2010
TV interview
WBOY-TV, Aug 9,2010: Scientists are developing better filters to make digital data useful.
Sat Aug 14 15:10:53 PDT 2010
New NSF Award
$500K, with Andrian Marcus, at Wayne State University
Better Comprehension of Software Engineering Data
The amount of data generated during the development of today?s software systems is staggering. It includes the source code, developer e-mails, bug information, testing results, analysis data, process information, requirements, etc. The size and complexity of this information make it impossible for developers to reason about it.
Data mining techniques are a common solution to extract what is relevant to developers and managers. The success and quality of these software projects depends on the software engineers? ability to customize generic data mining algorithms to specific software engineering data. This project will produce tools and techniques that will allow software developers and managers to easily customize and apply data mining techniques to a variety of software engineering problems. Such solution will become more practical and will help many existing approaches to migrate from the research lab into industry.
Under represented categories of students will participate in this research. The project will enhance the existing software engineering curriculum and facilitate the inclusion of data mining solution in the repertoire of future software engineering practitioners and researchers.
Specifically, the project will improve the state of the art solution to three important software engineering tasks: concept location in software, software defect prediction, and development effort estimation. The project will produce an algorithm customization methodology and a framework that will be instantiated for a variety of combinations of data mining algorithm x software engineering task x software system data. The customization problem is framed and addressed as an optimization problem. The resulting customization agent will assist the software engineering user in efficiently selecting the best configuration, which includes a set of algorithms and their parameter values, customized for a particular task and software system. All tools and methodologies will be empirically evaluated in academic and industrial settings.
Sat Aug 14 15:09:17 PDT 2010
The PROMISE'10 conference accepts two papers.
On the Value of Learning From Defect Dense Components for Software Defect Prediction by Hongyu Zhang, Adam Nelson, and Tim Menzies
Case-Based Reasoning vs Parametric Models for Software Quality Optimization by Adam Brady and Tim Menzies.
Sat Aug 14 15:06:58 PDT 2010
Automated Software Engineering journal accepts two papers
Stable Rankings for Different Effort Models by Tim Menzies, Omid Jalali, Jairus Hihn, Dan Baker, and Karen Lum
Defect prediction from static code features: current results, limitations, new approaches but Tim Menzies, Zach Milton, Burak Turhan, Bojan Cukic Yue Jiang, and Ayse Bener
Sat Aug 14 15:03:35 PDT 2010
Software Practice & Experience journal accepts paper
Sharing Experiments Using Open Source Software
Adam Nelson, Tim Menzies, Gregory Gay
Sat Aug 14 15:00:52 PDT 2010
IST Journal accepts paper
Practical Considerations in Deploying Statistical Methods for Defect Prediction: A Case Study within the Turkish Telecommunications Industry
Ayse Tosun, Ayse Bener, Burak Turhan, Tim Menzies
Sat Aug 14 14:57:51 PDT 2010
Webcast: Wayne State talk
Finding local lessons in SEFri Feb 26 04:43:45 PST 2010
Paper accepted to IEEE TSE
Genetic Algorithms for Randomized Unit TestingThu Feb 25 14:35:25 PST 2010
Wet fluffies vs warm fluffies
My cat thinks it can create its own reality, just by growling at it.
We've been house bound for 2 weeks now (lottsa snow). But Lucie (the cat) still wants to go outside for her ritual checking of the perimeter.
So we crack open the back door and flurries of snow blow in and land on her fur. There is nothing dry out there- no place to prowl.
She growls, and hisses, and waits. Nothing happens, except she gets more snow on her back. More growling, more hissing, more snow on her back.
Finally, she stalks off in a huff, back into the house. I'm sure she thinks the snow is worried that it has offended her.
As for me, I suspect the snow doesn't really care. But I won't tell Lucie.
Fri Jan 15 06:20:46 PST 2010
New draft paper
Exploiting the Essential Assumptions of Analogy-based Effort EstimationThu Jan 14 05:14:06 PST 2010
My body, the car
Godley And Creme warned us: our bodies are like motor cars that wear out:
My body is the car that I've been driving Around for forty nine years My body the car Slowly burning out the rubber and stripping the gears My body the car
Well, they were right. The results are in- my gall bladder ejection fraction is 11% (normal is 30 to 70). The doctor's don't seem worried- they just say its some low grade chronic cholecystitis (yawn). But they have
plans to make it better. Well, Billy Bragg advises against those plans:
The temptation To take the precious things we have apart To see how they work Must be resisted for they never fit together again
So I have my own plans. Bring on the low fat diet. Kind of like filling the car with the ultra-grade at the gas station. Milk shakes? Get behind me Satan!
Thu Jan 14 05:00:53 PST 2010
Emacs SVN mode = good
I've been using the svn mode in emacs. At first, I thought it had a bug: it made me write change comments for each committed file, one at a time.
Then I realized that this bug is a actually a feature. I no longer write "bunch 'o stuff" when I commit changes to dozens of files. Instead, I pause and reflect on why I changed each one (guess what: lots of changes are due to documentation fixes!).
So now I'm wondering about writing a blog reflecting on all this code work. Every day, write about the changes from 30 days ago. This lets me not say stupid things about something that seemed useful n Tuesday, but by Friday I'd undone it all and dumped it in the scrap heap. That way I can write sage wisdom today without having to eat crow tomorrow.
Wed Jan 13 05:44:38 PST 2010
Assaulted, by batteries
Decided to buy shares in lithium. Batteries are the future. Did a head count. My house in crawling in batteries:
And I'm sure I've missed some.
And when the lithium runs out, what next: Helium? Everyone driving round in cars with squeaky voices like Mickey Mouse?
And then what? Hydrogen cells? Could be a BAAAD idea. Oh, the humanity.
Tue Jan 12 05:06:15 PST 2010
Bad case of acute syllabi
So I can't read a calendar. I thought this week was last week. Or, should I say, that next week was the week after.
Perhaps I'm not being clear. I thought that school started NEXT week, not today. Which means I must now write multiple syllabi.
"Multiple syllabi" sounds like a disease. Eeek! My doctor looks
concerned. He tells me:
-
"I'm sorry Mr Menzies but you have have a bad case of multiple active
syllabi. You will suffer for 14+ weeks.
Your symptoms will be tiredness,
inability to explain even the simplest ideas,
and repetitive exasperation syndrome as you repeat the same thing over
and over and over and over again."
He tells me that there is no cure- that I will surely suffer a relapse in the fall.
Mon Jan 11 06:56:45 PST 2010
Raining, inside the house
I've never had such a house. It keeps raining, inside!
Hell, I've seen The Last Wave, I know what happens next. Everyone: time to buy scuba gear!
Sun Jan 10 06:31:44 PST 2010
Give us this day, our daily hack
I am revealing in a morning hack. Take all that
screwing around time. Do it for 1 hour first
thing in the morning. Maybe even get some cool
stuff done. But burn it off so the rest of the
day can be spent happily in administrivium.
Fri Jan 8 18:53:44 PST 2010
And the snow comes down
Weeks of snow. Wonderful.
Fri Jan 8 18:48:55 PST 2010
21-century: looking good
I love the modern world.
Last night I had insomnia, so in
one hour I added to a website
I'm working on:
All this was done using 3rd party free tools and less than 40 lines of javascript. (Note: search is broken till the Google spiders pass my way.)
Fri Jan 8 18:34:19 PST 2010
The web is closer than my hard drive
When I lose a file, I used to search for it using OS/X's "searchlight" tool.
However, I copied most of my files to an on-line publicly readable web-based version control system. So an alternative to searching for it in my 500GB hard drive is to ask Google to search for it amongst it's petabytes of data.
And the winner? Google search,hands down. This means that the web is now a faster place to explore than my local computer.
Fri Jan 8 18:29:53 PST 2010
Scripting 'r us
I've realized something. What I like doing is scripting. Scripting is what is what blew my mind in Beach Street Bondi in 1986 when I sat all day in my little room, hacking on my Taiwan AT clone.
And all I've ever done since is tried to find some business model where I can fund my scripting addiction.
Fri Jan 8 18:27:15 PST 2010
Building an IDE
I'm building an IDE for a language I like
and it is so NOT what I thought it would
be.
Occasionally I run to the cupboard to grab a coat , throw open the door, and find a whole vast room in there full of details to be retired.
Today I realized that I was going to spend a LOT of time importing all my old one-line functions into this IDE framework .
Fri Jan 8 18:22:58 PST 2010
Busy!
November and December was
extraordinarily busy. I mean busy2*100.
In those 2 months, I travelled to Auckland, Sydney, Beijing, Washington, New York to attend:
Then it was time to do end-of-semester grading, the annual P&T reviews, and 2 NSF grant proposals (CISE, Dec 17).
At the end of all that, my wife made a request that i actually was a husband for a while. So we went to New York to eat and do museums and some shows.
This got us home in time for Xmas and all that entails.
Rest assured that i have NO travel plans now for some months and months and months and months and ...
Fri Jan 8 18:11:27 PST 2010
New Journal article
Explanation vs Performance in Data Mining: A Case Study with Predicting Runaway ProjectsSun Nov 29 13:34:10 PST 2009
Slides of my Tsinghua University talk
Wed Nov 25 05:31:45 PST 2009
Slides from my ASE'09 talk
Understaning the Value of Software Engineering Technologies
Sun Nov 22 23:08:58 PST 2009
Invited to join editorial board
Automated Software Engineering JournalSun Nov 22 22:56:48 PST 2009
Automated SE Journal accepts paper
Finding Robust Solutions in Requirements Models
Tim Menzies with Gregory Gay, Omid Jalali, Gregory Mundy, Beau Gilkerson, Martin Feather , and James Kiper.
Solutions to non-linear requirements engineering problems may be "brittle"; i.e. small changes may dramatically alter solution effectiveness. Hence, it is not enough to just generate solutions to requirements problems- we must also assess solution robustness.
The KEYS2 algorithm can generate decision ordering diagrams. Once generated, these diagrams can assess solution robustness in linear time. In experiments with real-world requirements engineering models, we show that KEYS2 can generate decision ordering diagrams in O(N2 ). When assessed in terms of terms of (a) reducing inference times, (b) increasing solution quality, and (c) decreasing the variance of the generated solution, KEYS2 out-performs other search algorithms (simulated annealing, ASTAR, MaxWalkSat).
To appear ?January 2010.
http://menzies.us/pdf/10keys.pdf
Mon Nov 9 18:55:24 PST 2009
Invited to Tsinghua University, Beijing
Dr Hongyu Zhang, (School of Software, Tsinghua University) has kindly invited me to China to give a talk at his university.
Download the poster for the talk.
Mon Nov 9 19:04:53 PST 2009
New talk: Data Mining and SE
Here are the slides from my recent talk at UtDallas.
Data Mining: The Missing Link in Empirical SEView more documents from timmenzies.Mon Nov 9 19:12:17 PST 2009
Friendship Hill
Albert Gallatin build a mansion near Morgantown, WV.
Who was Albert? Well, he was a the longest-serving United States Secretary of the Treasury; founded the New York University and arranged maps and finance for the Lewis and Clarke expedition. He also helped balanced the US budget, which meant the USA could afford the Louisiana Purchase.
His house is amazing- acres of manicured lawns and a beautiful old house up on a hill.
The house and grounds are open to the public nearly 365 days a year. There are deer that wonder around the grounds- very nice in the snow.
On the way back, stop at Apple Annie's for PIE!!!
View Larger MapThu Oct 1 21:09:28 PDT 2009
Palace of Gold
Can you believe it? There is a massive massive massive Hare Krishna palace near Moundsville, WV.
The Internet map directions are somewhat incorrect, towards the end:
View Larger MapThu Oct 1 20:42:01 PDT 2009
Moundsville Prison
This one might want to remain a secret. But, for the strong of heart, a visit to the infamous Moundsville Prison is a very intense experience. This place had hangings, riots, etc,etc. It was shut down after the courts ruled in inhumane. One hour of touring inside and you understand why.
On the way back, just to neutralize the horror of the jail, go see the Palace of Gold.
View Larger MapMoundsville Prison is yet another Morgantown Secret.
Thu Oct 1 20:22:18 PDT 2009
Secret Morgantown
There are all these pretty places around Morgantown that are drop-dead gorgeous- but its taken me a decade to find them.
Here is list of amazing afternoon/ day trips around Morgantown:
Thu Oct 1 20:09:49 PDT 2009
Kentuck Knob
Want some Frank Lloyd Wright architecture?
Kentuck Knob has drop-dead brilliant architecture and amazing guided tours. Not as spectacular as Falling Water, but nowhere near as crowded. Also recommended is the walk back from the house down the hill to the car park, via "The Meadow" (contains 400 2-d red tin solders, a 12' high chunk of the Berlin Wall, and other cool++ stuff).
View Larger MapKentuck Knob is yet another Morgantown Secret.
Thu Oct 1 20:06:10 PDT 2009
Morgan Run Road
North end of Cheat Lake is a manicured lake front with easy walkways between lake and forest. Fantastic at sunset. Really nice drive down there.
View Larger MapMorgan Run Road is yet another Morgantown Secret.
Thu Oct 1 20:02:11 PDT 2009
I used to drink optimistically...
Now I drink misty optically.
Here's the Friday night drinks crowd. We gather to reassure each other that, yes indeed, we are marvelous and wonderful. And, gosh darn it, we are too!
Thu Oct 1 17:45:49 PDT 2009
Should I cheer this?
Some Australian accused me of being a dirty old man posting these pictures of cheerleaders to my FLICKR account.
I was about to protest that this was decent, wholesome, all American girls. Then, for a second, my cultural conditioning slipped and I looked again at these writhing tanned trimmed limbs over mini-skirts short enough to show their breakfast. All to the tune of some guy blowing his horn.
Now I'm confused...
Thu Oct 1 17:30:22 PDT 2009
PROMISE'10 web site
Now live.Thu Oct 1 10:02:54 PDT 2009
No news is good news?
I used to blog a lot here. Now, I am spread out over five sites:
Give me this day my daily web site .....
Thu Oct 1 09:58:19 PDT 2009
Data mining toolkit
A WVU scripting environment for data mining experiments.Thu Oct 1 09:56:10 PDT 2009
Agent-oriented SE at WVU.
New web site on agent-based programming.Thu Oct 1 09:54:24 PDT 2009
Awk.info
New web site on all things awk.Thu Oct 1 09:53:42 PDT 2009
Case-based software engineering
CBR at WVU.Thu Oct 1 09:51:50 PDT 2009
Talk at ICSM'09
Sonia Haiduc presented our ICSM'09 paper.Thu Oct 1 09:49:56 PDT 2009
Modeling Intelligence Lab
We have a new site for CSEE AI research.Thu Oct 1 09:48:41 PDT 2009
WVU Seminar
Mathew Valenti kindly asked me to speak at the WVU CSEE graduate seminar series (Oct 5). I'll post the slides here when I get them done.Thu Oct 1 09:47:21 PDT 2009
Invited Talk, UT Dallas
Eric Wong kindly invited me to speak at UT Dallas CS colloquium (Oct 23). I'll post the slides here when I get them done.Thu Oct 1 09:44:54 PDT 2009
Program committees
I've been asked to serve on the PCs of ESEM'10, MSR'10 and INFOS'10.Thu Oct 1 09:43:08 PDT 2009
Paper accepted to ASE 2009
Understanding the Value of Software Engineering TechnologiesWed Jul 29 16:23:52 PDT 2009
Paper accepted to ASE 2009
Assessing the Relative Merits of Agile vs Traditional Software DevelopmentWed Jul 29 16:16:13 PDT 2009
Paper accepted to ISSRE 2009
Cost Curve Evaluation of Fault Prediction ModelsWed Jul 29 16:09:02 PDT 2009
Yue Jiang receives Ph.D.
Incremental Development and Cost-based Evaluation of Software Fault Prediction ModelsWed Jul 29 16:08:12 PDT 2009
Invited to program committee
Mining Software Repositories, 2010.Wed Jul 29 16:05:33 PDT 2009
Dieting = tripping?
Sshhh... don't tell your minister but dieting is a buzz.
I'm on day 8 of South Beach Phase1 (no carbs, no sugar, no booze, no caffeine) and I just drove 1.5 hours on a sunny Sunday M'town to P'burg. The experience can only be described in terms Dr. Leary or Mr. O'Rourke might understand.
Soon I climb into a little metal bullet and try to fly higher into the sky. But that does not seem possible.
(Five minutes later.)
Oh god, I'm so depressed. Food! Give me food! McDonalds' hamburgers dripping in fat. A McHeartAttack Smoothie that is 10 parts sugar and 1 part some crappy flavor that I won't even taste as I snort the whole thing up my nose. Ice Cream... plates and plates of ice cream that I eat so fast most of it falls on my shirt. Then, for dessert from my dessert, I will pull my shirt off and thoroughly lick it clean.
(Five minutes later)
Hello birds, hello sky, may I join you as you fly by?
Sun Jun 7 10:16:59 PDT 2009
Dieting, again
Welcome (again) to day three of South Beach Diet, phase one. For the next 11 days, no fruit, booze, caffeine, or bread. Its just fine... I drift through the day occasionally taking pills for the coffee withdrawal headache. Don't feel hungry but if you put a pizza in front of me, I know I'd just inhale it.
Yes, I know that this is not a long term solution (the effect that diets do not, in the long term, work is sadly very well documented). But last semester was round and now I am round and something needs to trim a few pounds.
Tue Jun 2 09:16:13 PDT 2009
Shakespeare hates your emo poems
So says Gregory Gay, voice of the nation.
Sun May 31 10:00:07 PDT 2009
Paper accepted to ICSM'09
On the use of Relevance Feedback in IR-based Concept Location
Sun May 31 09:36:10 PDT 2009
The Zen of Gardening
Today we took all the bricks out of the garden path, weeded, laid some sheeting (death to weeds), then put
the bricks back. Total time = 3 hours. Net movement of bricks = zero. But we felt better when we were done.
One strange thing. Initially, we had N bricks on the path. Then, after much pulling and pushing, we had relaid N-6 bricks. The leftover six bricks now mock our revised geometry.
Thu May 28 17:18:17 PDT 2009
New grant
STTR (phase 2) from DoD, with Grammatech. Starts Sept'09. Details to follow.
Tue May 26 17:14:42 PDT 2009
Paper accepted to the SPIP journal
Accurate estimates without local data?
Tue May 26 16:48:02 PDT 2009
Paper accepted to PROMISE 2009
Can we build software better and faster and cheaper?
Tue May 26 17:05:20 PDT 2009
Paper accepted to PROMISE 2009
How to Share Experiments
Tue May 26 17:07:38 PDT 2009
Paper accepted to ICSP 2009
On the Relative Merits of Software Reuse
Tue May 26 17:06:48 PDT 2009
Student evals: cs472(AI)
Comments from 21 students.
instructor's clarity = 4.62 / 5; teaching effectiveness = 4.8 / 5; overall rating = 4.81 / 5.
"I really enjoyed the material we learned in the lectures. I wish we had time to learn more. I really enjoyed the relaxed environment that the teacher created while doing lectures. He was funny and really enjoyed what he was teaching, which helped me enjoy the subject and learn more. The projects were challenging, but they help me learn a lot. "
"Probably one of the best teachers I've had as a student at WVU. "
"Good grief, I have never had such an enthusiastic and passionate professor. Keep it up and never let your students get you down. "
"Honestly - one of the best professors I've had at the university. If you didn't understand the subject matter, the professor had no problem what so ever offering help as long as you asked for it. "
"(format t "i love lisp") "
"I really enjoyed the material we learned in the lectures. I wish we had time to learn more. "
"Dr. Menzies is an amazing professor -- he's very passionate about AI. The passion makes the class much more enjoyable. It is, without a doubt, my favorite part of the class."
"YAY"
"There is very little hand holding (which can be a good thing), so you have to keep up with the material or risk losing direction. "
Tue May 26 16:31:49 PDT 2009
Student evals: cs572(advanced AI)
Comments from 6 students:
instructor's clarity= 4.17 / 5; teaching effectiveness= 4.33 / 5; overall rating= 4.5 / 5.
"Dr. Menzies is a very good professor. He always gives us challenging things to work on, as well as encouraging us to go beyond whats required for the class to try and further our careers. He is available by appointment at almost any time, as well as before class, and with his office hours."
"Always tries to motivate the students by giving real-life examples of AI. Also allows the students to draw their own conclusions rather than giving them a strict set of guidelines. "
"Excellent course. It was very demanding but Dr. Menzies was very fair in his grading and expectations. Overall it was a very rewarding experience and I genuinely enjoyed Dr. Menzies as a professor."
"The team grading is unfair because everyone's grade can be lowered if one team member does not know something. I agree that the team should have their grade lowered if the project is not up to par, but only the team member who answers a question wrong should loose points. "
Tue May 26 16:22:04 PDT 2009
Student evals: cs310(Programming languages)
Responses from 35 students
instructor's clarity = 4.31 / 5; teaching effectiveness = 4.11 / 5; overall rating = 4.11 / 5.
No written comments.
Tue May 26 16:40:00 PDT 2009
Student evals, cs791(search-based SE)
Comments from 3 students.
instructor's clarity = 4.67 / 5; teaching effectiveness = 4.67 / 5; overall rating = 4.67 / 5.
No written comments.
Tue May 26 16:38:23 PDT 2009
Kindles rock
I used to be the guy who read. I read everything, all the time.
Lately, last few years, not so much. Sooooo busy on tenure track technical stuff.
But that has changed, thanks to my Kindle. I bought it for Helen, did a test configure, glanced at my first download, read some, read some more...
Two hours later I thought "Helen is NOT getting this as a gift". So I did the only honorable thing. Now Helen and I take our two Kindles (a.k.a. the twindles) everywhere.
Tue May 26 16:12:56 PDT 2009
The semester is over... finally!
This has been a HUGE semester. For years I've had the NASA buffer going and I only taught one grad class/semester. Now, that's all gone and I have a standard faculty load. Officially that is 2-1 and some committee work. But...
This term I taught 4 classes including one large 3rd year class that I had not taught before. Also, I got assigned to the high-effort promotion and tenure committee. The P+T stuff was NOT planned- I was stuck overseas and got a Facebook message saying that I had been appointed to P+T (thanks to the efforts of one Dr. Tim McGraw... who I know owe, big time).
I got by by a few admin tricks: two classes shared the same lecture series; one class was a seminar series that meet in my office Friday mornings; and my marker (Adam Nelson) did a great job of handling many of the subject details.
Still, I am exhausted and my "vacation" for this week involved exploring my inner couch potato.
Tue May 26 15:56:38 PDT 2009
A little Awk goes a long way
You may have been wondering why this site has been so static of recent times. My web obsession of late has been awk.info.
Awk is a decades old computer language- hardly bright and shiny life Ruby of OCaml or C# or.... But dude, it works. When I am prototyping ideas, this language is the best. I keep having attacks of other languages (Ruby, Python, Lisp, Smalltalk,...) but in the end, its dear old Awk that I keep returning too.
Most people see Awk as a language for writing two line scripts- buts its much more than that. At awk.info I've collected examples of Lisp interpreters in Awk, entire OO languages coded up as Awk pre-processors to "C", massive great word processing and macro language toolkits, etc etc.
Also, the language has another advantage- there is barely anything in it. No advanced data structures, pointers, memory management, etc etc. So if I explain an Awk program to someone, that explanation is not about the code. Rather, it quickly becomes about the thing the code is doing. And I like that- lets me use Awk to talk about other things.
Tue May 26 15:47:49 PDT 2009
Maria and Val: R.I.P.
I don't even know how to write this one. Last month I learned that two of the alpha-females in my nursing group from Jan'78 had died. Strangely, they did so within a month of each other. After Val died, Maria kept saying that she could feel Val tapping her on the shoulder. Their funerals, I am told, were very elaborate (700!!! people attended Maria's).
I remember them both very well and very fondly. They seemed so old when I meet them (nearly thirty... years... old....). It is to their credit that they were so kind and welcoming to a boy that was so young, he thought 30 was old.
We worked many a shift together and they were staunch comrades-in-arms. Together we fought and defeated tutor sisters and pressure sores. Maria and Val were both first-hand witnesses to the great misplaced swap incident of 1979 and the curious affair of the cholecystectomy (stories that I will share, another time).
I also remember, very well, the day they threw me to the ground, ripped off my shirt, and shaved off the four chest hairs I had at that time. There was nothing personnel in this act- they were just frustrated that their main mark (Tony) had gotten away. Their blood lust was unsatiated so they turned on the youngest fawn in the herd (and that would be me).
I hope I meet them again one day (for one reason, now I have more chest hair).
Tue May 26 15:28:12 PDT 2009
Conference overdose
Question: What does I C S E, I C P C, I C S P, and P R O M I S E spell?
Answer: Seven days of timm being "ON" at four back-to-back international conferences co-located in Vancouver Canada.
Some nice things happened.
But life is swings and roundabouts:
Tue May 26 15:19:28 PDT 2009
Worst use of technology, ever.
The new “Ocarina” app for the iPhone allows you to play a musical instrument by using the touch screen to tap out notes and the microphone to measure breath like a pseudo-wind instrument.
Since I am the resident experimenter of all things new (see my recent encounter with Mo’s Bacon Bar) I decided to stage a public performance of the Ocarina:
Sun Nov 30 08:58:27 PST 2008
I will never understand teenage girls
When I was a teenage boy, the biggest mystery in my life was teenage girls.
I thought I'd gotten over that but last night I went and saw Twilight. Here, the biggest thrill in a girl's life is
And I thought dimensionality synthesis in hyper-geometrical space was complicated.
Sun Nov 30 08:46:32 PST 2008
From jails to palaces of gold
You could not make this up. Saturday we toured hell and saw heaven.
Hell was the old jail at Moundsville, WV.
Now closed, this was an obscenity.
Conditions there were so bad that the State Supreme court ordered it closed, arguing that the facility was in violation of the Eighth Amendment (which prohibits cruel and unusual punishment).
And 15 minutes away was a slice of Heaven- a Hari Krishna
Palace of Gold (at Limestone, WV). Absolutely beautiful.
And the moral is? There is much to fear and much to celebrate about being human? People are as good and as bad as we can ever imagine? Dunno- but I'm thinking hard about it.
Sat Nov 29 21:15:09 PST 2008
Score: turkey=0, humans=1
Great Thanksgiving. Low key. Neighbors (Max and Lucy) came around and we boozed and stuffed ourselves all day.
Everything was fun- even watching a dog show where the dogs looked like toupees.
Sat Nov 29 21:10:44 PST 2008
"The class I've been waiting for."
I spent an hour reviewing my graduate data mining class with
the students last Wednesday. They pointed out some less-than-optimal
features of the class (I need some more intro material and I need
to change some aspects of the projects) but they also offered some
rather interesting accolades.
The class is hard going and one student said that if he'd had two such classes, he would not have made it through. So I asked them "should I make it easier" and they said "NO!" (resoundingly). Structure it better, they said, and prune of some of the superfluous documentation work, but don't dumb it down.
I asked "should I get rid of LISP?" and they also said "NO!". Functional programming rocks, said one student, adding that it changed his view of computation.
And I can think of no higher praise than from the alpha-programmer of the class saying that was "this is the class I've waited my entire degree for". He said that this was the only class, in his entire degree, that seriously challenged him and pushed him to the edge of his abilities.
I left that session with a long list of "to do"s but I was also grinning from ear to ear.
Mon Nov 24 08:34:35 PST 2008
Paper accepted to ESE Journal
Cross-Company vs Within-Company Data for Defect PredictionMon Nov 24 08:28:44 PST 2008
Paper accepted to ICSE'09
How to avoid drastic process change. 12% acceptance rate.Mon Nov 24 08:19:44 PST 2008
Managing Uncertainty in Value-Based SE
A presentation to the COCOMO forum 2008. October 28. 2008.
Managing Uncertainty in Value-based SEView SlideShare presentation or Upload your own.Fri Nov 7 14:27:08 PST 2008
Why Turing Machines?
Listen to the audio from my talk from the NSF Codeworks workshop, April 5 2008.
This written material from this talk is also on-line.
Fri Nov 7 14:07:49 PST 2008
Talk at Google: Learning Defect Predictors from Static Code Attributes; Oct 29 '08
Download slides (PDF)
For six years, I have worked on learning quality predictors from NASA data. Based on that experience, this talk offers the following lessons from the trenches:
Tue Nov 4 18:13:58 PST 2008
Parametric Analysis of a Hover Test Vehicle Using Advanced Test Generation and Data Analysis
Paper accepted to AIAA'09, 5-8 Jan'09, Orlando, FloridaTue Nov 4 18:10:17 PST 2008
Practical Considerations in Deploying AI: A Case Study within the Turkish Telecommunications Industry
New paper with Ayse Tosun, Burak Turhan, Ayse BenerSun Oct 12 02:57:32 PDT 2008
On the Relative Value of Cross-Company and Within-Company Data for Defect Prediction
New paper with Burak Turhan, Ayse Bener, and Justin DistefanoSun Oct 5 15:22:42 PDT 2008
Over-production?
By mid-October I will have:
Now, submissions does not mean acceptances and pride comes before a fall and I've been doing real bad lately with TSE.
On the other hand, nothing ventured, nothing gained.
Sat Oct 4 20:20:00 PDT 2008
Finding the walls of the fish bowl
Once upon a time, a few people knew where to park their cash so that
(hopefully) it grew into
more cash.
Then came the Internet and E-Trade and global knowledge. And now everyone knows all those places where we can park money, to make it (hopefully) grow.
So we we ran out of new places to safely park and grow money. And there was lots and lots and lots of money looking for someplace to park and grow:
So we parked the money in riskier and riskier places. Then guess what? Those risky places crashed and now everyone is running around screaming "the sky is falling, the sky is falling".
Now the market crash worries the hell out of me- threatens my savings, may cause depression in this country, etc. etc. But it seems to me that this economic event was inevitable; perhaps, even necessary:
So welcome to a brave new world where we've bumped into the walls of fish bowl, and the walls bumped back.
Sat Oct 4 18:15:39 PDT 2008
The world did not end
Last Tuesday was end of financial year. The world was going to end unless I got some reports written.
Heads up- in case you didn't notice- it didn't.
So you know how I spend Tuesday.
Sat Oct 4 18:06:09 PDT 2008
Accurate Estimation Without Calibration?
New paper with Barry Boehm, Steve Williams, Oussama Elrawas, Daniel Baker, Jairus Hihn, Karen Lum, and Ray MadachySat Oct 4 16:48:42 PDT 2008
Si, I am a dog
This dog, sleeping in the plaza at L'Aquila, seems to me to be quintessentially Italian.
It sleeps. It occupies a space but never doubts its right to that space. And the space it occupies is some tastefully selected region in the center of some graceful concentric brick work.
It ignores me as I walk towards it. If I was to disturb it, it would gaze at me, neutral, uncomplaining, waiting for me to declare why I dare to disturb its siesta.
If we talk, it might say "of course I deserve this space and if we have the espresso I might explain to you precisely why that is true, or perhaps which Italian car is best.
"But a first, I sleep".
Thu Sep 18 02:03:32 PDT 2008
Better than the real thing
I can't help noticing that this conference looks better on the video screen than
in real life.
Also (begin geek joke) this is a bounded model checker (only two copied of a recursive structure).
Thu Sep 18 01:09:10 PDT 2008
L'Aquila: a mountain Italian town
L'Aquila, 2 hours east of Rome, is not a popular tourist location. It is a small
bustling, living town. We work here, not entertain Americans.
This is a typical Italian town of tiny streets and even tinier cars.
The town has the ritual brassy nudes, cavorting in the water.
Even the toilets are stylish.
Of course, we have many Y.A.A.O.T.C. (yet another angel on the ceiling).
And what Italian mountain town is complete without a spooky hulking medieval fort?
Thu Sep 18 01:10:27 PDT 2008
Software Ecologies
Something has changed in software.
The Platonist and formalists are retreating (or retiring) and the ardent model-based reasoning people are singing another tune.
Its like last century's software was an oyster, alone in their little part of the beach, in full control of inputs and outputs. But today's software lives in some kind of jungle and must connect to a diverse ecology of other systems that slither and change all around. Before, the designer owned their world while now the designer must continually react and adapt to the changing environment around them.
Thu Sep 18 01:11:19 PDT 2008
Real-Time Optimization of Requirements Models
New paper with Gregory Gay, Omid Jalali, Martin Feather, James KiperSun Sep 7 18:50:43 PDT 2008
Automatic Estimation Techniques are Useful?
New paper with Omid Jalali, Jairus Hihn, Dan Baker, and Karen LumSun Sep 7 18:48:07 PDT 2008
How to Avoid Drastic Software Process Change (using Stochastic Stability)
New paper with Steve Williams, Barry Boehm, Jairus Hihn, Oussama El-RawasSun Sep 7 18:45:50 PDT 2008
Incremental Development of Software Quality Prediction Models
New paper with Yue Jiang and Bojan CukicSun Sep 7 18:44:26 PDT 2008
Using a Genetic Algorithm to Control Randomized Unit Testing
New paper, with James Andres and Felix LiSun Sep 7 18:42:44 PDT 2008
Best WV long weekend trip?
I was wondering, before the days grow shorter and cooler, what would be the best weekend trip in this part of the world.
Here's what i came up with. The following takes about four days. For a two day version, drop Greenbank, Greenbrier, and Skyline drive.
View Larger MapSun Aug 24 07:22:41 PDT 2008
Countdown to Prof. Timm
Welcome to the 3rd year tenure track grind. The next three months are exactly the time I put things in place to get promoted in 2.5 years.
And no one lets me forget that. Either by plan or accident, all these faculty people have been dropping into my office offering advice on how I can get promoted to full prof. at the end of my tenure track.
Normally tenure track is for assistant Professors and the end-tenure promotion is to associate.
But I am already an associate so the only way I can get a tenure pay bump is to go for full professor.
Sadly, due to the terms of my start up letter, none of my pre-2006 work counts to that goal. That is, those 29 journal pubs and $X million in grants are now forgotten, buried in the mists of time.
But I'm on track for promotion. I've been looking into other people's track record that got them promoted. I need 5 more journal pubs and 4 more completed masters and at, hopefully, one Ph.D.
That's a bit of a stretch in the next 2 years but it is possible. Right now I have revise-and-resubmits on three journal articles, one more written and ready to go, and one invited one for September.
Can't write anymore. Must get back to (re)writing journal papers.
Sun Aug 24 07:14:49 PDT 2008
Addicted to power
On this
road trip we carry numerous rechargeable devices- none of which use standard batteries.
Last night I forget to charge the camera so I write this while waiting for a little yellow light to turn green.
This gives me time to pondered all the electricals we carry.
Every night, we have to:
Most motel rooms aren't equipped for this kind of drain. As we plug in, we can see the lights of the town around us dimming.
Angry mobs of local villagers storm the motel carry torches (cause their electrical power has strangely disappeared). We fight them off with flashes from the camera- which means never letting that one run flat. Quick, where's that wall socket?
Sun Aug 3 09:00:57 PDT 2008
New Mexico beats Utah?
Utah rocks, but New Mexico does more.
In 2005, we went to Utah. Everyday, we found rock formations more amazing than the day before. Everyday, the camera filled up at 3pm. It was amazing.
But New Mexico is different to Utah. More diverse. Utah has rocks, New Mexico has so much other stuff. Each day, we've found something really cool. And not just rock formations.
Day 1, in Albuquerque, we found the Atomic Museum and were scared++ by
full scale life-size mock-ups of the atomic bombs dropped on Japan (and
the Trident missiles).
Then we cruised all the route 66 diners and signs that run through town.
Day 2, we rode the Sandia tramway thousands of feet into the air, then
drove to Santa Fe. There, we got a history rewrite.
Santa Fe is an old old old town.
America was not colonized east
to west. Long before cowboys, the Spaniards and the Mexicans had built
massive cities in the center of America.
Day 3 we went to Los Alamos. Along the way we found the Valles
Caldera (so beautiful).
Then we went on to Los Alamos to visit the Stepford Wives and Husbands that made the bomb.
Day 4 we went to Taos and found an Indian Pueblo village that is at least
500 years old (and the locals claim it dates back to twice that age).
Day 5 we went to Mills Canyon then drove for hours over grasslands rolling away to infinity. Also, we had a little more route 66 stuff at Tucumcari.
Day 6 was Roswell (yawn) then walking the desert around Carlsbad Caverns.
Day 7 we went down the Caverns (amazing), strolled the desert around
Gaudalope mountains, then drove along the smooth buttocks of America
(an eiriely deserted highway running over to El Paso).
Day 8 we walked over the sand dunes at White Sands.
That afternoon, we walked around
cool rockets at the National Space Museum, then checked out the National
Solar Observatory.
Still to come is the Very Large Array and what ever else is up the road.
So, in many ways, New Mexico is way more interesting than Utah.
Sat Aug 2 21:20:19 PDT 2008
New Mexico & monsoons?
Welcome to New Mexico. Desert sands. Atomic wastes. Searing heat. Drizzling rain. Can I have that last one again?
Ever heard of the New Mexico monsoon season? I hadn't. But every afternoon
here the clouds gather and the rain falls.
Sun Jul 27 21:37:45 PDT 2008
A nuke a day...
Welcome to the Atomic Museum, Albuquerque.
In the original version of this Nagasaki bomb, Helen stood nervously at its side. Then she demanded that the photo be deleted saying "I don't want to be seen anywhere near this thing".
What was scary was how small were these nukes.
The Hiroshima bomb
was tiny.
And an entire Trident missile (complete with dozens of warheads) was smaller
than an RV.
We asked for home delivery but they said they only had take out.
Sun Jul 27 21:31:19 PDT 2008
Temperance says "happy birthday"
For my birthday, all my dreams came true. Bones sent me a letter!
Hi TIm, Thinking of you! XXOO, TemperanceSun Jul 27 21:26:23 PDT 2008
Paper accepted to ASE'08
Using Simulation to Investigate Requirements Prioritization StrategiesSun Jul 27 19:48:03 PDT 2008
Paper accepted to ISSRE'08
Cost Curve Evaluation of Fault Prediction ModelsSun Jul 27 19:49:46 PDT 2008
Welcome to the 1980s
I'm so up to the minute. In the last week I got a rule-covering algorithm going (PRISM, 1987) and an iterative dichomization tree learner (ID3, 1982). Great breakthroughs in modern science? I don't think so.
Nevertheless, I am now the father of a 10-fold tree learner and I couldn't be prouder.
Mon Jun 30 07:12:01 PDT 2008
Relaxing with the murderers
Kathy Recihs' books are like my old kindergarten blanket. Familiar and very comforting when you curl up together.
The formula is simple:
Are we relaxed yet?
Mon Jun 30 07:08:07 PDT 2008
Summer rain
It rained all weekend. Gently falling, huge gobs of heat relief.
Like walking around inside a huge slow shower.
Very relaxing.
I've been fighting with LISP and data mining and when nothing was going right and my recursive tree builder was going crazy, I'd stare blankly out the window listening to a steady patter plop plop plop. Drops of water, not so much falling as drifting down.
Mon Jun 30 07:05:49 PDT 2008
Californication overdose
Gawd bless the 21st century. Don't watch any series week by week. I instead, buy them all on iTunes then eat them all in one weekend. We've been doing "24" each year at XMAS that way. Last year we got through "24" in 46 hours.
Just finished mainlining season one of Californication. I asked my wife "do I look as cool as David Duchovny?" and she patted my arm and said "yes dear".
Duchovny has developed this entire style of acting where he spouts rapid fire prose, without opening his lips more than a quarter of inch. If I tried it, Helen would say "what's that? stop mumbling and OPEN YOUR MOUTH". But when David (queue sound of heavenly choir), she leans forward closer and closer. She says its to hear what he's saying but I can't help noticing that it means she is getting closer and closer to.... David.
Mon Jun 30 06:59:34 PDT 2008
More forensic porn
Ok, Temperance Brennan and Bones is all well and good but she ain't the source
If you want the real forensic porn circus, it has to be Dexter. We spent last night doing 5 hours of the end of Season Two. And by the end we were jumping out of our chairs screaming. Exhausted, we went to bed and prayed we'd never be drug mules in the Florida everglades where Dexter prancing about is more dangerous than any crocodile.
And, of course, if you think that murder isn't a lark, then you should leave Dexter and go straight to Tony Hill (but only if you are not faint hearted).
Wed Jun 25 07:51:27 PDT 2008
Not reusable till reused
Been having much fun coding a data mining toolkit in LISP. Congratulating myself on
test-driven development (sort of), refactoring
all these reusable abstractions, then coding then directly in LISP.
The test-driven development stuff is a new work practice for me. No, I don't write the test first (not disciplined enough) but I now my work is peppered with little demos that illustrate what is going on. Good for teaching and regression tests. So that's a little success story.
As for the "reusable code", I've been here before. Thinking I'm doing great design. But the real test is the next thing I build. Is this supposedly "general" reusable framework really reusable? Will there be some quirk of the next thing that breaks the supposedly reusable framework? Time will tell.
And if I was betting man, I'd be better on more "refactoring" (newspeak for "patching bad ideas").
Wed Jun 18 19:42:04 PDT 2008
I love Temperance Brennan
In the last 15 days. I've eaten 3 seasons of "BONES".
58 episodes.
Serious corpse porn:
blood, gore, maggots, bodies melted to mush. Yummy.
My wife is in love with Hodgins. I am fond of Angela and would die for a date with Temperance.
But we both agree that (a)Zach's fate was so sad; and (b)we need to curse the writer strike for cutting short the season three arc.
Now we wait wait wait wait for season four.
Sat Jun 14 18:56:31 PDT 2008
Dieting blues
Welcome back to South Beach, phase 1.
Two weeks of no booze, no bread, and definitely no fun.
This is my first time back to phase 1 in two years. Last time, I don't recall feeling this crappy And crappy I definitely feel. It took till Friday morning (four days pf back pains, much glumness, loss of sleep) to burn off the enough of the required water, sugar, etc for my body to relax.
Now I'm sluggish, very couch-ish. Very very couch-ish.
Totally my fault: I choose to start it on a week with a surprisingly large admin load. Lots of buzzing around. Let me very very buzzed.
Got another week of it. Then Phase 2. Booze. Avocados. Dried apricots. Mangoes. Bagels. Wine. Yeah!!!
Sat Jun 14 18:14:53 PDT 2008
Why I Love Lisp
I've been having doubts about my choice of LISP for this data mining toolkit I am writing for the fall class.
Then, yesterday, I was writing the generic 10-way cross-way loop. And I realized that (a) there was a general pattern to that processing; and that (b) I could code it up as 20 lines of LISP, plus some funcall to particular functions passed in as arguments.
So, while other languages support this or that, LISP supports all things. IJAM (it's just another macro). Tee hee.
(defmacro square (x) (let ((x1 (gensym))) `(let ((,x1 ,x)) (* ,x1 ,x1))))Sat Jun 14 16:49:30 PDT 2008
A drive beneath me
Sunday, I was lost. Driving through empty streets in download Portland under
gray skies, threatening rain.
I found it... revealing. I came to this town in 2004, king of all I surveyed. Darling of NASA, the publication demon, fresh back from a great trip to Australia.
Pride cometh before the fall. Here, I was doing the soft money/adjunct professor life style. And I very quickly learned my (very small) place in the scheme of things. Adjuncts are a dime a dozen and rogue faculty without a uni position are quite irrelevant to the concerns of the uni. After 10 weeks here, I saw the list of candidates for 2 CS Portland faculty positions: 485 in all and I was lost on page 12 (candidate number 251).
This town took away my sense of entitlement to all the goodies of the academic life. In retrospect, I totally had it coming (never levered by soft money position into a faculty position, then the dot-gone bubble burst, as did my job security). So I shouldn't really blame this town- my own damn fault, really.
But, there are too many sad memories here. So , these last few days, I spent much of time in Portland grateful that I did not living in Portland anymore.
Thu Jun 5 21:56:51 PDT 2008
Playing with the rocket scientists
Another day, another day of data mining and software process theory at JPL.
Spent 3 hours today sitting under sun umbrellas at JPL geeking out. Pairwise testing rules. Data flows like wine. And data mining is cool.
Geek heaven.
Thu Jun 5 21:53:52 PDT 2008
The land that Internet forgot
L.A.: a primitive and savage land where the locals disfigure themselves in order to appease the gods (of fashion). Where the car is king and all travel is at 80 m.p.h. Including the Internet. My hotel has the wireless less connection from hell. Or, should I type, th e wir ele ss cooooooooonnnnn ee cti onn fro m hellll.
Thu Jun 5 21:50:48 PDT 2008
Greatest name of a function
I love LISP. Here's a function whose name is the formulae it implements.
(defun b^2/b+r (w before) (let ((best (wme-besth w)) (rest (wme-resth w)) out) (labels ((rank (b r) (/ (* b b) (+ b r))) (result (k v) (cons (float (rank v (gethash k rest 0))) k))) (sufficient w) (dolist (eg (wme-best w)) (inc eg best before)) (maphash #'(lambda (k v) (push (result k v) out)) best) (sort out #'> :key #'car))))Tee hee.
Tue Jun 3 17:29:59 PDT 2008
Are you PC?
I'm at the ASE program committee meeting. 40 people flying themselves in from around the world- how much would this cost if it was funded?
Such collegiality. It makes me want to sing (to the tune of WHAM's classic "wake me up before you go-go"):
Write me up before you go go, Cause I think your paper's so so, Fix this up to show that you know, Anything 'bout your field.
Tue Jun 3 17:20:06 PDT 2008
If clouds, then Pacific NW
Helen woke up, gazed out the hotel window at the low gray rainy clouds
that covered the sky, and said "there's an attractive picture". Yup, we're at a conference in the Pacific North West.
Oh, sure, the sun shines here, sometimes. It even blazed out one day
when all the conference people sat out in the sun and said nice things
about the weather in the state of Washington. The locals all laughed
and smiled- their suspiciously white skin beaming in the sun.
The conference is in Vancouver, otherwise known as Vantucky. We lived here for while in 2004 and I can't say that that I enjoyed it. Oppressive skies, strip malls stretching to the horizon. How can one city support need all that parking, all those malls?
Our conference hotel is in downtown Vancouver and it can be a ghost town down here. Some Australians at at the same conference and they asked "where are all the people?". All we could answer is "driving north, south on I5, racing to get to the mall".
I learned to hate pine trees in Vantucky. They suck the sunlight from the sky and walking beneath them is a very dreary experience. The locals are proud of this- they quote with relish from Lewis and Clarke's memoirs- the part when they sailed along the Columbian and were depressed by the incredible weight of massive trees flowing down to the shoreline in infinite numbers. "We have to cut these down", they said, giving birth to the Oregon timber industry.
We made some good friends here. Here's Scott who shocked me by saying "you don't let your students use
setf, do you?" and I am ashamed to admit that I do.
That night we went to our favorite Indian restaurant in the SE with Scott and Jen (his wife). We drank till we stopped then went to New Seasons to buy little cakey things.
Today, I go to Portland for breakfast. Portlanders are much prouder of their city their the Vancouver-ites and are found often striding around it. Either that, or escaping the rain in coffee shops where they... interact. All these caring and sharing over-educated middle class kiddies caring and sharing with each other, 24/7, 365 days/year. This is not a bad thing- Portland is ground zero of the international open source movement. But it makes this town so frigging ardent and self-righteous and smug.
One of my favorite sci-fi stories is set in
Portland. A hippy changes the world by dreaming. He doesn't have to
convince anyone of his vision, he just dreams it all up. Better yet,
our hero is not in control of his dreaming. Rather, he is told what
to dream by his shrink. The catch is that each dream is flawed so each
dream must be patched by another. An endless stream of failed best
intentions, all assuming that upper middle-class kiddies know
best. This story could only have been written in Portland, Oregon.
Speaking of irrational dreams that go horribly wrong, tomorrow I fly to California to move and shake with NASA people who, bless their hearts, have little funding and no control over who they give it to. This could be the last time I try to dip into the NASA bucket.
Sun Jun 1 08:17:25 PDT 2008
In the beginning...
In the beginning the project was without form, and void; and
darkness was upon the face of the accountant.
And the spirit of coffee moved upon the pen of the academic and he wrote stuff.
And NSF said "let there be funding" and, lo, there were green sheets.
And the accountants saw the green sheets, that it was good.
And the accountants divided the money into faculty salary and student salary.
And the promotions committee saw the green sheets, that it was good.
Fri May 30 09:02:15 PDT 2008
Can Data Transformation Help in the Detection of Fault-prone Modules?
DEFECTS 2008: Data preprocessing (transformation) plays an important role in data mining and machine learning. In this study, we investigate the effect of four different preprocessing methods to fault-proneness prediction using nine datasets from NASA Metrics Data Programs (MDP) and ten classification algorithms. Our experiments indicate that log transformation rarely improves classification performance, but discretization affects the performance of many different algorithms. The impact of different transformations differs. Random forest algorithm, for example, performs better with original and log transformed data set. Boosting and Naive Bayes perform significantly better with discretized data. We conclude that no general benefit can be expected from data transformations. Instead, selected transformation techniques are recommended to boost the performance of specific classification algorithms.
Wed May 28 17:13:51 PDT 2008
Wii, the people
And, of an evening, the humble village folks gather round the big-screen flat panel T.V. to bash the crap out of golf balls, bowling pins, and androgynous boxing partners.
Mon May 26 20:53:57 PDT 2008
A memorable day
Up at sparrow's fart doing spreadsheets. B'fast at Cooper's Rock.
Home for much hacking.
Dinner at Asian Garden with Eddy and Crystal.
Home (again) for dessert and much playing of the Wii.
Mon May 26 20:57:06 PDT 2008
Photos from Leipzig
Welcome to post-soviet East GermanySat May 24 17:15:55 PDT 2008
Photos from PROMISE 2008
An ICSE'08 conference.Sat May 24 17:14:37 PDT 2008
Let's write a textbook
24 people writing one textbook? Why not? Crowd sourcing rules!
I've started a Google Code project for my fall data mining students. One big old Latex file with hundreds of sub-directories, breaking up the whole thing into tiny little fragments. The idea is that student groups have to offer one or more of the following:
Sat May 24 16:33:18 PDT 2008
Summer resolutions
Gym, swimming, eating healthy, weight loss, not work obsessed, kinder to small children and animals, generous to charities, more time on the house and garden, get out of bed earlier, get more sleep, ....
(Memo: erase this page in 2 months time.)
Sat May 24 16:26:22 PDT 2008
Leipzig architecture
Just back from Leipzig, which used to be in East Germany. The architecture was... informative:
(Of course there is a minus-one error in this song. As of June 2007 the Australian state of New South Wales has a population of 6,889,100 so now there are nearly seven million candidate "loves of a lifetime", less the "was one" above, leaving six other candidates for life long soul mates.)
As with song, so to in industry. Robert Glass (2007) writes that software creation is a highly creative process. Yet in the software industry, it is possible to predict the time required to complete a creative processes such as those required to develop new software. In my other life as a computer scientist, I build predictors for the time required to complete software as well the number of bugs expected in the final product, see Menzies (2006,2007). This is important since the whole point of engineering is knowing how to surf cost-benefit curves. In a mature engineering discipline, it is possible to predict the quality of some future product based on trade-offs between process, product, and personnel decisions.
For example, here are the standard figures showing the relative importance of different factors on the effort (in months) to develop software:
1 Personnel/team capability 3.53 *********************************** 2 Product complexity 2.38 *********************** 3 Time constraint 1.63 **************** 4 Required software reliability 1.54 *************** 5 Multi-site development 1.53 *************** 6 Doc. match to life cycle 1.52 *************** 7 Personnel continuity 1.51 *************** 8 Applications experience 1.51 *************** 9 Use of software tools 1.50 *************** 10 Platform volatility 1.49 *************** 11 Storage constraint 1.46 *************** 12 Process maturity 1.43 ************** 13 Language & tools experience 1.43 ************** 14 Required dev. schedule 1.43 ************** 15 Data base size 1.42 ************** 16 Platform experience 1.40 ************** 17 Arch. & risk resolution 1.39 ************** 18 Precedentedness 1.33 ************* 19 Development mode 1.32 ************* 20 Developed for reuse 1.31 ************* 21 Team cohesion 1.29 ************* 22 Development flexibility 1.26 *************
Note that, as we might expect, individual capability is the single most important factor (see personnel/team capability). Which is not to say that combinations of other factors can't still dominate. For example, the benefits of using the most capability staff (3.53) are out-weighed by building an over complex product is too short a period of time ((2,38 * 1.63 = 3.88) > 3.53).
Along with many other researchers, I've shown that predictive models based on the above numbers are surprisingly accurate- which would be impossible if the creation process was as individual and idiosyncratic as is commonly perceived. And there is much support for this evidence:
Software engineering offers us other principles for engineering creativity. Some works of creation contain ambiguities- nuances that can engage and delight the viewer. These ambiguities are to be encouraged. Other ambiguities are not so benign. As soon as the creative piece executes then, before the viewer can enjoy the piece, the compiler must approve it. And compilers can be tricky. Imagine that your pen and paper argue with you saying "no sir, i don't like it. I don't like it at all" then stops you from mailing off your poem to a publisher because it won't compile, because it would interface with the commonIdeas (tm) v3.1 or because to distribute your poem you need newInk v2.1 and the company that marketed that one has gone out of business or has declined to port that library to mac os/x version 11.3.
And if compiler or licensing errors don't get you, never under-estimate
the power of your own stupidity.
- Debugging had to be discovered. I can remember the exact instant
when I realized that a large part of my life from then on was
going to be spent in finding mistakes in my own programs.
- Maurice Wilkes 1949Wilkes is point out that the time required to debug a system is alarming- over half the effort of building. According to Brooks (1995), the time required to build software divides as follows:
Brooks is cautioning us that we need to look beyond just the writing phase of software. So while we could focus of creativity in writing software (which is 18% of the task), we should remaining aware of the other 82% of the work. Many factors influence the environment around that 18% step. In several books, including "The Flight of the Creative Classes", Florida (2005) lists the social conditions that attract highly creative people to some local region. His triad is "technology, talent, tolerance". A moderately Bohemian environment, says Florida, is a necessary but not sufficient condition to attract talented software hackers to,say, Silicon valley. There, in an environment tolerant of new ideas, this talent can create innovative new technologies. Our 100 hour per week software hacker may never actually go an art gallery, lesbian bar, or join a drum circle but they want to feel that might be able to, if only they can get the next version of the software out the door.
Yet another advocate of engineering creativity is the international open source movement. The open source world takes the complete opposite approach to Florida. Rather than collect creative people in one place to work on a product, they take the product and ship it around the world to whoever and wherever creative extensions might occur. "Release early, release often" is the mantra amongst the open source community. Get parts of it going, get it out there, exploit the serendipitous effects seen when large groups use a new technology. Says Eric Raymond (2001), "The next best thing to having good ideas is recognizing good ideas from your users. Sometimes the latter is better." Strive for creative alternatives, says Raymond: "Any tool should be useful in the expected way, but a truly great tool lends itself to uses you never expected."
The open source experiment is both new and old. New, in the sense that it is remarkable that large, robust, innovative, complex software systems can be build by geographically disperse teams from across the planet. But old in the sense that it is well established that creation is a community process. The patent clerk Albert Einstein did not invented relativity in a shed on an isolated farm in the back woods of Tasmania (as proposed by the great Australian philosopher, Yahoo Serious). Rather, Einstein spent years reading everything he could of 19th century physics. And it was his detailed understanding of flaws with the old ideas that led him to conceive the new. So to create the next generation of creative thinkers, herd the newbies to university where they can study the oldies. "If you want new ideas, read old books" advises Ivan Pavlov.
If the British poet Ted Hughes was here, he'd object to this notion that creativity can be engineered, its products predicted in any precise quantitative sense, and that we can train a team to be some percentage more creative. In his 1957 poem "The Thought Fox", the paw prints of the thought fox are the marks left behind on a page after that elusive creature has run buy. This thought process is out-of-control to the creator- some primordial force that we cannot command
I imagine this midnight moments forest: Something else is alive Beside the clocks loneliness And this blank page where my fingers move. Through the window I see no star: Something more near Though deeper within darkness Is entering the loneliness:
In Hughes' view, we must wait patiently for the muse to arrive. No crowd sourcing, no storage of the poem in an on-line version control system, no developers releasing early versions of the poem for others to elaborate. In Hughes' world, the best we can do is to coax creativity into creation- not by action, but by our stillness:
Cold, delicately as the dark snow, A foxes nose touches twig, leaf; Two eyes serve a movement, that now And again now, and now, and now
Any wrong move on our part and the fox will bolt. We are frozen, like a hunter waiting for a clear shot or a rabbit trying not to draw attention to itself. Richard Florida would not approve of Hughes' imaginary landscape- a cold, isolated and lonely place with no art galleries or bookstores to visit after the poem is written.
Across clearings, an eye, A widening deepening greenness, Brilliantly, concentratedly, Coming about its own business
In Hughes' world, the artist is like Leonardo DeCaprio watching the in-coming iceberg. We are powerless to bend its business to ours. There are no start-ups we can spin off to explore different poem generation methods. We can't write multiple poems and throw away the one the don't work. All we can do is wait for a wham-bam visit by the muse, running faster than a clock can tick, leaving us reaching for cigarettes.
Till, with a sudden sharp hot stink of fox It enters the dark hole of the head. The window is starless still; the clock ticks, The page is printed.
One can imagine Clint Eastwood playing the Thought Fox, arrogantly gazing at us sweaty and anxious across the clearing, daring us to try anything else except to freeze in our tracks, interrogating us with "You've got to ask yourself one question: 'Do I feel lucky?' Well, do ya punk?"
Of course Ted Hughes wants to convince us that creation is special, the exclusive property of a special breed of poets, like himself. Heh, we all need to justify our salary somehow.
But even the Thought Fox process (which we might characterize as "please go shiver in the snowy forest of your subconscious till the muse lowers itself to deposit a creative turd in your path") is open to engineering principles. If:
Well, you get the idea. Creation can be engineered. We know that the probability of creation at any instant at any place, by a single person is very low. But decades of software engineering has taught us that given the right social conditions and the time and the tools needed to explore a set of ideas, creation is a predictable, controllable process.
References
Sat Feb 9 19:57:14 PST 2008
New Grant with Barry Boehm: "CRYSTAL BALL"
How are we to assess the benefits vs cost trade-offs on all these new methods? How are we to make future plans for the agency, given some much change in current practices?
Using traditional methods, there are no answers to these questions. Such a study is hindered by the local tuning problem. Software process models are most accurate after they have been tuned to local data. Unfortunately, the data required for local tuning is difficult to obtain, especially across multiple organizations. This is due to the business sensitivity associated with the data as well as differences in how the metrics are defined, collected and archived. Sometimes, the data is not even archived (e.g. after two years we were only able to add 7 records to our NASA wide software cost metrics repository).
Recently, we have developed a novel solution to the local tuning problem. We have found that, in terms of ranking different approaches to software development, a precise local tuning is not required. Rather, some knowledge of local circumstances may suffice to make the business case for some policy (A,B,C) vs some alternate policy (X,Y,Z). Based on research dating back to 1981, we assert that the space of possible tunings seen in actual practice is well known. This space can be explored to find features that minimize effort, defects, development time, and other threats to the development process. While we cant say exactly how much some policy (A and B and C) will change effort, defects, and development time, we can comparatively assess policy (A,B,C) against some other policy (X,Y,Z). Better yet, we can explore a very large space of different policies.
Our proposed research is to conduct this kind of analysis over a space of projected changes to NASA software processes.
This is one year project funded by the NASA Software Assurance Research Program.
Sat Feb 9 19:51:59 PST 2008
New paper
Parametric Analysis of ANTARES Re-entry Guidance Algorithms Using Advanced Test Generation and Data AnalysisSat Feb 9 19:46:42 PST 2008
Exhaling after 2 NSF grants
So Dec 10 and Jan 17 I had two NSF grant submissions (FYI: "NSF" is short for "no such funding").
And then there was a new subject to prepare for. And then this. And then that.
Third week in January, a strange mood came over me. Took me a while to realize what it was- I was relaxed! Not running round like an idiot. Empty in-box on my email (that won't last).
Memo: try to space out NSF grants to less than one very month.
Sat Feb 9 19:43:50 PST 2008
FaceBook rules
Joined FaceBook.
Its changed my view of the web- all that AJAX interfaces adds a degree of reactivity that blows away
older conventions (just a wiki, just a blog, just a some other thing).
Such functionality! Via FaceBook I've found old friends from high school and some old AI grad friends. I know that Ken Forbus is updating his knowledge bases, that Enrico doesn't have a beard, and what's new in LolCats.
There are a couple of strange things about FaceBook. Firstly, no more anonymity. It would take more effort to create a fake identity on FaceBook that maintain your real one. Gone are the days of BigMan007.
Secondly, you actually want to use it. I read an interview with the FaceBook CEO and he offered the following telling comment "half of our users log on FaceBook at least once a day". Its true- I've been obsessing over it. Even found a Twitter/FaceBook connection that lets me update my status at high frequency which is both exciting for me and boring for the rest of the world.
Thirdly, because it is so usable, you use it. So I'm seeing my world split between emails and FaceBook messaging. Tim Menzies, split it two: blogs are so 20th century. Onwards to FaceBook.
Sat Feb 9 19:42:43 PST 2008
The LISP attack
My LISP attack continues. I've broken some threshold where now the code flows off
my pen.
LISP was my language of choice in the mid-1980s. Then I had a love affair with Smalltalk that lasted till Smalltalk died.
But it turns out that LISP is worth a second look. In the 1990s, the compiler technology changed. Measured in terms of runtimes and code size, LISP is looking nearly as good as "C".
And my American AI students seem to get it- they seem to riff on the LISP code I'm giving them- exactly the way Prolog never did.
I told a west coast AI friend mine about the LISP attack. He snobbed me about "time to get to the 1990s dude! Use OCAML". So I read up on OCAML and it is a great language. Still, the maturity of the LISP books is hard to compete with:
And my favorite piece of LISP:
(defun shuffle-list (l) (loop for i below (length l) do (rotatef (elt l i) (elt l (random (length l))))) l)
Sat Feb 9 19:39:58 PST 2008
Beyond Pairwise Testing
The website pairwise.org shows the effects on data set size when a simple constraint is imposed on the input space. If no two tests can share the same pair of settings (var1=value1 and var2=value2) then millions of randomly generated tests reduce to hundreds of tests or less. And there are fast algorithms for generating those hundreds of tests.
Now I'm wondering if we can use those hundreds of pairwise tests (PT) as the centroids of a pseudo-clustering algorithm:
If all that works then that would be a unified framework that subsumes treatment learning, anomaly detection, test generation, planning- and does so in lightning fast time. Its a long shot but heh, you never know if you don't give it a go.
Sat Feb 9 19:38:01 PST 2008
Torchwood is soooooo gay
Dr. Who was always camp and BBC's Torchwood series is Dr. Who squared.
It turns out that Cardiff (where it is set) is a pretty sexy place. Well, pretty frequent sex. Plain old sex, bisexuality, homosexuality, sex-with-aliens, sex-with-robots, sex-with-bacteria, sex, sex, sex, sex. Anything is fair game just as long as it it moves, breathes, or has a shadow. Genitals not essential.
And its not even good sex- good in the sense that it enhances the plot. Sometimes, it is pivotal to the story (like in the sex-with-bacteria one, where the bug was driven by an the urge to reproduce, at any cost). But usually it is puerile- the kind of sex where you sneaking off round the back of the toilet block for a quick shag, when teacher is not looking.
And be warned: anywhere else on the planet, sex is followed by snuggling and maybe a cigarette. But in Cardiff it seems required to condemn to death the person/alien/robot/microbe that you just bonked. Shoot them dead, chuck them into a time vortex, freeze them for eternity, send them back to 1918 where they will be shot for cowardess, whatever.
So here's a tip for anyone travelling to Britain. When in Cardiff, best to keep it zipped.
Sat Feb 9 19:35:51 PST 2008
New Journal Paper
Learning Better IV&V Practices.
To appear in Innovations in Systems and Software Engineering, ?March 2008.
Wed Jan 9 21:03:12 PST 2008
A good day
A morning writing NSF grants. Phone calls with folks across the country.
In the afternoon, a meeting with grad student "Z" to discuss his new results. He's been working on an idea that's been simmering around since 2002 (Kareem's stuff on IQ) and boiling since XMAS'06 (stochastic beam search).
(And don't tell anyone but the problem is kind of urgent. There is BIG hole in my Jan'07 TSE paper- that no on else knows of except me. Oh, and student "S".)
So I brought along student "S" to audit the results. Three of us realized that Mr.Z was not borderline state-of-the-art, if not over the line. And the problem with the TSE was solved. Killed. Totally.
Emailed current holder to state of the art. Phone call with them the next morning and he agreed we might indeed be state of the art.
Meanwhile, Mr. "Z" did lots more plots. And we are looking good.
What a day. What a great day.
Wed Jan 9 19:56:34 PST 2008
New NSF proposal
Exploiting Islands of Certainty in Systems Engineering.
Tim Menzies (WVU); Martin Feather, Steven Cornford (JPL); James Kiper & Gerald Gannod (Miami U.)
Model-based design is widely advocated as a way (some go so far as to say the way) to meet the ever-increasing expectations levied on complex systems and their software. Substantial focus by the software research community, and substantial investment by commercial field, is rapidly advancing the tools and techniques that underpin model-based design and its pragmatic application.
However, we see a looming problem. The figure at right is an extrapolation of the size of design models seen at JPL. This figure shows that we are on the brink of running into severe limits in three fundamental areas:
Our proposal is to research a phenomenon that shows great promise as the means to simultaneously circumvent all three limits.
Intellectual Merit
This proposal also address core challenges set out in this Programs Synopsis:
Broader Impacts
Wed Jan 9 19:29:26 PST 2008
On the 12th day of XMAS
My true love gave to me, yet more to write.
I was just tallying the words generated by me over the XMAS "holiday". This kind of gives a little insight into the life of an academic:
Time to rename my web site, I think, to "spell check 'r us".
Tue Jan 8 06:43:47 PST 2008
Rain, rain, go away
Last two winters, driven to Chicago and Chincoteague in mid-winter. got lucky- unseasonably warm in both cases and sunshine to burn your eyes.
After XMAS, we drove
town to the Atlantic ocean.
No complaints driving down.
But
I guess two years of luck was all we get.
Down at the beach, we scored all the rain and all cold we'd missed last two times around.
But with the had comes the good. We had the whooooole beach to ourselves.
It must be some law of physics. On the last day of the trip, the day we had to leave early,
the sun came out.
All the satellite dishes at the local NASA base unfurled their flowers.
And with the sun, incredible colors in the ocean marsh land.
Sat Jan 5 21:42:39 PST 2008
Video: Markov Logic
Inspiring talk from Pedro Domingos: Learning, Logic, and Probability
See also his papers at his AAAI'06 or his 2005 Machine Learning Journal paper.
Sat Jan 5 21:26:38 PST 2008
The Bay Bridge Schmolze
Take spare parts from 14 other bridges, string them together any which way, and you get the wonderful mess that is the Chesapeake Bay Bridge at Annapolis.
Sat Jan 5 21:20:13 PST 2008
Student evaluations, Data Mining'07
Good results. High overall ratings, high teaching effectiveness. Highest scores:
Click to enlarge:
Sat Jan 5 21:05:57 PST 2008
Revenge of the sugar
Sugar-fest'07. The aftermath. Not pretty.
Chocolate headache. Coffee shake. Vaguely unsettled.
World a little value and murky. Can't tell you the plot of the last three episodes of "24"- I think Jack is torturing someone but can't say who or why.
Wed Dec 26 13:21:24 PST 2007
Twas the night after XMAS
And all through the house, no chocolate remained. Or cheesie-poops or potato chips
or wine or turkey or stuffing or gravy or....
The day was a celebration of technology.
Two
couch potatoes on two couches with two
laptops. First, we
decorated the computers
Next, we watched 14 hours of "24" (Jack started
out being anti-torture, then missed it so he tortured his brother to death).
Then we
video SKYPEd over
to New Zealand to talk to Dot.
Finally we
went to bed carrying:
Two hours of web hacking followed. Helen ported someone's post-structuralist masterpiece into her on-line journal while I bashed away at some tricky PHP. After which I wrote RSS feeds (this article). Clearly, I married the right woman.
We wish you some happy hard drives, and more silicon next year.
Tue Dec 25 23:06:24 PST 2007
What are you scared of?
Helen woke this morning after her standard nightmare.
She was staying at the New Chicagoquerque Town Holiday Best Western Seasons Holiday Inn to attend the 3rd Season of the Four Cees conference.
She's just discovered that she was due on stage in 20 minutes to talk before her peers on some topic she'd never heard of before. Chilling stuff.
So we started trading worse fears:
Mon Dec 24 15:47:06 PST 2007
And the stockings were hung
(By the chimney with care.)
Mon Dec 24 15:39:26 PST 2007
All you'll ever (k)neeeds
And today's score:
Mon Dec 24 15:31:54 PST 2007
My favorite vampires
Hooray for re-runs. I overdosed on Buffy during her day
but I've only ever seen one run (or less) of the Angel stuff.
But my Tivo has caught all the last two seasons of Angel. Love it. Like that Smile Time episode when Angel gets turned into a puppet and must do battle against an evil kids show.
Watch that furious felt fly!
We're building up
to the perfect ending. Our "heroes", morally confused, has been working
for the enemy (trying to change the system from within). Soon they will
stake out some dark alley and turn to face all the legions of hell.
Their cause hopeless, trusting only in each other, the immortals raise their blades to start yet another possibly final battle. Roll credits. Ends series.
We'll may not know how this one ends- but we have hope.
Sun Dec 23 21:18:59 PST 2007
Tis the season to be humble. Not.
Have I been naughty or nice?
Well, at this time of year I am obliged to document
that I have been just frigging awesome!!!!!!
Faculty productivity files due Dec 28. Time to list all that stuff you've done all year. Plus a little spin for the some of the stuff you've only half done.
Of course, us faculty love and trust each other so much that we don't believe a single word of what any of us say. So everything (and I mean every single thing) has to be somehow documented.
Printers run hot, printer ink runs dry, piles of paper grow on the dining room table.
Sun Dec 23 08:33:12 PST 2007
I said what?
Do I believe in this empirical result?
The paper was due in 162 minutes and I was
staring at 300 lines of shell script trying to work
why on earth did I even compare that
with this. Maybe the whole paper,
that I'd been working on for a week, was a
crock?So I stared and stared and stared, remembering that old joke about your own code, 3 months later, is as alien as anyone else's.
After a while, I found that I believed in 1/4 of the results- and starting writing those up. Then I realized I had a baseline issue- my new method was generating values in the range X to Y but what did any other method do?
"O.K.", thinks I, "need to baseline these deltas against.... ooohhhh so that's why I did that screwy thing."
"Gosh, I don't know who I was when I wrote this code but darn that guy was clever."
Sun Dec 23 08:24:45 PST 2007
Tough job, but someone has to do it
Peyton
Fireman, legal
eagle, golf star, moonshiner,
asked
for help.
Seems his golf club requires him to spend $X00 per year in their club and he was running way under budget. So, he said, want dinner?
The wine list proved trickier than expected and there was some dispute over what to get. "But why fight?", we decided, "Get it all."
For starters, I tried the trout salad with potato leek soup. For mains, the crusted sea bass with shaved fennel did not disappoint, but my partner found the grilled salmon engaging.
The desert menu was challenging before we released we were too drunk to care. So there was some cheese cakes and something else that I can't quite remember. Then a drive home, with a short digression to buy cigars, singing all the way. Saturday, a little blurry.
Sat Dec 22 21:56:44 PST 2007
XMAS comes but once a year
"God bless us, every one!"
Sun Dec 16 19:23:12 PST 2007
The gathering storm
There's this moment, about 20 minutes before the party starts, when the place is clean, the food is ready, the wine is chilled, and the flat never looked better. And at that moment, the selfish thought crosses your mind- why share?
What if instead of opening the door to the guests that I lock it instead, kick back on the couch, drink all the booze, and eat all the food?
Then it passes, your friends arrive, the volume goes up and this unworthy thought goes away.
5 hours later you thank your friends for your latest memories, bid them goodbye, face piles of washing up and bags of fresh garbage, and you're happy they came.
Sun Dec 16 19:14:00 PST 2007
Home for the holidays
No more commuting to Baltimore, ever again (at least, not for another 6 weeks).
Fri Dec 14 14:45:35 PST 2007
ICSE accepts PROMISE workshop
The ICSE workshop chairs just sent word- the next PROMISE workshop will run May 12-13 at Leipzig, Germany.
And, for the first time ever, PROMISE and the related Mining Software Repositories workshop are not on the same date!
Fri Dec 14 14:40:48 PST 2007
Inside the ex-ISR building
From the New York Times, April 18 2006
The most ambitious effort by the congressman, Alan B. Mullohan, is a glistening glass-and-steel structure with a swimming pool, sauna and spa rising in a former cow pasture in Fairmont, W.Va., thanks to $103 million of taxpayer money he garnered through special spending allocations known as earmarks.
The headquarters building is likely to sit largely empty upon completion this summer, because the Mullohan-created organization that it was built for, the Institute for Scientific Research, is in disarray, its chief executive having resigned under a cloud of criticism over his $500,000 annual compensation, also paid by earmarked federal money.
18 months later, the building stands empty: a huge empty echoing space, only ever partially used
The building is like some massive airport foyer, but with not a plane in sight.
Fri Dec 14 14:32:49 PST 2007
PDA rules!
Got a new Blackberry curve. Finally broken through the PDA barrier. Now I wake up in the morning and I read the news, on my little screen. Found a good Gmail mobile reader- use that all the time now.
Welcome to the 21st century. How would of thought it would be less than 2" square?
Thu Dec 13 20:27:28 PST 2007
Paper accepted to IEEE Software
May, 2008: Application of a broad-spectrum quantitative requirements model
Tue Dec 11 08:05:35 PST 2007
New paper:
"Increasing Confidence in Verification Results by Combining Diverse Tools: an Empirical Study"
Automated verification tools are capable of detecting subtle errors in models of complex software systems. Unfortunately, it can be difficult to use these tools effectively. Input models must correctly represent essential system behavior. Models must also enable efficient verification in terms of their time and memory requirements. Satisfying these two requirements can be challenging even for experts. But, given a correct model, can a user be confident that the verification results are correct?
Mon Dec 10 19:09:41 PST 2007
Donald Boland to defend master's thesis
Title:
When:
Downloads:
Data discretization is a commonly used preprocessing method in data mining. Several authors have put forth claims that a particular method they have written performs better than other competing methods in this field. Examining these methods we have found that they rely upon unnecessarily complex data structures and techniques in order to perform their preprocessing. They also typically involve sorting each new record to determine its location in the preceding data.
We describe what we consider to be a simple discretization method based upon a randomized binary search tree that provides the sorting routine as one of the properties of inserting into the data structure. We then provide an experimental design to compare our simple discretization method against common methods used prior to learning with Naive Bayes Classifiers. We find very little variation between the performance of commonly used methods for discretization.
Our findings lead us to believe that while there is no single best method of discretization for Naive Bayes Classifiers, simple methods perform as well or nearly as well as complex methods and are thus viable methods for future use.
Mon Dec 3 16:40:33 PST 2007
Daniel Baker to defend master's thesis
"A Hybrid Approach to Expert and Model Based Effort Estimation"
Download thesis: (pdf format).
Sun Dec 2 15:31:14 PST 2007
New research proposal with Uni. Texas (Dallas)
Intelligent Process Definition for Agile Product Line Requirements Engineering
Finding hot spots in questionnaire responses using data mining.
Sun Dec 2 15:09:37 PST 2007
New NSF Proposal
Automated Quality Prediction-- Exploiting Knowledge of the Business Case
Co-PIS: Tim Menzies, Bojan Cukic
It's time for the annual NSF lottery. Lately, NSF is short for "No Software Funding". So why do I submit? Well, as any person with a terminal illness will tell you, hope remains despite massive evidence to the contrary.
PROJECT SUMMARY: Over the past twenty years, the development of fault prediction models has been a very active research area. The reason for such a significant attention to automated quality predictors lays in their practical importance. Current models are useful, as they allow software project managers to better guide the allocation usually meager quality assurance resources to artifacts which need them the most. However, a number of recent results indicate that current research paradigm which relied on relatively straightforward application of machine learning tools in the context of diverse projects and typically code or process measures has reached its limits. Building software quality predictors via data mining is essentially an inductive generalization over past experience. According to Mitchel's classic model of data mining , any inductive generalization explores a version space of possible theories. All data miners hit a performance ceiling effect when they cannot find additional information that better relates software measure with fault occurrence.
To build better quality predictors that break through ceiling effects, we must introduce more topology into the search space. Standard machine learning algorithms lack the business knowledge which characterizes software projects. To add that business knowledge, we propose to investigate which business level project concerns will be the most promising so they can provide guidance for automated fault prediction models. We plan to utilize human-in-the-loop rule elicitation and maintenance combined with automated rule inference. We expect that this approach will allow software managers to safely focus on the application of quality assurance techniques of choice, confident that automated quality predictors will raise alerts about the artifacts where quality issues actually exist.
INTELLECTUAL MERIT: It is time for a fundamental paradigm shift in the way we develop models for software quality prediction. In order to improve software quality prediction models, we will develop the methodology and supporting tools which will make quality prediction a joint endeavor between software quality experts and data mining algorithms. Specifically, we will shift the focus from data mining algorithms towards the software engineering data passed to those algorithms. This proposal will create a new paradigm in automatic quality prediction, adding the information to the models search space and allowing them to improve their predictive performance along the dimensions that make the most sense for software engineering projects.
BROADER IMPACTS: The proposed research addresses key concerns with the current generation of defect predictors which, when resolved, will serve to advance the understanding of data mining in general and software defect prediction in particular. Fault predictors are widely used in industry to focus verification and validation efforts. Better fault predictors would better focus usually scarce resources leading to the detection of the most consequential faults earlier in the development lifecycle. This, in turn, will lead to production of higher quality software using fewer resources. The projects educational activities will include involvement of undergraduate and graduate students in research, the development of new and the enhancement of existing courses. We have a good track record and remain strongly committed to bringing research and educational opportunities to students from the underrepresented groups.
Sun Dec 2 15:02:59 PST 2007
A present from OZ
News from Ron. D.:
Well - Thanksgiving has come and gone and here in OZ we received the
BEST Thanksgiving Day gift you could ever want - yes a NEW LABOR
Government.
After 11 years we were finally able to go to bed drunk and happy instead of drunk and sad.
Its hard to pick the "best moment" - could it be...
1. Howard's Defeat Speech?
2. Costello taking his bat and going home - with tears in his eyes?
3. Alexander Downer staking a claim for his "next life" - "I am after all the longest serving Foreign Minister and have many contacts that would be useful in business and academia? - (if I hadn't turned away splitting my sides with laughter I would have thrown up)
4. Mark Vaile (Nation Party "leader") also throwing in the towel with tears in his eyes (yes, the drought was broken over the weekend) and handing over to one of the 10 remaining members of the "once great Country Party" to find a new leader?
5. Janet Howard NOT being able to host Christmas drinks at Kirribilli House - or even The Lodge - well at least they have Padstow - the new "Whitehall of Sydney"?
6. The famous conga line of Liberal Party suck ups from the media who now have to find new jobs?
7. The thought of John Winston Howard - cleaning out his office and not being able to take his beloved pictures of the Queen, Menzies and GWB.
Yes, life IS good!
Wed Nov 28 15:23:15 PST 2007
Upgrades to web site
New page added,
just for my talks.
Social bookmarking added (see icon, bottom left).
Page titles added, specific to each web page (see top of window)
Search engine upgraded to Google's new Custom Search beta (try typing stuff into the "Find" box, top right).
(Oh, and I've been playing with SQLite, trying to convince myself its time to take the plunge, leave big XML files behind, go fully relational. Another day, perhaps.
Sat Nov 24 22:29:56 PST 2007
Hey, maybe I do have a life
In the last three days:
Mucking around in the garden with my wife...
...great boozey Thanksgiving dinner with the neighbors...
...Massive photo attack- driving around for hours catching a sunny winter's day in rural Pennsylvania...
...much XML hacking: new web site...
...planning an AI subject for the spring: collecting cool materials...
... and lots of fun LISP hacking
:-)
Sat Nov 24 14:09:56 PST 2007
Title
DESCRIPTIONSat Nov 24 14:09:48 PST 2007
Time, once again, for ego repair
Email today. "Decision on your paper". Thrilling!
Then you read the review. Thumbs way down. Oh dear...
Begin stages of grieving:
How to gauge your academic experience: how long does it take you to run from denial to acceptance.
Me, I'm getting better: down to three days.
Sat Nov 24 13:47:36 PST 2007
Graph theory? Art?
Point Marion Bridge, Pennsylvania. Massive, scary, hunk of rusting steel.
But how to describe it? Struts over a bridge? Graph theory? Work of art? All three?
FYI: due for demolition early in 2008.
Sat Nov 24 13:40:33 PST 2007
The Mathematician's Castle
A very strange building, just down the street, opposite the dog park.
No, I don't know its story. But damn, I like it.
Sat Nov 24 13:34:19 PST 2007
Getting good at the XML
Last December, it took me days to get my first RSS feed going
In July it was about 2 weeks to get this web site up and running.
In August, it took about a week to write a subject page, using the same RSS/XML techniques.
But I'm getting faster. In the last 24 hours, working from scratch, I've built a new site with a Google search engine, bookmaking tricks for all the social bookmaking sites, RSS feeds, multiple categories per site, CSS formatting, PHP rendering the site from XML, etc, all the good stuff.
The next hurdle is SQL. Right now, my home brew stuff works just fine for 100s of XML items. Note sure when that will fall off the edge.
Fri Nov 23 16:26:32 PST 2007
Title
DESCRIPTIONFri Nov 23 16:26:16 PST 2007
Spring 2008 AI subject, now on line
Please see http://menzies.us/csx72.
Fri Nov 23 08:52:06 PST 2007
Invited to IEEE ASE 2008 PC
I'm now a member of the IEEE ASE 2008 program committee. L'Aquila, Italy - September 15-19 2008.
If past trends are anything to go by, this will mean reviewing a lot of papers.
Wed Nov 21 10:42:59 PST 2007
$1.4M proposals to NASA
This week, WVU researchers submitted their 2008 University Initiative proposals to NASA. Totaling more than $1.4M, these proposals review and extend the state of the art in software V&V with a particular focus on early life-cycle model-based software engineering.
Sun Nov 18 07:58:04 PST 2007
Research planning with JPL
This week, much time was devoted to exploring new research ideas with requirements engineering researchers at NASA's Jet Propulsion Laboratory.
Sun Nov 18 08:00:51 PST 2007
Fame is so fleeting
For 48 hours, they all knew my name. Now, I'm so last week.
Monday night I uploaded an old talk of mine on Open source- can you ignore it? up to http://slideshare.net. Someone wrote a nice comment on it and it got featured on the slideshare home page.
Success begets success. The slides rose to number 5, dropped a little, then held firm at number 7.
For Tuesday/Wednesday, that was the status quo. For the record, I shared the limelight with slides shows about:
Then, on Thursday, disaster struck. Someone posted Living in a garbage
truck and I knew the wild ride was over.
By midnight, I was off the home page. Currently, I'm number 43 and falling, lost in obscurity. Just another faded former rock star.
(Update, Monday, now #56. Eeek! Lights, fading. Oxygen, giving out....)
Sat Nov 17 19:42:18 PST 2007
Help. I like Peter Frampton
Stumbled over a recent concert by Peter Frampton on HD TV. In his day, Frampton was the best of the best. "Frampton Comes Alive!" went 6 times platinum, and is now the fourth best selling live album of all time
And he still does a great show. He did the talk box thing, played some nice guitar, charmed the audience with his refined British accent. Looked like he was having fun++.
But what fascinated me was his age. In his younger days, he was pretty serious guitar-god eye-candy.
Now he's beyond middle aged. Or is he? I started at him, eyes squinting, puzzled.
Guess I'm feeling my age (47). I stared at old-Frampton, much like I look at myself, trying to see if that young fella is still in there.
Sat Nov 17 19:25:13 PST 2007
Bumblebee Bat
According to the Animal Diversity Web, Thailand's bumblebee bat (a.k.a. Kitti's hognosed bat or Craseonycteris thonglongyai) competes with the Etruscan pygmy shrew for the title of world's smallest mammal.
At issue is whether skull size or mass defines "smallest"; shrews are lighter but the bumblebee's skull, shown right, is smaller at 11mm.
These bats are so-named since they're about the size of a bumblebee, weigh about as much as a dime, and have the ability to hover like hummingbirds. Their roosting habitat consists of the hot upper chambers of caves in limestone hills. They are most active at dusk when they fly around the tops of bamboo clumps and teak trees to feed on insects.
Solitary
While bumblebee roosts in colonies, an individual is solitary. Though close to others in its group, it hangs alone, rather than clustering tightly. Due to its tiny size, and its predilection for separation from the rest of the population, the bumblebee bat is the perfect symbol of a Bayesian Treatment learner.
Famous
The bumblebee's fame far exceeds their size. Cuter than a button but less glamorous than, say, blood-feeding vampire bats, only the bumblebee's extreme size ensures a cult following. No survey of terrestrial bats is complete without a reference to the humble bumble. Little girls around the globe adore the bumblebee (e.g. http://kids-learn.org/stellaluna/florin.htm).
Predictably, bumblebees don't attract little boys who seem to prefer ghoulishly fantasizing about the dripping fangs of vampires.
Rare
Unfortunately, the bats are small both in size and in number. The species was unknown prior to 1974 and is found only in the Karmhanaburi Province in Western Thailand. Their habitat has been highly affected by deforestation and unsustainable levels of teak logging. In 1982, the Royal Forest Department of the Thailand Government could only found 160 of them living in 3 caves, despite extensive surveys.
Endangered
Sad to say, bumblebee bats may now only exist on the web pages of young girls. Reports on their population are ambiguous and quite dated:
The Zoological Society of London lists the bumblebee bat as one of its "top ten actions" but is not clear it on-line donations will go just to the bat, or to other endangered species in Thailand, or to endangered species around the globe.
For more information, see thewebsiteofeverything or www.helium.com or Hello, Bumblebee Bat.
Sat Nov 17 18:51:55 PST 2007
CocoMonster
I love this bit in Dan's thesis:The 3rd bagging experiment tried involved the creation of an algorithm that creates a learner using every permutation of the feature set. This algorithm was named CocoMonster for its ridiculous computational complexity.
There is some kind of intrinsic appeal to overly complex algorithms. One may think, It must be better; its just so...complicated.
This is much like the snob factor in economics which leads us to believe that a diamond must be rare and valuable because it costs a lot of money.
Sat Nov 17 18:39:16 PST 2007
Relationship advice
If someone is starting out on an academic career, they really
should be warned about the "two-bodied problem"; how it can take years
(if ever) to find some place where a Dr. Spouse and Dr. Spice have jobs
in the same zip code.
And, if its not too late, they need to be get the right advice about where to search for a spouse:
So, now you know that, you can sale off on a successful career, two by two, in love that's true.
Sat Nov 17 18:24:41 PST 2007
Attack of the thesis
Autumn. Trees go red, days grow dark, students race
to write their thesis in time to graduate before XMAS, and supervisor read hundreds of pages
Tuesday night was Dan (120 pages), then DJ (40 pages); Wednesday night, Justin (80 pages).
Sent back all these edit notes.
Which means tomorrow I should get back Dan (140 pages), DJ (80 ages), Justin (100 pages).
And so the cycle of (academic) life continues.
Sat Nov 17 18:13:59 PST 2007
2cee at COCOMO forum
Dr. Jairus Hihn (JPL), Prof. Tim Menzies (WVU), and Karen Lum were highly visible at the 22nd International Forum on COCOMO and Systems/Software Cost Modeling where they made two presentations describing their research results (2cee: A 21st Century Effort Estimation Methodology and STAR: a novel solution to the local tuning problem), demoed 2cee at the tools fair (where Karen lost her voice because of the level of interest) and participated in two panel sessions one of which was on the Future of Cost Estimation.
2cee is available for free to U.S. government agencies, universities, and companies on official government contract. To request a license and copy of 2cee, send an email to softwarerelease@jpl.nasa.gov with your name, citizenship, and affiliation.
Sat Nov 10 14:12:39 PST 2007
This beach is closed
Ozzie boy meets North American autumn. Autumn wins.
Sat Nov 10 13:04:47 PST 2007
Rewriting AI history
One of the standard stories of AI is that a system called DART saved the Army so much money, that it repaid all the bucks they'd ever invested in AI. DART was a real-time just-in-time scheduling system used to move materials into Iraqi war (version 1.0).
Now one of sad side-effects of knowing too many people is that some of them have long memories. I was at a conference last week where one and two of attendees were living off DoD grants in the early 1990s. And I made the mistake of telling the DART story. They exploded into laughter.
"That thing?", they said, "just a quick and dirty GUI that replaced a ridiculously awkward manual system. The army couldn't tell the difference. They were just happy that anything worked better than their old system. "
I asked, but why DART's reputation? And the answer was obvious.
"No sane AI hacker was going to tell the funding agency that their baby was a dud."
Sat Nov 10 13:29:08 PST 2007
ASE'07=excellent
ASE 2007 was my definition of a good conference.
Only downside was that my talk was not till midday, day 3. So the whole time I was sneaking off to practice/ rewrite/ practice/ rewrite the talk again and again and again.
Afterwards, I could get home, easy as pie. In the old days, international conferences meant international travel, locked in teenee weenee tubes, eating airline food, watching bad movies, and getting dehydrated like a prune. Weeks of living packed into a few days elapsed time, followed by a zombie week waiting for the brain to arrive back from overseas.
That was then. This is now. Travel to conferences in the same time zone. One short plane flight start to finish. Yesterday i cam back from ASE 2007. Slipped back from Atlanta, easy as pie.
Fin
Sat Nov 10 13:13:47 PST 2007
Bernd Fischer isa muppet?
Right: Bernd Fischer.
Left: a fraggle muppet.
Your call.
Sat Nov 10 13:55:31 PST 2007
Learning Near Optimum Inspection Policies
Wed Fed 6,11:30am, 2008
WHERE: room 225b, IVV building
WHO: (Dr. Tim Menzies, Zach Milton) wvu, csee
ABSTRACT: High assurance software requires extensive and expensive assessment. There are many forms of software assessment, ranging from manual inspections to automatic formal methods. These assessment methods differ in their effectiveness and the effort required to apply them. Typically, the more effective methods are more expensive. Hence, project managers often "bias" the assessment resources and apply more effort where they think that extra effort might be most useful.
If most of the assessment effort explores project artifacts A,B,C,D, then that leaves a "blind spot" in E,F,G,H,I,.... Blind spots can compromise high assurance software. It is therefore important to discuss the bias introduced by the inspection policy. To but the matter in a nutshell, we need to ask "how blinding is our bias?"
This talk contrasts three different kinds of "bias" in selecting what code modules to inspect:
We find that #1 usually outperforms #2. This result calls into question many years of research by the speaker (translation: "oh dear.....").
But we also find that #3 almost always out-performs #1 or #2 (translation: "phew!!").
In fact #3 works so well that we speculate that it could be used as a proxy for determining the actual number of defects remaining to be found, after inspecting Z% of the code.
Sat Feb 9 20:08:13 PST 2008
New talk: The Business Case for Automated Software Engineering
IEEE Automated Software Engineering, Nov, 2007, Atlanta, USA. ASE 2007. Download paper on-line.
Adoption of advanced automated SE (ASE) tools would be more favored if a business case could be made that these tools are more valuable than alternate methods. In theory, software prediction models can be used to make that case. In practice, this is complicated by the "local tuning" problem. Normally, predictors for software effort and defects and threat use local data to tune their predictions. Such local tuning data is often unavailable.
This paper shows that assessing the relative merits of different SE methods need not require precise local tunings. STAR1 is a simulated annealer plus a Bayesian post-processor that explores the space of possible local tunings within software prediction models. STAR1 ranks project decisions by their effects on effort and defects and threats. In experiments with NASA systems, STAR1 found one project where ASE were essential for minimizing effort/ defect/ threats; and another project were ASE tools were merely optional.
Sat Nov 10 12:42:59 PST 2007
The chair, the chair
Every conference I go to has this chair. Exactly this chair. I think it is in love with me.
Sat Nov 10 13:03:12 PST 2007
New talk: Nighthawk: Two-Level Genetic-Random Unit Test Data Generator
ASE 2007. Paper available on-line.
Randomized testing has been shown to be an effective method for testing software units. However, the thoroughness of randomized unit testing varies widely according to the settings of certain parameters, such as the relative frequencies with which methods are called. In this paper, we describe a system which uses a genetic algorithm to find parameters for randomized unit testing that optimize test coverage. We compare our coverage results to previous work, and report on case studies and experiments on system options.
Sat Nov 10 12:35:30 PST 2007
Best picture I ever took
Talk 'bout dumb luck. This shot took all of 10ms to set up.
Atlantic lighthouse off-shore from Portland, Maine.
View bigger and even bigger.
Sat Nov 10 14:30:34 PST 2007
Helen vs the hurricane
Had a little Pacific northwest experience last week. But in the Atlantic northeast. Portland Maine in the late autumn. Some sunshine but days of overcast gloom, thanks to the tail end of Hurricane Noel. And I was back in Oregon winters again going "get me outta here!".
Me and Helen were there to do a panel at SLS. It didn't matter 'bout the weather- she shone, she shooked, she moved, she nudged, she prodded, person of influence were influenced, planets were re-aligned. Persons sucked up to her to get a publication in her hot new on-line journal hyperrhiz. It was great to see.
Our panel was fun. At one point, half the audience was bursting out of their seats to make comments. And such a discussion- incredible parallels to software engineering. A theoretical acceptance of how knowledge is shifting, that our ideas live on sand. But actions speak louder than words. In what they do each day, the tools they use, they show a belief in that it can all be written down right, first time. Fascinating!
Tue Nov 6 18:53:37 PST 2007
New talk: If you fix everything, you have nothing left to lose
On the Value of Stochastic Abduction (if you fix everything, you lose fixes for everything else) International workshop on living with uncertainty. 2007.
Creative solutions can sometimes be found in larger space of possible qualitative behaviors than in the tighter space of precise quantitative behaviors. To put that another way: If you fix everything, you loss fixes for everything else. We illustrate this effect with an example from software process theory.
Mon Nov 5 19:52:35 PST 2007
Missing: ghosts and ghouls
What if you threw a satanic ritual and no one came?
Halloween'07 I made the mistake of moving from our porch (on the corner) to a neighbor's place (just a little ways down our side street). We sat there with our candy waiting to see 1000 cute kiddies.... and no one came! 100 yards in either direction we could see the main streets teeming with kids. But without our corner porch luring them into a little alley, we were off the grid.
Across the way, Max and Max snr sat with their HUGE bowl of sweeties. Same story there- no kids. To pass the time we threw barbs and candy at each other.
Eventually, we gave in to the inevitable. Lara and me moved our chairs and bowl of candy to the end of the street. Max came along and gave us his HIGE bowl of sweeties. Then he went back to keep his Dad company, back on the porch. We did a roaring trade- cause we were now in the ghostie trade winds.
Next year- bear traps. And we'll move cavity-central to our porch.
Sat Nov 3 12:59:33 PDT 2007
New talk: Multi-implications of multi-dimensional authoring: or, everything you wanted to know about geek herding, but were afraid to ask
If a stone tablet doubles in size, it weighs eight times as much. What is true for tablets is true for other media. As the dimensionality of our media grows, the authoring effort increases super-linearly. With open source tools, and utilizing crowd souring, massively multi-dimensional multi-media products can be produced on minuscule budgets. In open source, all products are available at all times for copying, or branching the project in an alternate direction. Unless the geeks work together, they will splinter into less productive sub-groups. Gangs of geeks generating new media become reliant on each other. Individuals learn to serve the needs of the group, often constraining their own ideas to those endorsed by the group. In one sense, this is old news. Open source multi-media generation is just relearning a centuries-old lesson that crowds can generate more than individuals. Shakespeare's portfolios are excellent examples of crowd sourced content generation. The plays themselves were remixes of older stories. Actors improvised portions of the plays before they were recorded in the portfolios.
But whats old is now back in the news. Our legal institutions, fixated on ownerships or corporate property rights, actively block crowd sourcing. Academic institutions (read tenure committees) give little credence to team players. Yet modern media authors must enlist in an army to complete multi-dimensional masterpieces. Do you like the ten people sitting next to you? You'd better, they've just become as important as your heart beat for completing your next project. But before you get together, you'd better generate an acceptance of modern models of accreditation that move beyond concepts of I, and that acknowledge us.
Sat Nov 3 12:51:59 PDT 2007
New paper: Application of a broad-spectrum quantitative requirements model to early-lifecycle decision making
During the early phases of projects life cycles, detailed information is scarce; yet there is frequent need to make key decisions, especially those concerning tradeoffs among quality requirements. Trading among competing concerns occurs in many fields, e.g. systems engineering, hardware engineering, and software engineering. Here, we report on our design and use of a model to help make requirements decisions that applies to all of these fields. To do so, we employ a model based upon coarse quantification of the factors involved. We illustrate our approach with fragments taken from some of our studies of software technologies.
Sat Nov 3 12:49:12 PDT 2007
End of Spam?
About 3 weeks ago,
I replaced my email text (see above) with an image file showing the same information.
And my spam has fallen right away. No longer, it seems, can I make money from an accountancy mistake in Nigeria. And an 8" penis (that satisfies her all night long) it out of the question.
Fine by me. Nigeria can keep its cash and she if she wants those eight inches, she can take the pills herself.
Just as long as I don't get all that damned spam.
Sun Oct 28 11:13:46 PDT 2007
On a road to somewhere
In Australia, there is one (or two) places to drive. The Princess Highway runs down the east; and the Hume and New England Highways dive inland. So the road signs all count down to one or two places: Sydney, Canberra, Melbourne.
That's the first thing you notice as different when you come to America. More population, more towns, more places to drive, road signs all telling your that Haggerstown is 47 miles away.
So, on an American highway, the sign in this photo is really unusual. Shown here is the most easterly part of I-70 which runs, as the sign says, for 1,700 miles west to Denver then on to Cove Fort.
1,700 miles to somewhere? Beauty. Time to floor the Holden, sink a Tooheys, turn up 2JJ, and run down a few roos.
Sat Oct 27 20:08:23 PDT 2007
Very stimulating week
And by "stimulating", I mean I can barely remember it all.
The week started in Baltimore and ended in M'town. There is a lot a NASA stuff going down right now- meetings with lobbyists, strategy meetings, etc.etc. Meanwhile there is teaching to do, papers to write, grants to prepare. And next week I leave for SLS, then IWLU, then ASE. Still haven't started on those talks. I guess this is my life right now but its a little (a lot) busier than I expected.
Sat Oct 27 17:03:17 PDT 2007
Post-modern poem
I love this postmodern poem. At first glance, it looks like a shopping list left on a dining room table by my neighbor about to leave for a long trip. But, when read again, it changes.
Saline (log)Cotton padsQtipsdeod, toothp> Zuldroman synthroid - dreser *better tray with pill low of G 3 heau 3 pimple bag of bottles clan on 7 bag of bottles ---------------------------- nardstromi's boxdrop chest- jeans- khoh'ssweaters wardrobebluegreensjumper (prothe) 24 fleeceI1white tim throdentdrawers (5th)socks for tums 3<Act of Treason> bills trenchcoatNote how the author has married text genres from traditional forms (the spy's cloak and dagger and trenchcoat) with post-textual representations. "<Act of treason>" is template notation from the C++ language. It directs the compiler to take a generic container (in this case, "act" which is short for "action") and populate it with N treasonous acts.
Here, the author is stressing both the reader and the compiler- the poem does not say how many treasonous acts to add and must assume the hard case of an infinite number of treasons.
Sat Oct 27 16:42:31 PDT 2007
New Paper: Just Enough Learning (of Association Rules): The TAR2 Treatment Learner
An over-zealous machine learner can automatically generate large, intricate, the- theories which can be hard to understand. However, such intricate learning is not necessary in domains that lack complex relationships. A much simpler learner can suffice in domains with narrow funnels; i.e. where most domain variables are controlled by a very small subset.
Sat Oct 27 16:37:06 PDT 2007
New paper: Cross- vs Within-Company Defect Prediction
In a recent May 2007 IEEE TSE article, Kitchenham et.al. explored effort estimation and found contradictory evidence about the value of cross- vs within-company data. Those contradictory results may have been the result of effort estimation features, some of which are subjective in nature.
Static code features are different than effort estimation features. They can be generated in an automatic, rapid, and uniform manner across multiple projects. Therefore, in theory, the conclusions reached from such features may be more uniform. This paper tests that theory by searching for uniform conclusions using cross- or within-company static code features. Whereas Kitchenham et.al. explored effort estimation, this paper explores defect prediction.
Cross-company static code features will be found to generate higher false alarm rates than within-company data. Hence, cross-company data is best used for mission critical software where (a) the extra costs associated with high false alarm rates is compensated by (b) an associated increase in the probability of predicting fault modules. For other classes of software, false alarm rates can be decreased using a very small amount of local data (often, just 100 modules). In our experiments, the use of within-company data halved the false alarm rate while decreasing prediction rates by only 10%. Hence, for non-mission-critical software, we strongly recommend using within-company data for defect prediction.
Thu Oct 25 18:06:15 PDT 2007
New talk: "STAR"- seeking new frontiers in cost modeling
2007 COCOMO forum.
STAR has 3 key advantages over traditional methods and even 2cee:
STAR is an abductive inference engine that applies simulated annealing to a treatment learner.
Thu Oct 25 17:52:32 PDT 2007
New talk: "2cee" a 21st century effort estimation method
2007 COCOMO forum.
It became quickly apparent in the early states of our research that there was a major disconnect between the techniques used by estimation practitioners and the numerous ideas being addressed in the research community. Many fundamental estimation questions were not being addressed. We had to fix all that.
Thu Oct 25 17:45:14 PDT 2007
Spiders vs Volcanoes
Some things never change. I was at dinner the other night and
the North Americans starting going on and on about what a dangerous
place is Australia- snakes, spiders, crocodiles, sharks,.... That's rich- this was a Vancouver Canada dialogue. These folks live on an fault line with active volcanos on the horizon. There are no, zip, zero volcanos in Australia. The whole continent lies inside one tectonic plate. Geological activity? Just say no.
So let's do the maths shall we? Shark attacks last year in Australia vs ... Kobe! Or worse, Pompeii ! See, a crocodiles will eat you- if you are stupid enough to take a boat into their marsh lands. But one volcano can trash a whole city!
It's so unfair. My wife Helen speaks of Australia as death trap. And she grew up under the shadow of New Zealand's Mt Ruapehu. Ruapehu is Maori for "twin peaks exploding into pit". Or "get the hell out of here, she's going to blow... again". Sharks and crocodiles attract tourists but when Ruapehu went BANG in 1996, it killed the local skiing industry for years and years.
There's all this geological records for west coast USA. One big earthquake (I mean, really BIG) every 200 years for as far back as we can see. But no big earthquake for the last 400 years. It's coming- everyone knows it. And when it does, the Californian economy (third biggest economy in the world) is going to get a savage dent in it's cash flow (lava flow does that to cash flow).
And that will be nothing, I repeat nothing
compared to pending Yellowstone explosion. The reason Yellowstone
bubbles is that it sits on a huge super-volcano, which is active. The
ground there is 74cm higher than in was in 1923- i.e. the magma is
pouring in underground and the pressure is building. When Yellowstone
blows, magma will be flung 50 kilometers into the atmosphere. Nearly
all life within a thousand kilometers would die from falling ash, lava
flows and the sheer BANG of the explosion. Now lets compare that with, oh- I don't know- how about, snakes! Well, there are anti-venoms. I ask you, is snake bite plus anti-venom worse than volcano plus clouds of poisonous gases plus molten rock falling out of the sky? I don't think so.
And while we are on it, earthquakes are not the only thing to make west coast USA dangerous. What about guns and shooting sprees in high school and citizens going postal? Heck, I'd take a crocodile over Joe-trucking-cap and his AK47, any day of the month!
Sat Oct 13 20:56:07 PDT 2007
218,000 times faster
Omid Jalali has a nice new result,
a 200,000 fold increase in runtimes out of incremental treatment learning by a few tricks,
some smart, some just obvious (but we didn't try them before).
Some simple systems stuff:
Some simple locally guided search:
Some knowledge compilation and caching:
Then you hit ""go" and what took 40 minutes for Jeremy now takes 0.011 ms.
And, just to save you working it out, 40*60/0.011 = 218,182.
Now there are some cheats in the above. Our control vars are only binary (dozens to hundreds) and the extension to attributes of arbitrary arity is an open question. BUT the real world models we are getting from JPL don't need those elaborations.
Even with that all, you gotta say, 200,000 times faster? Party-eee!
Sat Oct 13 20:32:38 PDT 2007
Grass is always greener
Helen called me from D.C. She was at dinner at someone's
house right downtown. From their front steps, she could
look down the road to the capital building, all big and white
and massive, glowing in the night.
I took the call driving round Morgantown. Just the way she described the Capital building, it hit me how darn big is Washington D.C. As she spoke, suddenly I was a Sydney-sider again, only at home in a 3,000,000 strong city. The colors started fading, all around me and Morgantown became a very very small town.
Its a strange world. Helen works in Baltimore and likes Morgantown and I live in Morgantown and crave the big city. Grass is always greener...
(And the cure? Best way to change my mind about Washington would be to live there. Traffic jams, high housing costs, crime, long lines at health care....)
Sat Oct 13 20:07:09 PDT 2007
Still state of the art
I wrote a TSE article in January on defect prediction
(Data Mining Static Code Attributes to Learn Defect Predictors
http://menzies.us/pdf/ 06learnPredict.pdf).
Ever since then, I've been getting other researcher's work to review about better ways to do defect prediction from static code measures. And I keep trying to do better myself. Right now, I have a whole swag of informal results along the lines of :
- Lounge Room
- He proposed to his wife under the gates on the Randwick race course.
- He's a dedicated couch potato sportsman who'll yell at the screen so hard that the batsman turns round and says "sorry, mate, won't happen again".
- Never afraid to try another glass of wine, providing that (a) it is not a Merlot and (b) there is no animal on the label.
- That the King will send troops to smash the Puritan movement or burn Washington.
- That they are better dead than red.
- Or even worse, that somewhere, out there, there is some country that may not think that the U.S. of A. is the greatest country on earth.
- Written a new xml-based content management system
- Created a new way to do my data mining tutorials
- Submitted an STTR
- Interviewed eight students and hired four GRAs
- Started two new research projects
- Wrote a TSE paper
- Worked with researchers in Turkey to find sensational results that will be the next TSE paper
- Helped Helen start up at Baltimore
- Organized new renovations at the NASA Eisland Lab
- Gone to Baltimore
- Wrote two talks for NASA and the ACM
- Set up for class
- Generative design The goal here is to build a planner that automatically constructs a least cost design of some assembly of STEP parts.
- Old informs the new. The goal here is to take a new design, access a related design that is decades old, then find some flaw/enhancement in the new design using the old design.
- STEP-XML ( ISO 10303-2)
- OWL
- OWL + the JESS rule-based programming language + the PROTEGE ontological editor.
- Known libraries of STEP parts would be mined from available on-line EXPRESS examples.
- Old rule (from cars): don't brake too hard
- New rule (for aircraft): too many G-forces can hurt the pilot.
- abstract:
- In five case studies with real-world NASA data, automatic data mining methods found clear quality predictors (effort, defects). Strange to say, in only one of those cases is that data source still active. Why do organizations ignore all this, clearly useful, data? What to do?
- speaker:
- Tim Menzies
- when:
- Wednesday August 23, 12:30am
- where:
- Eisland lab G29
- download:
- http://menzies.us/pdf/07goodbad.pdf
- a tester that generates test cases from a range of inputs,
- a simulator that allows step-through of model execution,
- and a validator that ensures models do not violate requirements set forth by the user.
- Critical parameter analysis is performed using the machine-learning tool Tar3, a treatment learner.
- All simulation parameters have (and should have) an influence on the outcome of the simulation.
- Tar3 is used to identify those variables that are most critical, i.e., changes in their value determine if the craft is landing on target or not.
- Running Tar3 provides a weighted list of treatments, i.e., ranges for simulation parameters, which result in better success rates.
- Critical parameters are very important to guide further analysis, e.g., to establish operational and safety margins.
- First you need cars- intricate pieces of kludgey machinery that consume huge portions of our income.
- Next, you need the roads for the cars to run on- huge black rivers that scar the landscape, slash away mountains, rape forests, and damn rivers.
- And all this only let us drive faster and straighter to our next Fat Mac!
- Faculty retreat last Friday.
- Departmental picnic last Thursday.
- Been interviewing graduate students, trying to find my next young gun. Hard task.
- The Mermaids Singing (1995)
- The Wire in the Blood (1997)
- The Last Temptation (2002)
- The Torment of Others (2005)
- (read excerpts)
- Crowded airport, booming announcements, bad phone lines, very fast speaking Indians, hard to hear.
- I spoke to Eesha (or was it Naija?). After twenty minutes on one call, she said "oh- this is a ticket with an international carrier. you'll have to speak to our international desk".
- Sigh. Back to the beginning.
- More "on hold" listening to an insanely cheerful recorded voice trying to seduce me into a trip to Lisbon.
- After 50 minutes of this, I started to think that a plane to Lisbon (or, in fact, to anywhere except here) would be a very good idea.
- So I planned a day or two of eating in downtown Philly and sleeping in a BIG BIG hotel bed.
- I had no booking;
- That I could not get on the flight without me contacting the carrier that took me across the Atlantic (British Airways);
- But that was pointless because there were no spare seats;
- That I would have to fly with another airline (me paying the fare);
- etc, etc.
- As requested in his will, his body was preserved and stored in a wooden cabinet, termed his "Auto-icon".
- The Auto-Icon is kept on public display at the end of the South Cloisters in the main building of the College.
- For the 100th and 150th anniversaries of the college, the Auto-Icon was brought to the meeting of the College Council, where he was listed as "present but not voting". Tradition holds that if the council's vote on any motion is tied, the auto-icon always breaks the tie by voting in favor of the motion.
- The Auto-Icon has always had a wax head, as Bentham's head was badly damaged in the preservation process. The real head was displayed in the same case for many years, but became the target of repeated student pranks including being stolen on more than one occasion. It is now locked away securely.
- coffee,
- sugar,
- talk,
- seek agreement when you can,
- politely argue otherwise,
- repeat till dinner time
- then go to pub.
- Until recently the topic of what data are used to evaluate various competing prediction systems has scarcely been considered.
- An important initiative by Menzies et al. has been to promote a systematic collection of properly documented and catalogued data.
- This has been motivated by various concerns not least the difficulty of replicating other researchers work when the data are not made publicly available.
- -- Software project economics: a road map - ICSE 2007
Fri Aug 3 13:01:22 PDT 2007
Unmentionables
When I grew up I was told that some subjects were "unmentionables", not part of polite society. I never really knew what they were but I remained curious.
So I started thinking what kind of society it might be if those unmentionables were left outside. In a box, maybe put out by considerate hosts at parties. A box full of everyone's unmentionables, all squirming together in the box. Like eels.
And since these would be polite parties, there will be etiquette- rules of behavior for dumping things off in the box. You have to cast it off, quickly, quietly, no fuss. And once must never look at what is already in the box. That just wouldn't do.
( But of course, everyone looks AND talks about it afterwards. Inside the party, little groups fix brave smiles on their faces and whisper in awestruck tones through clenched teeth, "Did you see, did you see? We gotta try ourselves some of that.")
And I'm guessing it might be humbling, standing there at the box. You've emptied all your pockets and wallets and shaken your coat twice. And you're still staring at the box saying "I'm sure, I'm just sure there was something else".
Your friend walks by, glances into the box, and says "Nope, that's all". And you say "that's it? that's all? that's why I can't stop (insert your most annoying flaw here)?".
And cause they are your friend, they take the time to check. They rummage around a little, count it all up, and tell you "yup, that's everything, that's all your stuff".
And you reel back, horrified that they know all your stuff. Then you check their stuff, see that it's all there, and realize that everyone knows everyone else's stuff- we only think we hide it.
And if it's all gone, all dropped off in the box, what do you do then? No hidden issues or agendas or hopes. No reason to live or reason to die. "Hello, how are you?" would have no reply and would not even be asked. Cause you would have no drive to find out. No reason to even think about the answer. Everyone would stand mute and still as statues (while the hosts of the party attended politely and unobtrusively to everyone's food, fluids, and toiletry needs).
Just a tip- it's not polite to drink at these parties. Or wise. Loss of focus? Bad idea! Imagine searching through the box and getting it wrong. Peering bleary-eyed at a hundred that look like familiar, but is it really yours? Could get ugly- think what you might catch if you took someone else's.
So why would you go to such a party and how long would you stay? Well, it might be kind of nice, leaving it outside. Aristotle said he liked being an old man since youth has its disadvantages: "like being chained to a madman".
Once there, you can't leave early. That would be rude, just not done at polite parties. But that won't stop everyone. You know who I mean. That couple who arrive late, eat all the food then sneak out early. And they steal the best bits out of the box as they go! Or worse,they double dip!
Which means as the evening wears out, the party gets tense. No one wants to be first, but you can't be last. Or you won't find a matching set in the box. Or worse, you'll get stuck with a bad one (at least till the next party). At the very last minute, everyone bolts! One second: a room full of mild mannered men and women. The next, dust swirling in the air as everyone runs to squabble around the box.
And someone loses out- forced to leave the party with something they didn't bring. And what's wrong with that, I say? Why always have the same unmentionables? Experiment! Live a little! Mix and match!
I can't say more. The party's breaking up and there's some stuff I want to try on, for a few days anyway.
And meanwhile, outside, all the rogues with the stolen mentionables are roaming wild in the streets. The best and worst selection of everyone's egos, dreams, and schemes- are all mixed up and running around and freely available.
Out there, people struggle daily with the nameless to pick what desires they indulge or disdain. And how you are judged tomorrow depends on the choices you make today. Now that's my kind of party. Gotta go!
Thu Aug 2 21:09:04 PDT 2007
Guess what happens if you eat
For the last two weeks I've been doing this South Beach diet phase 1 thing. Talk about dragging the chain. I was trying to write a paper with Omid and it was like giving birth, with the clock stopped.
Well, Saturday the phase1 stopped and the eating started. There was a birthday party and much silly talking and too much wine. Then, bleary eyed and feeble of body, I sat down to do papers on Sunday. And eight pages fell out of my pen. Monday Tuesday were data mining super hacks. Wednesday I rested then more data mining super hacks on Thursday.
So, amazing to say, eating gives you energy to do things. Who would have thunk it?
Thu Aug 2 16:36:53 PDT 2007
Sssh... don't tell anyone, but I seem to be up to date
This has to be a mistake:
- Research deliverables all on-track,
- 2007 fall subject advertised,
- grad students working on rewriting drafts of journal papers,
- money all sorted out for the fall,
- grad students appointed,
- advertisements our for new grad students,
- travel sorted out for London next week,
- up to date with all my data mining tasks,
- email in-box empty,
- to do and waiting lists tamed
I even have time to blog!
Hang on, now I remember, I have to write a talk for Wednesday and I have around 20 pages of two failing journal papers to rewrite. And the fall subject has to be sorted out (lectures, code).
Oh well, it was nice feeling while it lasted (for all of 32 minutes).
Thu Aug 2 16:28:27 PDT 2007
Towards a mental model of NASA-based research
So with Lisa Montgomery and Frank Gmeindl, we starting brain-storming a mental model of the NASA SARP research program.
Nothing official, just some tentative jottings.
At first, we were just doodling but after a while a rich cognitive map started emerging on the board- and when we tried to break it, most of our new ideas fell into the existing map.
We started identifying rich structures between ideas and changes to NASA practices. Before, when we'd done these exercises, the push was "make the program relevant to projects". In that light you either were with projects or you were useless.
Now we found all these partial deliverables that are on the path for getting to projects, but might in themselves be useful deliverables to other projects (e.g. empirical studies checking what works best in a particular context).
So this has no purpose, as yet. But it was most exciting seeing a fresh structure to something I've been working with since 1998. We're going to iterative it, a lot, in the weeks to come and it might be junk that should be thrown away. But right now, we are rather proud of it- it seems to be a mental map of industrial-based research in general rather than just a reflection of immediate 2007 concerns (that might disappear next week).
Thu Aug 2 16:18:54 PDT 2007
It's my birthday and I'll eat if I want to
47 today (well, yesterday actually, 16,158
kms away, where I was born) and I've just done South Beach, Phase 1
. Two weeks of no booze, no bread, no fruit, no fun (actually, not so bad after day 8, once the carb-cravings stopped).
Clearly (birthday) + (end of Phase1) = a good enough reason for a bit of a bash. So a hearty party was convened on the deck of the Boathouse Bistro. And we were merry.
Now its the morning after. Guess what happens if you don't drink for two weeks then go out with four witty friends who like the grape? 'Nough said.
Here are all the photos, mostly thanks to surfbabe69.
Sun Jul 29 04:58:54 PDT 2007
WVU researchers visit "Rocket City"
"The speed of light is too slow", says Dr. Reagan Moore. "It takes 60ms,
at least, for a computer in San Diego to talk to Washington D.C. That's
too slow for fine-grained collaboration when processing terabytes of data".
Dr. Moore was speaking at the opening of the Electronic Records Archives (ERA) Research Lab at the Allegany Ballistics Lab, Rocket Center, West Virginia. The lab is devoted to the big problem of maintaining the records of the republic of America, for the life of that republic. Moore spoke of systems that dispatched batches of operations to remote sites, so that they could then return the results without time-consuming interactions.
WVU researchers, along with NETL were in attendance as part of the new NETL/WVU collaboration
with ERA. The problems faced by ERA are immense, and a new generation of technology is required
to support it. While other applications talk are measured terms of megabytes or gigabytes,
ERA looks at Terabytes2.
"Just scanning our volume of data, even once, is a technology achievement" said Robert Chadduck, Principle Technologist, ERA Program.
Also in attendance were representatives from several branches of government.
Staffers from the offices of the state's Senator and Congressman were on hand to stress the strong support for ERA
from Capital Hill.
Letters of support
were read out from Congressman Mullohan, and WV state governor Joe Manchin and Senator Byrd.
"Democracy starts with preserving the past to protect democracy in the future", read Byrd's letter.
Wed Jul 25 18:44:26 PDT 2007
WVU CSEE one of world's top 50 SE schools
The June 2007 issue of the Communications of the ACM ranked WVU CSEE in the top 50 of all SE schools, in the world.
The table on p84 of the article tells the story. Little old WVU with its 5 SE researchers scores nearly as well as certain powerhouses of SE research such as the NASA Ames Research Center.
Wed Jul 25 15:16:24 PDT 2007
Renovations to commence for NASA/ATL lab
Currently, the ATL lab has all its desks around the walls. This is not supportive of group work so NASA has approved funds for renovations to create island work spaces.
Wed Jul 25 14:49:53 PDT 2007
Got a minute?
When you get a second, if you've got a minute, can you spare me some
time? I'll buy it and store it in my new time machine.
It's a nifty box that holds time in cold storage. If I've got a few hours to kill, I shove them in this freezer. ZAP! The hours disappears for me and I have a little time up my sleeve.
Which is really handy when I get busy. Deadline approaching? Too much to do? No problem! Grab some spare time out the freezer, thaw it in the microwave and ZAP! My day suddenly grows by a couple of hours.
Course, if everyone used it, things could get a little tricky. Time would become a commodity to be bought and sold. Companies would only stay competitive if they can buy lots of overtime for their workers. Each family would have one "designated worker" who starts a new job Monday morning and retires by Friday after working 40 years in five days. Think of the retirement package- four decades of income!
And who would be selling the time used by the workers? The airlines! Where else can you find thousands of people who would willingly sell hours and hours of their lives in order to avoid a tedious flight? Air travel would become free- and instant! You board the plane and ZAP you arrive. And the airline sells your time in between.
Such air travel would become a civic duty. A child's first flight would start with great pomp and ceremony. On landing, the child would be presented with a scrapbook showing everything that was accomplished using their time. Oh and the shame and humiliation if, somehow, all that time was wasted.
But imagine how this sort of thing could tear families apart. Johnny brings home his report card. Mother looks, and turns pale. Father asks nervously, "What's wrong?"(but he already knows). Dramatically, mother turns and shows Father the report card. Johnny got an "A" in maths! The parents break down in tears- their boy is going to become a worker! Oh why couldn't he have been stupid like his parents? Now they're going to have to bury him!
Then they brighten up when they remember that a week after Johnny's graduation, and just before Johnny dies of overwork, their bank accounts are going to look very healthy indeed. So the initial shock gives way to sage wisdom. "Who are we?", they tell each other, "to stand in the way of the lad's career?"
This would be an interesting world. All the workers would live at lightning speed, never seeing the results of their labors. No, that privilege would be left to the drones who would tour the centuries like hungover football fans on the bus.
"Hey look, we're passing by a period of political oppression run by a Feng Shui neo-communist matriarchy using anti-gravity technology where voting rights are only given to tall nude redheads who have proved their superiority by wining stock car races."
"Again? Yawn! Zzzzz."
Tue Jul 24 19:08:05 PDT 2007
How did we land on the moon?
Last week, on July 20th, was the first time we walked on the moon.
Now if you ask how we go there, then that is another question.
On the descent phase, the guidance computer went nuts. Things were tense, then it wasn't, then it got tense again. For a time, like all of three seconds, they were going to abort the landing (in full view of all the media of all the nations of the world).
They didn't abort because an archetypal computer-nerd named Steve Bales said it was ok to ignore the warnings and land. He was right, thus allowing him enter into the computer-nerd hall of fame and winning a Presidential Medal of Honor.
(I used to have one grainy photo (now lost) of that event- three brave astronauts and one geek (?with pocket protector) getting ribbons pinned to their chest by President Nixon.)
As part of the 25th anniversary of that landing, I wrote a presentation about how that decision was made. It was hard to sort it all out. Bales had back room support (Jack Garman) and some of the folks I spoke to said Garman actually made the call (it didn't appear on the voice record since it was voice-activated mike, direct to Bales).
An alternate story comes from the biography of the mission controller. Gene Kranz said that Bales (or Garman) knew so much about the that codes since, a week before launch, Kranz had set them the task of understanding all the unlikely ones.
Yet another story was one I was told after I gave the presentation . Someone took me aside and told me what really happened. His claim was that it was a requirements error- there was meant to be a built-in auto abort to handle extreme events. That the error was a true error and if something else had happened then that would triggered the other thing and bidda-boom bidda-bang, no landing. Or a bad landing. Or no take-off after landing. Or something.
Confused? Me too. We can't say with any certainty what really happened just 25 years ago. And this is in a room where every breath was measure and monitored amongst people who knew that every breath they took was being measured and monitored.
And the lesson? Well, how much of history is misunderstood, confused, or distorted? What can we say, with certainty about any other historical event?
I dunno, beats me.
Mon Jul 23 19:52:57 PDT 2007
Harry Potter magic
"Attention muggles. Please register your broom for the broom judging contest by 10pm."Barnes and Noble on Friday night was insane. Harry Potter #7 was due out at midnight and every car park was taken. The store was packed with Hermonines, Harries, Hagrids, any number of nameless ghouls witches and goblins, and eight year olds display an encyclopedic knowledge of the whole series (e.g. knowing the password in book one to the Gryffindor common room).
We didn't buy a copy- we were back-ordered on Amazon.com. Next morning Helen rushed downstairs looking for the UPS parcel. Nothing! It came, hours later (UPS must have put on special staff for the occasion) and Helen started, non-stop, till she was done.
My turn now- begin....
(9 hours later)
.... end. Nicely done Ms Rowlings.
Postscript: As if we weren't Potter-nerdy enough, I can report that 24 hours later, Helen re-read the whole thing (just to check what she missed the first time through). That 759 pages. Twice. In three days.
Sun Jul 22 13:50:17 PDT 2007
Where have all the blokes gone?
The car's iPod decided to play lots of Cold Chisel (try saying that in the 20th century) and I realized that they were my "blokes" music.I'm more the nerdy coach potato type- not really a bloke ("bloke" is strine for, er, "manliness" and "strine" is Australian for "how blokes talk"). My idea of a good time, before shin splints, was long distance running. Not the kind of thing that leads to lots of male bonding. So I've never been in a drum circle, chased wild animals with the other apes, etc etc.
Most I ever came to all that was riding to Melbourne with Michael Priest in 1985. 14 days, 1200 kms, Nutrigran-powered pumping machines. But while Michael is a good bloke, he's not a real bloke (remember the apples on shoe laces?).
Strange to say, whenever I did have blokey moments, they were usually with Cold Chisel playing Khe Sanh in the foreground. e.g.
- Peter Bitmead told me earnest tones that "this was the best song ever written in Australia" as he took the LP out of its sleeve, applied the anti-static brush, and played track2, side1 of Cold Chisel's firs album. We listened in awed silence.
- Sherman Young and me on one of our "tweety" weekends in the bright yellow convertible VW, top down,
driving
between nowhere to nowhere on a road marked on no map, singing as loudly as we possibly could:
I left my heart to the sappers round Khe Sanh
And my soul was sold with my cigarettes to the black market man
I've had the Vietnam cold turkey
From the ocean to the Silver City
And it's only other vets could understand
Well the last plane out of Sydney's almost gone
Yeah the last plane out of Sydney's almost gone
And it's really got me worried
I'm goin' nowhere and I'm in a hurry
And the last plane out of Sydney's almost gone
Now I can't say that I miss being a bloke but I miss the other blokes (Peter Bitmead, where are you?) and I miss sitting round with the blokes and pretending that I might be bloke-ish.
Sun Jul 22 13:19:08 PDT 2007
Special issue: Repeatable Experiments in SE
Edited by me.Sat Jul 21 08:58:54 PDT 2007
Special Issue on Information Retrieval for Program Comprehension
Edited by Letha Ezthorn for the EMSE journal.Sat Jul 21 08:04:01 PDT 2007
Member of 2 new workshops
Wed Jul 18 17:17:15 PDT 2007
"Geez, this owright."
(Translation: the current situation is above satisfactory).So sayeth Ron Davenport after 4 hours on our back deck on a hot summer evening. We agreed.
Rain was sprinkling down but the tree above was kind enough to stop most of it. Tiki torches were lit, wine was drunk, followed by moonshine then gin & tonic (I abstained but detailed plans were made for my end-of-diet feast, in 10 days time).
We'd talked and talked and talked into the evening. Max came over and told us strange tales of forensic deaths (we got the snorkel and the plastic wrap, but the boxing gloves? Now that was really weird).
After quoting parts of the Screwtape letters to us, Max lead a discussion on whether or not haggis was kosher (fyi: it is).
Lara came over and told us tales of fighting for hotel room in Leeds during race carnival times.
Helen demonstrated the value of iframes from displaying Flickr slide shows.
Ron discussed Australian cuisine, in particular pie floaters from Harry's Cafe de Wheels.
Meanwhile, her royal highness, Queen Lucie of Cat, chased the fireflies.
Inside the house, my email overflowed with stuff to read, things to do. But I checked- apparently all that stuff will still be there tomorrow.
Tue Jul 17 20:44:50 PDT 2007
TeraByte Tim
What do you do with a terabyte of on-line storage? Please, I want to know.
Each week, my ISP gives me another 2GB of storage, free of charge (just another example of the excellent service at Dreamhost.com).
It accumulates and now I have, wait for it, over a terabyte of storage (well, 1,024,256 MB to be exact).
The thing of it is this- I only use 18GB of it all. So all those bits at my command and I have nothing to say to them.
Do they get lonely?
Tue Jul 17 06:07:49 PDT 2007
Return of the Diet
On the holiday, I started real good but got read bad, real fast.On day one, I was prepared! Cheese sticks and almonds at the ready.
But by day seven, it was all about the Waffle House (man, those waffles are good). So I swore to start the diet again when I got home.
Timed it, just so. Two weeks of South Beach Diet (phase1)= no starch, no bread, no booze, no fruit (sigh, no fun). This ends exactly on my birthday so Helen now knows what to buy me this year- apples!
So now it is day three of phase one and I am spaced out. Not hungry, but strangely calm.
But you'd never know it to talk to me- I find I can't stop blurting out whatever I am thinking. Rapid fire, no diplomacy.
Helen is meant to be on the diet too but I know better.... she's in the next room, loudly eating her granola. Tempted? You bet!
Mon Jul 16 18:22:08 PDT 2007
Visiting NASA, west coast
I was at JPL July 9,10,11 then at NASA AMES July 12
JPL:
At JPL I checked in with Dan Baker (WVU student spending the summer there). He's working on getting our effort estimation methods deployed into JPL.
Also, with Robyn Lutz, he's doing some research into product lines at NASA and how NASA evaluates interface integration issues.
When Dan and Robyn went into meetings, I meet with Jairus Hihn and Karen Lum (part of JPL's cost estimation team) to discuss recent results from Ous Elrawas and Omid Jalali. Ous is telling us how to make decisions without having to first tune the effort/defect/threat models. Omid is telling us what estimation methods not to use.
Much time was spent with Jairus trying to come up with business patterns for COCOMO-like parameters at JPL. We found 10 such patterns and plan to get Ous to simulate them all in one program.
AMES:
At AMES I meet with Johann Schumann and Karen Gundy-Burlet to discuss data mining for NASA simulators. We did some new data mining on simulations of NASA vehicles re-entering the atmosphere and day dreamed about new interactive algorithm that would offer more feedback to the users.
Sun Jul 15 06:29:59 PDT 2007
IEEE TSE accepts "Problems with Precision"
To appear, September 2007.
Sat Jul 14 22:31:53 PDT 2007
Kids=no
Honest, we ain't kidding. No kids. Ever. Just not
interested. Our genes aren't selfish, they're just
slackers. We've hit the "pause" button so hard, we broke our
biological clock. The lifeguards have stop us diving deeper
into our gene pool.
Now we're trying to figure out why. So if you have a minute, maybe you can help us out.
Empirical results
We did a study. 100% of all idiots were babies once:
- That guy who nearly swiped your car at the corner? Ex-baby.
- That couple upstairs who play loud music? Kept the nursery awake with their crying.
- The really unhelpful woman at the insurance office? Used to wear nappies. Bet they were soiled all the time.
Alternatives to kids
Maybe the whole baby/kid thing is wrong-headed. How do kids compare with, say, cats? Well....
- Kittens take six weeks to toilet train.
- Kittens still look cute if they haven't had a bath this month.
- When they grow up, they never bad-mouth you to their therapists.
- No one will question your abilities to function normally at your job when they hear you just got a kitten.
- Also, they kill vermin. (Well, actually, they half kill, then bring inside to proudly show you. Your role in this little dance is to scream in fear and take a broom to the bleeding snake/ wriggling bird/ screeching bat/... For her part, your kitty will gaze at you all perplexed. Then, understanding dawns. and cute kitty will go outside to get you something else to play with.)
On the other hand, there's kids:
- People with kids have to present them in public, screaming, usually in crowded supermarkets.
- People with kids start acting members of some weird brain washing cult: They try to trick others into joining.
- And babies are just stupid. According to Institute president Molly Bentley, in an effort to determine infant survival instincts when attacked, the babies were prodded in an aggressive manner with a broken broom handle. Over 90 percent of them, when poked, failed to make even rudimentary attempts to defend themselves. The remaining 10 percent responded by vacating their bowels. source: The Daily Onion (Studies Reveals: Babies are Stupid).
Clearly, given the choice we'd get cats, not kids. Hmmm... maybe babies are some sort of involuntary thing. Something we can't control. Like an air-borne disease that we pick up from just walking around. And these been lots of baby infections lately. World population is currently 6,278,627,225 (as of 2007).
(Six billion looks pretty bad, right? But look again- the increases are decreasing: 100% increase 1950-2000 but only a 50% increase projected from 2000-2050. We shouldn't get much past 9.22 billion [gasp] before things start dropping again (2075). And while that's a lot of folks, its only double the current population: we might just be able to squeak past the peak without collapse.)
They protest too much?
But by now, you might be offering the Lady Gertrude objection: methinks they protests too much? Why so anti-baby? Hell, we were babies once and we just spent an amiable and fun couple of hours working on this page. So kids can grow into things you don't want to shoot on sight or run from screaming whenever you see, hear, or smell one.
Its a puzzle- why are we so vehemently anti-baby? Could it be that we're over-compensating for our appalling record at nurturing?
- Helen got a Tamagotchi and it died in a week- starved to death.
- I played The Sims and bought a baby. Social services had to take it away!
- Helen always kills every plant she buys, and I'm no better.
The truth is revealed
Then the other night Helen broke down, confessed all, and the anti-baby mystery was solved. Helen's great-great-great mother three times removed married the sister of... dramatic pause.. Clive of India! Robert Clive was an Englishman who lead a band of mercenaries that conquered Bengal in 1757 and executed its leader.
Dangerous genes indeed! And mine are no better. One of my relatives was this arch-conservative Prime Minister of Australia. Robert Gordon Menzies championed the white Australia policy; called out the army to break up miner's strikes; tried to outlaw the communist party; and best of all- kept helping other people to bomb his own country. Just before WWII, he sold pig iron to the Japanese who promptly used it to build bombs to drop back on Australia! And if that wasn't bad enough- after the war, he gave away bits of Australia to the British so they could do atom bomb tests. Some people never learn!
Mystery solved. Our genes are just too dangerous, too imperious, too English empire-ish. The 21st century is not place for a Robert2 Clive-Menzies. These days, we don't need any more atom bombs, white man's burden, black holes of Calcutta, her gracious majesty the Queen, and six different kinds of forks.
Sat Jul 14 21:54:34 PDT 2007
How much do I believe in open source licenses?
Dilemma: someone want to use one of my digital products (hooray) without it's viral creative commons license (boo). The item in question is the photo, shown here, of the UMBC library.
I presented his case to the zeigeist (well, to be accurate, just my wife . She was the only member of the zeitgeist available at that time)
Here's what me and zeigeist decided (more coffee, dear?):
Dear XXXX,
Your question has prompted much debate in my household (well, 5 minutes over breakfast).
The most important thing to say is that you have read the license and thought about the issues. Thank you.
Licensing matters. Do we really want to be edited out from the future? Our digital world may become the lost world, locked away by proprietary licensing, lost on dusty powered-down hard drives in formats no one can read any more, sealed within rooms with forgotten keys. People use what they can touch, and change. As Lessing says, "Creativity is based on the past" (watch video: mov; mp4; mpg)
Open licenses allow our ideas to (maybe) live on, after us. So do I care about your work? Is it valuable enough to keep it accessible for the future? If not, then I don't care if you are select a license that will reduce the odds that the future can touch and extend your work.
Or should I take the Stallman's viral position? Demand that my work and all subsequent work (like yours) is sharable? If "yes" then I take it my pic won't appear in your work which, in all likelihood, will be the only time that it ever gets used.
Or do I allow you to modify the license into a product where the content: gets used a few times (in your work) then lost on some hard drive, never to be accessed again in the future?
(In this case, I note that the ORIGINAL photo will still remain in the public domain. So your decision to be closed in your work will only effect the future of your work. You make your decisions about sharing your new work, I make mine about the original, we both are happy.)
I can't decide so I leave the entrust the matter to you. My license says "Any of the above conditions can be waived if you get permission from the copyright holder.". Accordingly, with one proviso, I grant you the right to distribute the work under any license you like and I invite you to be creative. Perhaps we should stop thinking about "all or nothing". Perhaps there is some halfway point you can find where parts of your material is fully secured and locked away from the future, while other parts are free to access and free to improve in the future.
The one proviso is this- please send me the license agreement that you select to use. I would be interested to read it.
Best wishes. Sorry this response was so long.
Tim Menzies
Sat Jul 14 09:49:39 PDT 2007
San Francisco- what a quaint little village
After 3 days in L.A. I came to S.F. and drove on Highway 101. After the madness of L.A., it all seemed small and quaint. The traffic on the freeway slowed to a halt. Imagine that! It would never happen at 80 m.p.h. (or more) L.A.
I drove from the airport to friends in Palo Alto, slipping in through not-so-crowded streets. It was 5pm so I guess it was rush hour but you could barely tell. Lots of little gardens on the road side. Major thorough fares as tree lined streets. A gentler light than L.A. (not brash in-your-face spotlight but a gentler kinder light).
In 1993 I'd spent a month in this very spot and found it alien and hostile. I vowed never to come to America and went home to finish my Ph.D. and life in the antipodes. Maybe, in 1993, what I needed to do is spend a few days in L.A. BEFORE coming to S.F. And if I'd done that, my life and career would have been radically different.
Thu Jul 12 21:56:00 PDT 2007
L.A. = a mirage of an illusion in the desert
What does it mean that I find NASA's Jet Propulsion Lab more "real" than the rest of L.A.? That a place building robots that will fly out of this world is more grounded than the mirage built on the illusion that is the desert of L.A.?
I confess to some biases. Every since seeing the collapse exhibition I am painfully aware that southern California is a wafer thing layer tacked onto a dying desert ecology. A few dry years and the whole thing will peel off, like a scab, scattering millions of refugees across America.
But even without that in the back of my mind, I would still probably find L.A. an consensual hallucination built to perpetuate an illusion.
In Manhattan, I saw dusty people forced into tiny space, all wriggling to get along despite and because of their diversity. Say what you like about that place, but one thing is clear: there are few illusions here. The whole city acknowledges the "in -your-face" and "here" and "now" of that crowded island.
But in L.A. I see less "here" and more "not here" and "there". The whole place is "not there" in a desert. Every overly green lawn and palatial mansion is alien and screams "Greetings earthlings. Bow before us! We come from a distant and bountiful planet to conqueror you.".
In Manhattan, I saw a vibrant diversity but in L.A., I see rich and poor neighborhoods separated like oil and water, ordained never to mix.
Maybe I can blame it on the cars. Manhattan is a small island- there is no place to hide from the teeming millions. In L.A., 10 million people each day hide in their cars, jump on the freeways, rocket round like pin balls, never to smell their fellow travelers, and rarely have to get out of their way.
So this is the wild west, paved and watered and laundered, but somehow the wild remains. The paper today told the story of border town with Mexico. When the armies of drug lords mounted up and rode into town, local sheriffs all rushed to surrender. I read the story with growing puzzlement. It took a while for me to understand why the writer was reporting this in such bland terms. There was no surprise or shock in this story- their can't be. Of course the rebels are pawing at the castle walls. That is what walls, and rebels, are for. Both are needed- there is no need for walls without rebels and without the rebels we could not justify building the walls. And without the rebels we'd be tempted to connect to our fellow apes. L.A. needs illegal immigrants, meth labs, drive-by shootings in order to justify ostentatious and outrageous displays of wealth where the locals choose their clothes, and their cosmetic surgeon, with all the care of set designer.
Mon Jul 9 20:06:09 PDT 2007
L.A. : mega-myths made real
Today, driving round L.A.:- Warm weather and blue sky. Too blue. Suspiciously blue.
- Massive palm trees lining the street rearing up into the sky. I mean really, palm trees are never that tall.
- Flocks of cars like ballet companies on the freeways, engaged in some elaborate opera.
It occurred to me that it looked like a place made up to look like L.A.
Sun Jul 8 21:20:30 PDT 2007
Darkly Dreaming Dexter (of Data)
Flew Pittsburgh to Pasadena today, all the way reading Darkly Dreaming Dexter. Meanwhile, I was looking forward to running Omid Jalali's Mann-Wilcox' scripts on Yue Jiang new ROC curve data.
So while I was reading of a serial murderer lusting over his next victim, I lusted over data with (almost) the same intensity.
Sun Jul 8 21:14:03 PDT 2007
Summer drive to Gulf of Mexico
Last week of June, first week of July, we jumped in the car with the aim of getting to Key West.
Heck, its only two thousand miles, there and back again. We can do to Pittsburgh to Portland in five days (including deer strike)- and that's 2,500 miles.
Course, we got lost, many times.Didn't get to Key West but ended up on beaches in the Florida panhandle. All up, we drove 3000 miles and made some sensational discoveries. For example:
- Did you know that Florida is bastard hot in the summer?
- And that car air-conditioning is the BEST thing ?
- And that it takes a LONG TIME to drive to Florida?
Yes, all the above are true (write it down, there will be a test).
All in all, it was pretty easy. Highlight was Apalachicola. Other good stuff seen was Spruce's Knob, Harper's Ferry, Fell's Point (Baltimore), Charleston, and the beach at St George's Island.
The winner for "most insane" has to go to the mini golf places at Myrtle Beach.
Low lights were Kitty Hawk and Panama City- two places that precisely describe what NOT to do with a delicate coastal ecology.
Sarah Vowell's audiobook "The Partly Cloudy Patriot" kept us company. She is America's greatest living writer and patriot. I'd gladly join her America. Oh, and the one with that has lots of Apalachicola.
For all the photos from this trip, see here.
Fri Jul 6 13:50:26 PDT 2007
Yet another web site
New look and feel to the my web site.
All content here generated from XML files and imported via PHP scripts calling XSLT or XPATH tricks.
Mon Jul 2 19:30:06 PDT 2007
IEEE Computer accepts "Strangest Thing About Software"
For the Jan '07 issue.
Sat Jul 7 14:17:55 PDT 2007
IEEE TSE accepts 2 articles
IEEE TSE accepts defects and effort estimation papers, pending minor revisions.Mon Aug 13 19:18:18 PDT 2006
Vacation to hills and oceans
Just back from fantastic vacation: hills and oceans of (West) Virginia .Mon Aug 12 19:20:58 PDT 2006
New draft: Making sense of requirements, sooner
Draft to IEEE Computer: Making sense of requirements, sooner .Mon Jun 22 19:22:21 PDT 2006
RE'06 panel of requirements traceability
RE'06 panel briefing: Requirements Traceability .Mon Jun 6 19:23:51 PDT 2006
New paper: Bayesian anomaly detection
New ICML'06 workshop paper: Bayesian Anomaly Detection (BAD v0.1) .Mon Jun 6 19:25:07 PDT 2006
White paper to ONR
With Ali Feliachi and Karl Schoder, white paper on model validation and verification.
Mon May 3 11:24:42 PDT 2007
White paper to DOD
With George Trapp , white paper submission to a DoD Multidisciplinary University Research Initiative on Assured Trusted Information.
Mon Jul 15 11:22:26 PDT 2007
Working the cash flow
NASA IV&V set me chasing the cash flow of the WVU grant funds inside NASA's financial system. After ringing the contact names they gave me, I found (a) who was looking after this one and (b) that there was a little extra information needed to get the $$$ moving.
So IV&V got the extra information together and I nudged the person looking after it.
Net result, hopefully: the funds will be transferred in week(s), not months.
Sat Jul 21 14:00:03 PDT 2007
Talk: Open source- can you ignore it?
- What is open source?
- Where did it come from?
-
Why can't you ignore it?
- Anthropologically: has been with us for centuries
- Economics: too powerful to ignore
- Legally: no future without it
- Technical: unavoidable
- So what social institutions will be adopted to handle it?
Sat Nov 10 18:30:15 PST 2007
Also, I've got some new results that suggests that there are some incredible simplifications to the whole defect prediction process.
All in all, the Jan'07 methods are holding up pretty good, despite lots of work that might have refuted it.
Sure, pretty soon some grad student will blow those results away but for now, for the first time in my life, I am demonstrably state of the art in something.
Sat Oct 13 19:42:39 PDT 2007
Once again, the busy season is here.
UIs to write by Oct 23; two NSF proposals for Dec 10, Jan 10; journal papers to finish, new papers to write, SLS and ASE conferences in two weeks....
Sat Oct 13 19:38:49 PDT 2007
New paper, "The Value of Stochastic Abduction", to appear IWLU'07
Back in the 1980s, the model-based diagnosis (MBD) community explored qualitative representations. Since they are not overly-specific, such representations can be quickly collected in a new domain. Indeed, in domains where information is limited, there may be no alternative to qualitative representations since their is not enough information to build precise quantitative theories.Qualitative theories are inherently nondeterministic but, argued MBD researchers, they can generate a wider range of options for diagnosis and repair. Creative solutions can sometimes be found in larger space of possible qualitative behaviors than in the tighter space of precise quantitative behaviors. To put that another way:
If you fix everything, you loss fixes for everything else.
We illustrate this effect with an example from software process theory.
Sat Sep 29 21:23:06 PDT 2007
Photos at Helen's Fell's Point flat
Ain't nothing wrong with coming here every few weeks.
Sat Sep 29 19:31:54 PDT 2007
Photos from NASA's 2007 Software Assurance Symposium
September 25, 26, 27, 2007
Sat Sep 29 19:28:12 PDT 2007
Saving it for 'ron
Once there was an Aussie called Tim who lived for years
in Morgantown without any other Aussies to play with.
Then, along came... Ron.
Ron is just your average sort of bloke:
Ron and me are so different. He said he had a new G6 and I said I thought Apple only had released up to a G5. He replied "well I don't know about Apple but Pontiac has". He talks cars and I talk computers and that little story just about sums up our different slants on life.
Now, he is gone. Took his wine collection, lovely wife Sandra, luxury apartment down by the waterfront, and now he's buggered off home to Oz. Fair dinkum, what's a bloke to do?
There's nothing for it, really. We must now save all wine for later on. Or, as the
Aussies say it, for 'ron.
Come back, Ron, or
the candles will burn down to a stumps and the wine glasses
will stay forever dusty, empty, and useless.
Sat Sep 29 19:09:14 PDT 2007
Why the question mark in the American anthem?
Went to Port McHenry today where, in 1814, the Brits shelled the Yankees.
A week earlier, Washington had been badly
burned.
Now, was Baltimore to be next?
A fearful audience watched the battle, scared that their
flag and their city would fall to the invaders.
The British were repelled, the flag stood, and Francis Scott Key wrote a poem that was to become the national anthem. At every football/ baseball/ basketball game, you can hear 10,000 voices singing proudly:
And the rockets red glare, the bombs bursting in air,
Gave proof through the night that our flag was still there.
O say, does that star spangled banner yet wave
Oer the land of the free, and the home of the brave?
Now I've always thought that the inclusion of the question mark in the last line to be quintessentially American. My American hosts are individually charming but as a group, nervous and somewhat scared about the rest of the world.
In their somewhat blinkered view of the world, they fear that they are missing something.
So a national anthem that poses a question, a call to check if their country still exists, that requires a body count and a glance over the shoulder is, in my view, quite in their national character.
Strange to say, the original anthem has five more verses.
The
question mark in verse one is a rhetorical device to get you tense
enough to read the rest. Now, at the end, of three of the other verses, you will be relieved to
hear, the flag does in fact still wave,
o'er the land of the free and the home of the brave.
But I've never, ever, heard an American sing anything but the first verse.
So I'm going to try an experiment- ask 20 Americans what is the last line of their national anthem. Watch this space.
Sat Sep 29 18:47:17 PDT 2007
We'll have a gay old time
Seem in Baltimore harbor- cargo ship with a fabulous paint job.
As a matter of fact, the use of this color scheme was discontinued after the sister ship to this vessel had a near miss with an amorous blue whale.
They decided to stay just good friends and meet for coffee, whenever they're both in town.
Sat Sep 29 18:41:57 PDT 2007
R.I.P. Uncle Gordon
My uncle, Gordon Donaldson,
was an assoc/prof of electrical engineering-
actually lectured me when I was a first year. I got a poor grade, much
to my shame.
When I was 12, he tried to teach me Boolean logic, down on his farm. "See this gate", he'd say, "and the next one? Well, by the time we are through both gates, we have opened one plus one gates to achieved one thing. So one plus one is one".
He really tried to teach me lots of cool stuff, but I didn't get it. Actually, I think I learned more from the example he set, rather than anything he tried to teach. He had a super-conducting lab across from my first year EE lab and sometimes there'd be this "whoosh" sound followed by "oh dear...".
Some unplanned electrical spike had just boiled away one-tenth of Gordon's annual supply of liquid helium. I'd poke my nose in to see a mushroom cloud rising from this metal can in the middle of the floor, expanding up and over the ceiling. Underneath, Gordon and his grad student, sat there shaking their heads. Back to the drawing board.
Gordon showed me that science was a very human endeavor. His lab was filled with strange hand-crafted breadboards and wires higgeldy-piggeldy everywhere. "Come look at this," he'd say, "and showed me his lab note book- all covered in scrawl and rough notes and coffee stains. "Imagine it- from this we write international standard research papers. Boggles the mind."
He showed me also what it means to be an international academic. First time
I can remember meeting him, I was seven years old, living in Canada, and he his way to Oak Ridge for a sabbatical on
fusion research. He had his whole family with him and I saw
(rather than I was
taught) that academia meant that a larger world was open to you and your family,
through out your career.
He had this farm back of back of Canberra. It had no power, no farmhouse, no phone, no plumbing, and the road would wash out several times per year. For a homestead, there were three shacks down by the river at a place called Wombat flats. One shack had a door, one shack had insulation, and one had four bunks.
For a kitchen, there were two large hoops at right angles- one on the ground with paving stones and one over the top to hold the pots over the fire. We'd sit there, rain or shine, squatting on logs or folding deck chairs, brewing tea, frying eggs, toasting bread. Of a winter's evening, we'd sit all rugged up, sipping tea, stirring the pots.
(Here's my lovely Aunt Jane, in the kitchen with her tea and biscuits. Behind
here are the water buckets we brought up every day from
the river.)
No one ever told me that this was all totally insane so I just thought it was natural- what every adult does on their weekends. Up every morning to dig thistles out from the lake paddock and spread the superphospate (fertilizer). Round up cows to dose them against heart worm. Dig a new latrine hole every few days. Bathe in the river or, in winter, under a little shower that was can with some holes dribbling warm water from the camp fire.
Uncle Gordon died last month. He was in his eighties and had been sick for a while. I'm sad he's gone but glad that he isn't sick anymore. I don't really understand his legacy to me but I thank him for the example he set.
Now, every time someone comments that I am a little unusual, I just think "well, in my family, I'm kinda dull."
Sat Sep 29 17:36:58 PDT 2007
New Talk: Software Development Cost: How Much? You Sure?
"Cost" is a quality issue. If development costs is underestimated, developers will be forced into many quality-threatening cost-cutting measures. A major reason for poor software cost estimation is that NASA's managers don't have enough relevant information. Also, current cost models are brittle and improperly tuned. We fix all that.
Sat Sep 29 17:29:21 PDT 2007
New talk: Improving IV&V Techniques Through the Analysis of Anomalies
We seek an active monitoring framework where data collected during an active IV\&V project will trigger an alert if a project becomes "unusual" (and defining "unusual" is one of the goals of this project).
Sat Sep 29 17:25:21 PDT 2007
The key to it all
When I was 12, the pinnacle of hi-tech Sci Fi was Crichton's "The Andromeda Strain". And this was the key to it all- the arming device of the nuclear bomb that was meant to blow our heroes away if the alien bug got out.
One of the technical consultants on that film was Bill Koselka. Then, he worked at JPL and did things like shot lasers at butcher's meat to show the art department what a laser burn would look like.
Anyway, these days Bill works at IV&V and he's been bragging for months about his role in the film. I didn't believe it- I wanted proof. So, to secure his place in history, Bill obliged. "The arming device for a fictional nuclear weapon good enough for you?" he asked.
Mon Sep 24 06:47:27 PDT 2007
Young hopeful (?hopeless) writer, lost in the big city
Here's a photo from the vault. That's me on the left at a conference I went to when I was 16.
I don't know how the mother hens at this conference let a 16 year old get his hands on whatever I am drinking in this picture. Nor do I know why I thought that particular outfit was worth wearing or who was the young woman next to me. But I can guess what she was thinking and so can you so we'll just move on.
In 1976, the Shopfront Theater for Young People ran a young playwrights conference. I was quite the hopeful writer in those days and sent them some film script idea I had about someone doing long distance running late at night. At the time, I was doing that sort of thing and the script tried to capture what I felt, running around each night. There was something strange and powerful about running in dark places with your heart exploding out of our chest while you beat down dead every steep hill in town.
How well did my script capture that? Badly! I worked the thing out in my head, then wrote it out, sorted by scene location. So anyone reading it had to play "follow the link" to pull it out into a linear sequence. Regardless, the conference took the script and invited me down to Sydney.
At the time, I was 16 and living with my parents in a small country town. Coming from that, this conference was a blast! 70 like-minded fellows with their sleeping bags sprawled all over the floor of some big lounge room. No parents! Trains late at night into the Nimrod theater to see a REAL PLAY. Meeting REAL directors and REAL playwrights. So stimulating and overwhelming and overloading! I got 4 hours sleep in the three days there. Even feel asleep one afternoon when five of us had a session with some young great playwright who was twenty, but just had a PLAY on at the NIMROD. I awoke, bleary eyed in an empty room going "hey- where did everybody go?". Still don't know if my snoring drove them out.
After this three day blast, I returned home to months and months of unimaginative studying. When that was all done, I moved back to Sydney and hung out a lot at Shopfront. In fact, I matured into a mother hen looking after the 1981 Young Playwright's Conference. I was the chef that year- this was when I was going to cook for a living (another story for another time). First night, me and one of the other mother hens took a gang of 12 year olds into town to the see "Indiana Jones and the Raiders of the Lost Ark". I gotta say, that was the BEST crowd to go see that film with!
So did I go on to become a great writer? Nope. I grew up to be an academic who only wrote technical papers for scientific journals and the occasional web page. I went to many other conferences, but none as fun as that first one. But the Shopfront experience gave me a lot of good stuff. Like acting in some Harold Pinter plays in front of a real paying audience. That final speech of the barman when he explodes and collapses: wow. And I think I NAILED it (IMHO). Some of that dramatic training survived. I don't have any fears of speaking in public. I learnt to give flamboyant conference talks and energetic (over-energetic? rambling?) lectures.
Through Shopfront, I ended up living in a tiny tiny flat-with-no-clothes in Short Street with Carol and Clare (of the strange but successful study habits). Clare saved me from a life in the suburban flats around UNSW for which I will always thank her.
Shopfront also introduced me to the world of politics such as the great (failed) assault on the Shopfront board of directors, 1984.
But in an alternate universe the person in this photo went on to a career as a writer. The great fresh-faced playwright acclaimed throughout the land whose words make the middle class wince with their biting accuracy. Surrounded by women friends who are both attracted, yet repelled at his problems with drinking and sleeping. Who can't write a simple idea in a simple linear manner. With bad dress sense.
Sounds like I did the world a favor by doing something else, heh?
Sat Sep 22 08:37:21 PDT 2007
I itch
Helen took out some poison ivy in the yard, then left for Baltimore. The cat, bless her, inspected
the carnage then came to me for some loving. "Mummy is not here", she whined, "who will worship me?".
Well, it turns out that Lucie the Cat is a vengeful god. After I preened and petted her, all the poison ivy has made my arms and chest erupt in little welts that feel so GOOD to scratch then scream at you for hours afterwards.
Helen! Come back! Face the wrath of Lucie! And let the garden grow, dammit!
Sat Sep 22 08:25:26 PDT 2007
Ain't nothing wrong with Baltimore
Helen and me are playing city mouse and country mouse. Weekends are spent either here or
in Baltimore. Helen has a shorter working week than me so our current mix is two weekends in WV and the
third in Baltimore. Unless I screw up the Baltimore freeways, it is 4 hours from here to there.
I was really worried about this arrangement, before it started. Now, it all looks good. Sure, the house deflates a little when she leaves but its only for a few nights then we are back together. And her new flat is wonderful and the food around where she lives at Fells Point is just sensational. Yuppie over-the-top wall-to-wall bars/clubs/bistros/ etc etc etc and to hell with the diet.
I might have a different view on all this, come winter. The road from WV to Baltimore is I-68 and it is a mean mother in winter- isolated, scary, random animals blundering onto the freeway, blizzards, the whole nine yards. But, right now, it is as pretty as all get out. And fall hasn't even started yet so we're going to get a month of riotous colors. There's something to look forward too.
Sat Sep 22 08:18:05 PDT 2007
Busy as a bumble bee
Rate of blogging falling off. What can I say? I think "start of term" says it all. In the last six weeks I have
Summer is now a distant memory. A halcyon time when hours flowed gently between cups of tea and pleasant bicycle rides discussing the zeta function with that nice Mr. Turing. If only he'd tinkle his little bell once more, I'd be off again in a shot.
Sat Sep 22 08:04:40 PDT 2007
New paper: Software Effort Estimation and Conclusion Stability
This paper revisits the "conclusion instability" problem identified by Kitchenham, Foss, Myrtveit et.al.: i.e. conclusions regarding which software effort estimation method is "best" is highly contingent on the evaluation criteria and the subset of the data used in the evaluation. Using non-parametric methods (the Mann-Whitney U test), we show how to avoid conclusion instability. This paper reports a study that ranked 158 effort estimation methods via three different evaluation criteria and hundreds of different randomly selected subsets. The same four methods were ranked higher than the other 154 methods regardless of which evaluation criteria or data subset was applied.
Sat Sep 22 07:31:18 PDT 2007
Talk: Data Mining, Truth, Justice, the American Way, and the Flying Spaghetti Monster
President Bush advises that students should be exposed to many different schools of though. In this matter, he is less than totally correct.
An important part of science is to expose people to the critical and continual (re)evaluation of ideas. Otherwise, any nonsense can be shoved down our throats such as pirates are causing global warming.
So this talk concerns notions of certainty and standards for debate. One surprise will be that nothing is "true" (in a scientific sense) but many more things are false. Implications for humility, tolerance, and justice will be discussed.
Sat Sep 22 06:56:42 PDT 2007
Are our systems "safe"?
SAVVY is a joint venture between Global Science Technologies and LCSEE WVU. A NASA STTR proposal, SAVVY addresses the issue of "our our systems safe"? SAVVY watches an operating system and learns what "abnormal" means. Incremental Bayesian anomaly detection then alerts the operator if a problem is seen and contrast set learners propose repair actions.
SAVVY has an ultra-low overhead: it is both highly memory efficient and can operate using a single pass through data.
Abstracts accepted to the 2007 COCOMO forum
Two abstracts on 21st Century Effort Estimation and the STAR decision support tool have been accepted to the 2007 COCOMO forum. The authors are the usual suspects from WVU ( me, Dan Baker, Omid Jalali, Our Elwaras) and JPL (Jairus Hihn, Karen Lum).
Sat Sep 22 06:39:14 PDT 2007
New project: Understanding STEP
STEP is a standard for engineering documents recorded in ISO10303. This project would explore generative design and the use of old designs to inform new ones.
Generative Design
Tools for accessing those descriptions would be assessed. Candidates tools include:
Sample assemblies would be created, with their own particular design goals. Issues with the above process would be documented. If required, extensions to STEP would be offered to enhance the ability of STEP to generate new designs from old parts.
Old informs new
The first step of this part of the project would be to generate ontologies from the step procedures. This hierarchy can then be used to map rules of thumb from one design to another using standard semantic rewrite rules (rules going up the hierarchy must be generalized, rules going down must be specialized). In this way, we could, e.g. move a rule of thumb from car braking to aircraft design:
Fri Aug 24 11:58:02 PDT 2007
Talk: when good data goes bad
Thu Aug 23 08:09:02 PDT 2007
Applied Technology Lab a "Virtual" Success
This summer, the NASA/WVU Applied Technology Lab hosted four interns from Fairmont State University, Shepherd University, University of Idaho, and WVU, all working on leading-edge software equality assurance tools for the NASA.
Project leaders
- Wes Deadrick, NASA IV&V- Justin Morris, NASA IV&V
- Dr. Tim Menzies, WVU
Tools
- REACTIS- SAVE
- SDA
- Robustness
- SpecTRM/ RA
Interns
- Adam Anderson, SU- Robert Ball, FSU
- Jonathan Meyers, WVU
- Ben Ridgway, UI
WVU support
- David Krovich, WVUPhotos
"This was the first operational test for the lab", said Wes Deadrick, the IV&V NASA lead on the ATL project. "We wanted to see if the lab had the capability to support a wide variety of tools."
"We also wanted to check if the lab was flexible enough for the different work habits of students," added Justin Morris, the NASA intern summer supervisor. "Students tend to work irregular hours and can't always get into some central lab."
Remote access to the ATL was leveraged by the students this summer. Often, students weren't even in the lab, choosing instead to work from the NASA Facility at Fairmont, 20 miles to the south. There, they could have regular contact with their NASA clients while still having virtual access to the resources of the ATL.
"Even the systems administration uses virtual access," says David Krovich, LCSEE systems supervisor. "Once the next round of updates gets in, we'll be able to run those wipes from the Evansdale campus, even though the lab is downtown."
"This is the age of the Internet", said Tim Menzies, an associate professor at WVU's Lane Department of Computer Science and Electrical Engineering (LCSEE). "People can work, virtually, anywhere. On the Internet, ideally, there is no there, there. But reality sometimes intervenes. Some of the tools explored this summer had 'smart licenses' that rejected login via virtual access.
"And that's no problem,", said Deadrick, "The lab can handle virtual access, and if we actually need to go there, it is a good working environment. Its a comfortable place to hang out- big meeting tables, a couch, coffee making facilities, ten dual boot windows/Linux machines, a printer. all that good stuff."
"The lab serves an important function," added Deadrick. "One of the goals of the NASA IV&V Facility's Research and Development program is helping the infusion of bleeding edge technologies into NASA. These interns are the support team for these new tools."
While a comfortable environment, the lab is also secure. Swipe card access secures the doors. All the machines are wiped, on a weekly basis, and fully re-installed from a master image that is strictly controlled.
"It's right downtown in Morgantown," added Morris "so it really benefits from the relaxed and energetic university atmosphere."
The interns spoke highly of their summer experience. Robert Ball, a Computer Science senior at Fairmont State University, remarked that "Overall this has been an excellent experience and I would gladly do it again. This internship will no doubt be an instrumental part in building my career and portfolio."
"The tools that we reviewed provided us with experience in producing technical documents that could aid in software engineering work being doing within NASA", said Jonathan Meyer, a WVU computer science senior.
Adam Anderson, a computer science and mathematics student from Shepherd University, added "This internship has given me an opportunity to learn and improve on many skills, and I am extremely grateful to have been given the opportunity to work at the NASA IV&V Facility."
The tools
The ATL seeks to assist NASA to better assure the quality of their software. Software is built in cycles. Early in the life-cycle, the emphasis is on checking that requirements are correct or that the overall architecture offers the right functionality. Later on, the emphasis moves to code written by programmers.
New life-cycles are always being created. For example, a recent trend in software engineering is model-based methods where programmers don't code- rather they sketch the requirements and the code is auto-generated.
What ever the life-cycle, one constant is that any development method requires management oversight and process control
This summer, tool support for all these cycles is being explored at the ATL.
Process support with SDA
SDA is being researched by Ben Ridgway
The Software Developer's Assistant (SDA), by Tietronix Software Inc, is a web-based program which aims to provide tools and information to assist software developers to develop quality software.
SDA isn't a coding and analysis tool. Instead, it is a time and process management tool. Its purpose is to help delegate and manage the multitude if individual tasks which must be completed in a large development environment. All in all, it should ease the burden of remembering and adhering to a software development process, while still reaping all of the benefits thereof.
Such a tool is quite important to NASA due to the often complex design process that goes into the creation of high assurance software systems. In the NASA environment, there are many pieces to the development puzzle which much fall into place in a certain order. The SDA should in theory provide a good deal of assistance at every step of the process, and thus help NASA create better software.
Robustness Testing
Robustness testing is being researched by Robert Ball
Architectural analysis with the SAVE tool
SAVE is being researched by Robert Ball
Architecture is one of the most crucial artifacts in a software system. The architecture defines the system's components and how they interact with one another, and it also facilitates the reuse of code and other components.
Architectures are important in the business world where time is money and reuse of code can save valuable time. While architectures are very important, they are also difficult to follow, and commonly there is some "shift" from the original model.
The SAVE tool analyzes code and allows developers to avoid this shift. It allows the developer to compare the planned or intended architecture with the actual product architecture, and identifies architectural violations so that they can be corrected.
It saves time during reviews because it highlights areas of the software that need the most attention. These processes can be completed manually. However, using SAVE, removes this error prone and tedious process while saving valuable man-hours reviewing and modifying incorrect programs.
Model-based reasoning with REACTIS
REACTIS is being researched by Jonathan Meyer
REACTIS is a tool designed by Reactive Systems for the testing of models.
Software Engineering has evolved over the course of the last 40 years to meet the need in software development for reliable and efficient software. The software in use today and that which is continuing to be developed is of such large magnitude that the methods employed by early developers are no longer feasible.
Model-driven development has emerged as a strong contender in the field of software development as a result of the visual approach to programming. With Model-driven development, systems are created using graphical and textual models and then code may be generated from these, as well as allowing testing to be done on them.
REACTIS provides automation of the testing and validation portion during the development of embedded control software. REACTIS consists of three components:
REACTIS is a tool that has the potential to greatly reduce the time and expense involved in any project using Simulink/Stateflow models.
Requirements analysis with SpecTRM and RA
SpecTRM/RA is being researched by Adam Anderson
The main purpose of SpecTRM is to help with the development of software systems, particularly ones that are safety-critical. SpecTRM provides a new, simplified approach to organizing system specifications by using the SpecTRM-RL formal modeling language, a specialized executable requirements specification language, and intent specifications, which explain the system specification in detail, or in other words, what it does, how it works, and why.
After a complete requirements specification is created, SpecTRM can simulate the systems behavior directly from the requirements. The main benefit of SpecTRM, as its focus is system requirements and specifications, is that it can help detect errors and potential safety hazards early on in the development phase of the software. This brings system safety into the development process of the system and can allow detection problems before they become more costly.
Requirements Assistant (RA) seeks to analyze requirements written in a natural language. The program requires a text document and a set of input files containing the aspects that the user wishes to evaluate. The program outputs a file containing a list of defects and remarks that can then be considered for improvement before the document is needed in the next phase of development.
The Workers
Adam Anderson
"To err is human, but to really foul things up requires a computer."
-- Farmer's Almanac, 1978
Adam Anderson was born in Suffolk, England in August of 1986. In 1990 he moved to the United States and graduated from Martinsburg High School in 2004.
He now attends Shepherd University in Shepperdstown, WV and plans to graduate in 2008 with a degree in Computer Science and Mathematics.
Adam works on SpecTRM & "Requirements Analysis" tools
Robert Ball
"The bricklayer checks her courses with a level, the programmer with a
test"
--Bezier
"A computer without COBOL and FORTRAN is like a piece of chocolate
cake without ketchup or mustard."
--John Krueger
Robert Ball is a senior at Fairmont State University in Fairmont, WV. He is pursuing a degree in Computer Science (and a degree in Computer Security, with a minor in Mathematics).
His current research area is steganography, which is the art of hiding data inside of other data, i.e. a text file in a jpeg file. He has also completed work with both white-box and black-box testing.
Currently he is an internship with NASA. Previously, he has completed internships with Backbone Security and the Institute for Scientific Research (ISR), both in Fairmont.
One day, he hopes to work in the video game industry.
Robert works on the SAVE tool and robustness testing.
Jonathan Meyer
"If you give someone a program, you will frustrate them for a day; if you teach them how to program, you will frustrate them for a lifetime."
-- Anon.
Jonathan Meyer was born not far from Lake Michigan in 1984. After moving to West Virginia at the age of 6, he soon became interested in computers and what made them work. While playing with his father's old IBM XT he gained invaluable experience in BASIC programming and the inner workings of DOS.
During Jonathan's teen years he ran a small web design business, in addition to being organist at 2 of the local churches.
Currently entering his Senior year at WVU, Jonathan is pursuing a degree in Computer Science. During his spare time he enjoys reading, playing the piano, and programming.
Jonathan works on REACTIS tool.
Ben Ridgway
Research is what I'm doing when I don't know what I'm doing.
-- Wernher Von Braun
Ben Ridgway is entering his senior year of Computer Science/Information Assurance at the University of Idaho in Moscow, ID. He is the recipient of the National Science Foundation's Federal Cybercorps: Scholarship for Service. The SFS program's aim is to bring top Information Assurance students into the Federal Government.
He was raised in Hamilton, Mt in the shadow of the Rocky Mountains. His interests are downhill skiing, motorcycling, and naturally computers.
It has been his goal to work for NASA since becoming hopelessly hooked on Star Trek TNG as a child.
Ben works on the SDA tool.
Photos
At the end of summer, NASA invited the interns down to the beltway for
training on some advanced software tools, plus a guided tour of the NASA
Goddard Space Flight Center (GSFC). The training provided the interns with
an opportunity to meet and work hand-in-hand with highly respected
researchers.
The GSFC tour highlights were seeing the largest clean room in the free world, seeing a Hubble Space Telescope mockup, visiting the Science on a Sphere display, and visiting the mission control center.
Both
experiences were both educational and inspirational.
Thu Aug 23 06:24:45 PDT 2007
Johnson Space Flight Center expresses interest in Treatment Learning
Historical note: The TAR3 "Treatment Learner" is a special kind of minimal contrast set data mining developed by WVU researchers using funds from NASA SARP. For more information, please see Data Mining For Busy People.
From memos between JSC and AMES research center:
Up to this point, we've focused on demonstrating overall reliability of the skip entry, without too much concern of quantifying the underlying dependencies between the simulation parameters and specific results, e.g., sensitivities of fuel usage, loads, accuracy. Your last paper identified a couple of techniques that might help us quantify the critical dependencies:
Source: Karen Gundy-Burlet, Ph.D. Research Scientist NASA-Ames Research Center, M/S 269-3 Moffett Field, CA 94035-1000
Date: April 13, 2007 2:03:11 PM PDT
To: Karen Gundy-Burlet <gundy@email.arc.nasa.gov>
Subject: Re: ANTARES Software Release (Hill)
The work being done by the Robust Software Branch to support parametric analysis of large data sets has shown promise in real- world applications. As computer performance and disk storage capabilities improve, the problems of reducing and analyzing very large data sets will become increasingly challenging. The Ames RSB team has been working with Orion GN&C engineers to help reduce Monte Carlo results for both ascent abort and entry mission phases. They have used clustering and other data mining approaches to find interesting correlations between input parameters and analysis results. They have also worked with GN&C engineers to investigate visualization techniques that support data reduction and improve evaluation of these large data sets. The application of data mining to engineering analysis has shown promise, and has potential benefit across many different applications.
-Bill
Tue Aug 21 20:04:56 PDT 2007
The Horror of Drive-Thru
#include pinchOfSalt.h
Somethings are so common, we don't give them a second glance. But sometimes, if we do, the shocking truth is unavoidable.
Consider trampolines. Or, more generally, the human obsession with gymnastics. Why? Is it just we like looking at trim, lithe bodies in scant costumes performing physical feats for our enjoyment? Obviously not. Clearly we are fascinated by gymnastics since it lets us act out race memories. I refer of course to all those generations spent in the zero-G colony ships that first brought us to Earth.
Not convinced? Then let us apply the glare of reason to our darkest psychosis.: drive-through food on highways.
Now what is that all about? It can't be be food- its too fatty, destroys family life, and the coffee is crap.
Not only that, think of the ridiculously complicated infrastructure we have built and maintain, just to get us to our daily McDonalds:
So, why do we do it? After minutes of research I have made the following incontrovertible conclusion:
- Life on
this planet was seeded by a race of very thin invisible ribbon
people with a fondness for red meat, fresh off the grill. This
race wrote into our genes the overwhelming desire to
build the ultimate drive-thru.
Crazy, you say? Well, the evidence is everywhere. Ever notice how confetti blows in the wind? And power lines run alongside the highway? Now put it all together. The ribbon people wrap themselves around the poles and wires to stop the wind blowing them away from their food.
And what is that food? Why- us! Every day, we jump in our cars and spend hours driving around in front of the homes of our masters, the ribbon people. "Pick me! Pick me!" is our subconscious plea. And if our cars are pretty, bright, and shiny enough, or if our driving antics are daring enough, then we are plucked by the ribbon people who grill us on the power lines.
Of course, that was all stage one. We're now into stage two. Ever notice that there are more roads in the city? And the buildings are taller? And where the buildings are tallest, the traffic is slowest, the crime rate is highest, and there are more ticket-tape parades? Need I say more?
OK- if you need it all laid out for you. The ribbon people, tired
of the high winds of the open highway, have forced us to build
huge windbreaks. Oh, the fiendish cunning of their plans. Every
day we do our commute, carrying more and more of them into the
downtown area. Then we mill around there, in huge tempting
crowds. They pick and choose which of us to eat- and hide their
carnage within the usual death rate of downtown.
And sometimes they even flaunt their presence by raining down on
us during parades. And like the fools we are, we run through the
ticker-tape, laughing and cheering while ignoring the sizzling
smell of our hapless fellows grilling over the power lines,
feeding the monsters that feed on us.
Every day we play out our innocence, hiding, lying to ourselves, for the reality is unbearable. We struggle to raise ourselves above our murdering masters. But our imagination is stunted by the horror of our existence. All we have ever managed is pale parody of the slaughter yards of the ribbon people.
Worst of all, in our drive-thrus,
the participants gratefully pay to join the massacre
while ordered to "have a nice day. Now do you understand why Ronald McDonald's smile is forced and
his mouth is ringed in blood?
Oh the horror, the horror.
Mon Aug 20 21:12:14 PDT 2007
Welcome to the start of term
Can't even begin to list last week. Such a blur of activity.
Had this explosive idea about using a BASH programming trick to build interactive tutorials. Its really going off in my head right now. And sooo simple to do.
I got obsessed with writing web sites as reports of XML files. Did same for the cs591o data mining web site. It was very exciting, and code was so succinct. But I am numb from the mono-mania required to get it all together. Tired am I. But proud of the product. It looks like a normal web site but under the hood is some sweet XML.
There was some strange bug keeping me up at nights- I could not reproduce my Jan 2007 TSE results. Last week I found the bug and now I can do things like (a) reproduce old work (b) try it on more data sets; (c) try more learners, etc. Very exciting.
Still getting ready for class next Tuesday. Eeek! that's tomorrow!
All I can say is "bring back summer".
Mon Aug 20 20:59:54 PDT 2007
WVU data miners deployed at JPL
The "2cee" software effort estimation tool
combines Monte Carlo simulations with feature and instance selection to infer the range of possible software development
costs.
This summer, WVU CS graduate student Dan Baker spent the summer building software effort estimation tools for NASA's Jet Propulsion Laboratory, Pasadena, California.
Effort estimation is an issue that plagues much of software development. Getting the initial estimate wrong can be disastrous- when the money starts running out, the first thing jettisoned is (usually) any quality assurance work. As a result, under-tested quirky software is often released.
This problem has long been recognized at JPL. Numerous effort models have been developed- some of which required a significant development effort by scarce personnel.
Using data miners developed at the Lane Department of Computer Science and Electrical Engineering, Mr. Baker was able to show his tools did better than current JPL methods.
"Our results are great. " said Dr. Tim Menzies, associate professor at LCSEE and Mr. Baker's supervisor. "JPL wants to use Dan's tools alongside their existing methods for a 12 month trial period."
"After that," he added, "there is a very real possibility that after that time JPL will change their current effort estimation methods and rely solely on tools developed by WVU."
JPL scientists were most impressed with WVU's training.
"Dan's technical skills and professionalism were
exceptional, " said Dr. Jairus Hihn,
head of effort estimation at JPL.
Said Hihn, "His task was to turn a data mining rig that only a CS major could use into a windows based tool with an analysts mode and a simple estimation mode that even a manager with very little training could still use. Many times he was caught between Professor Tim (try this new idea) and Manager Jairus (get the job done). This was surely a most trying experience. In the end with only a little help from his friends, Dan pulled off the near impossible. We now have a tool that works and it loads with one click. We estimate that the model will save us months of work."
Mr. Baker was also very positive about his summer experience. "Working at JPL was a very surreal experience," he said, "To be surrounded by so many intelligent people on the frontier of human exploration made every day seem like a science fiction novel. Working closely with the JPL business users was very productive, both in development of the new estimation tool and in the research. I met some great people at JPL, and living in California was fun as well. "
For more on this research, see here.
Mon Aug 20 20:41:14 PDT 2007
New research project with GrammaTech
One of the long standing problems in software quality assessment is the rapid recognition of dangerous software. Our feature extractors for software are decades old- lies of code measures, simplistic intra-module metrics, etc. Working with GrammaTech's advanced compiler technology, WVU researchers will seek better measures, that can be quickly generated earlier in the coding life cycle.
Tue Aug 14 20:44:17 PDT 2007
BOOK(S): Val McDermid's Tony Hill novels
Welcome inside the mind of a serial killer. Are you enjoying the view?
Val McDermid writes tartan noir; that is, crime fiction from the land of kilts and haggis. These books are broody and , says Wikipedia, dwell on the good and evil within our souls, redemption and damnation. For example, I've just finished four of her Tony Hill/ Carol Jordan novels:
The books are formulaic. In every one, Dr. Tony Hill, famed international clinical psychologist, has at least one scene where he is at the mercy of yet another psycho (and D.C.I. Carol Jordan has to save his sorry ass, yet again). In every next book, he's sworn off his old ways yet McDermid finds some way of dropping him back into the fray.
The formula is the back story. The real story is that Jordan and Hill are hopelessly in love. They orbit each other, unable to get closer, desperate when they are too far apart. They share a love of serial killers. Jordan wants to find them and hang them (while, not-so-incidently, getting promoted). Hill wants to climb inside their skins so he can finally understand what makes people (like him) distant from the rest of humanity.
When one accepts a dangerous assignment or is damaged, the other gets drawn in and shares in the bleeding. By the way, I mean that quite literally- these books are not for the faint of heart. There is a horrifying completeness to McDermid's terrorverse. She'll write about the stuff that other male crime writers balk at.
In the Dexter books, murder is fun, a pixie-like dance through a wonderland of blood splatter. But in the Hill/Jordan world, no one is laughing. In fact, many folks end up bound and gagged, squirming and trying to scream, as their doom stalks closer and closer and raises the knife while crushing their arm in the vice and... well, you get the idea.
Sun Aug 12 08:07:52 PDT 2007
BOOK(S): Roddy Doyle's Barrytown Trilogy
Relentless poverty. Foul mouth Irish. Little hope for the future.
Reproducing like rabbits cause they are old-school Irish Catholics and there is fook all else to do.
In the first book, the young'uns start a band (The Commitments) and it fails. In the second, the sister of the founder of the band gets pregnant and has The Snapper. It the third, the girl's father learn the evils of capitalism and fast food as The Van tears apart a friendship.
The Commitments was fun. Very short. Had a story to tell, told it all, stopped.
The Van is described as a comedy but I just saw waste. Wasted lives, talent unexpressed, no resources to build a better tomorrow. I thought it was a good book, but not one I enjoyed.
The Snapper was all about loyalty- the girl's parents are upset by the news but they stand by her. Dad takes her down t'pub for a drink. The girl's friends down t'pub are explosively supportive. Come to think of it, the girl got pregnant in the car park of t'pub. That's some t'pub.
In the USA, this kind of poverty sends people over the edge. Street gangs, wars between drug lords, drive-by shootings, jails bulging at the seams. But in Doyle's Barrytown, Irish poverty has a gentler, kinder side.
Doyle writes a world where there are no drugs, families and communities stay together to support each other, and nearly no one goes to jail or gets shoot. I don't know if that world really exists, but I'd like to think so.
And Barrytown's poor remain proud, even powerful. The heroine of The Snapper turns on the father of her child who's been bad mouthing her around the town. She goes to his house and orders his silence, or she will reveal all to his wife. In American, this scene would involve guns or lawyers. In Barrytown, its two people hissing at each other over a cup of tea.
Sun Aug 12 08:06:43 PDT 2007
BOOK(S): Jeff Lindsay's "Dexter" series
A nice bit of light-hearted homicidal mania
In Darkly Dreaming Dexter, our favorite psycho works as a forensics expert for the Miami police. Dexter is a well-adjusted psycho. By night, his "Dark Passenger" emerges but, under orders from his dearly departed father, Dexter only unleashes the passenger on those that deserve the most horrible of deaths. By day, this bloody thirsty killer spends his time happily gazing at blood splatter- and gets paid to do it by the Police Department.
In Dearly Devoted Dexter, a killer is leaving grossly mutilated, but still living, victims all over Miami. Dexter's unique insights into psychopaths is desperately needed to catch the bad guy. Meanwhile our hero is being stalked, by the only cop in Miami who wonders why this guy understands so much about psychos. So Dexter has to lie low which, for Dexter, means playing the game of being normal. Sublimating his Dark Passenger, all Dexter's energies go into pretending to love the girl friend that he keeps for camouflage. He pretends so well that.... no, I can't say. But man it was cool.
Sun Aug 12 08:06:07 PDT 2007
Never fly with U.S. Airways
Oh
the horror, the horror, that is U.S. Airways customer service....
Philly. I'm still only in Philly. Waiting for a plane flight, getting softer. Every minute I stay in this room I get weaker. And every minute my luggage squats in a hanger, it gets stronger.
Everyone gets everything he wants. I wanted a flight out of here, and for my sins they gave me one. Brought it up to me like room service. It was a real choice flight, and when it was over, I'd never want another.
It had been a perfect trip. England, ASE PC (two full papers accepted, yee hah!), London all vibrant and exciting. Then time to go home.
52 hours later I was still lost in transition, still wearing the same clothes (you don't want to know), and constantly amazed at the awful customer service at U.S. Airways.
When I got in from London, all flights west from Philadelphia were cancelled (very bad weather). Called U.S. Airways customer service- what a disaster:
I was told that there were no other flights for two days (massive cancellations). So, like I always say, if life sends you lemons, send it back and ask for a whiskey sour:
Well, I got the bed but it turns out that the hotel was nowhere near downtown Philly. Not sure if it was anywhere at all really. Some generic Pennsylvania. town way out beyond the burbs. Really humid outside and I was no change of clothes (lost luggage, remember?). So I hid in my air con room and did lots of hacking. On the whole, not a bad way to spend a few days.
But hacking bliss melted away when I called U.S. Airways, just to check my booking. Turns out, U.S. Airways lost the booking. I spoke to Pari (or was it Darpana?) who told me:
At this point, I climbed onto a monumentally high horse and asked to speak to a supervisor. Madhuvanth (or was it Reshma?) left me on hold for about 20 minutes then came back with, wait for it, the flight I was offered in the first place.
Its hard to pick the worst moment of customer service but, overall, I think it was the "help" I received from the lost luggage folks. What was most fun was their overall tone of "its your fault if we lose your bags". When I asked about my luggage, I was told that they could do nothing without a reference number. I asked for a reference number and was told that I had to contact a U.S. Airways representative.
"But I am speaking to a U.S. Airways representative", I said.
No- apparently I have to go to the web site and start a query.
"Oh", I said thinking "don't you guys work for the same airline?".
So I logged onto the U.S. Airways web site, recorded my lost luggage. The next day I got this email:
Dr. Menzies,
Thank you for contacting US Airways about your baggage.
We apologize you have been without your luggage, but
could not locate your claim under the reference you
provided in your original contact.
Please submit a new inquiry by entering either the
ten alphanumeric file reference(i.e. PHXHP12345), or
six character bag tag number, so we may access your
file. We look forward to hearing from you.
Patricia XXXXX
Baggage Call Center
Well Patricia, I have to say that "Menzies" is not a common name in USA and the space of "Menzies" arriving from London on BA67 last Thursday must be very very small. But I guess you can't be expected to find me. Its not like you have a computer, or you do this everyday, or that you work in baggage call center
(BTW, its not just me that thinks U.S. Airways has crappy customer service- it was rated nearly worst in a 2007 North America Airline Satisfaction Study.)
Sat Aug 11 13:38:51 PDT 2007
Proof that the English are insane
Behold the pickled body of Jeremy Betham, one of the founders of University College London.
Now I am not making this up. Here is the story:
Thu Aug 9 02:38:16 PDT 2007
In England, fantasies can be local.
Late at night I lie in my London hotel room, trapped by jet lag, and listen to the distant police sirens. They sound
so much more real than the ones i knew at home- they sound like the ones on the tele.
In Australia, fiction was elsewhere. I was 22 before I read a story where the hero light a cigarette under the Coca Cola sign at Kings Cross. Before that, Central Park, Trafalgar Square, the Mullholland tunnel, the Eiffel tower, or the Statue of Liberty featured larger in my fiction that any Australian icon.
The lesson was clear: Australia wasn't important enough to deserve its own illusions. If you want to dream, this desert must be deserted for New York, Paris, London, whatever.
So now I'm here in London and reading British crime dramas and watching BBC science fiction on the DVD drive. And the heroes climb the grimy steps of Barbican station (200 yards to the south) or race to save the world at the Tate Museum of Modern Art (a few miles away, on the South Embankment).
Once again, the lesson is clear. This place is important enough, mentally BIG enough for you to play our your wildest fantasies, all without needing a passport or a plane ride to a distant land.
Wed Aug 8 10:41:51 PDT 2007
M'Town to London Town
Racing to the airport and got there late. Wrecks on the freeway to scare you senseless, holding up the traffic. Impatient at slow lines at security, hoping from foot to foot. Door on the plane shut just as I took my seat. Phew!
Intense relief to be on the plane. Marvelling to be on a Boeing 777 (never flew one before) and to have my own LCD screen right there on the seat in front of me, complete with 18 channels, at 40,000 feet. Watching crap movies then realizing you have crossed over to Ireland and are nearly there.
London, 7am, sunny Sunday. Hotel won't let me check in till 2pm. Hyde Park, fall asleep, cuddling back pack.
Woken by
the police who were checking I was breathing (in this heat wave, some drunks like down and never get up- most awkward- lots
of paperwork).
Went to sleep in an empty park. Woke up midday surrounded by the British. They'd sprung up, like mushrooms.
So un-American.
Pale, gormless, and graceless. No American university dorm finishing school
to groom them into tanned plastic clones (with blinding smiles and
an illogical belief that Mexico is an foreign country). They (still) smoke in public. They have bad teeth,
anemic tans and
have ugly little dogs-
that walk without a leash (imagine that!).
I love London public transport. All those big red buses and bicycles and people walking to work. And the train
stations with the funny names: Angel, CockFosters, Elephant & Castle, St Pancreas (what random word dictionary were they using to
name them?).
But riding the tube? Forgettaboutit. After 3 days of riding, changing, riding again, I started walking. Turns out that London is a small town with pleasant strolls above and spaghetti tubing below.
Very little English in England. Maybe it was hanging round universities but I could walk around for hours and never hear the Queen's tongue.
Later on, at pub. So dehydrated, one beer got me drunk. Reading the newspaper on a leather lounge,
sunlight spilling in all around me. Where else would you want to be in the whole world?
Ate breakfast the next day at a greasy spoon. The menu was "set1", "set2", or "set3". Not quite spam, spam, spam, tomatoes and spam but near enough. Baked beans for breakfast (gross). Sausages. Fried egg. Coffee in a dirty mug and huge bowel of white sugar on the table, all serves not individually stored in separate sachets (so un-American).
Then onwards to days of meetings with academics from around the world. It is a simple formula:
Somewhere out there are some of the greatest Museums and Art Galleries in the Western World. But no time to see them. Thursday I fly home in time for my second dose of jet lag in 5 days.
Postscript: got back to the US of A and found massive thunderstorms between coast and home. All flights cancelled. All flights the next day cancelled. Spent a strange stray day in an out-of-town Philly hotel. A little day in a little place with just little old me. My own private wood between the worlds.
Wed Aug 8 10:26:09 PDT 2007
I'd forgotten about back packs
I'd forgotten what it was like for you to be the mule and the pack to be the bag. Many's the time
I've stuffed a month's worth of underwear into a backpack and headed off to north coast,
Tasmania, Victoria, western Australia, New Zealand, Bali, Thailand, Phillipines, Borneo, USA, UK...
The first time you put on a shiny clean new pack, you tie "the knot". The waist band is usually too long so after you sort out the buckles, you tie the excess strapping into "the knot". Four years later, you throw away the rotting stained back pack, and "the knot" is still tied.
In the meantime, you keep traveling away for a week of four. Every morning you wake up to ALL you stuff spread around some tent/hostel room and you wonder how it can all fit in the pack. On day 1, there is too much stuff to get into the bag. By day ten, stuff has flattened out some (or you have learned some packing tricks or a fourth dimensional portal has opened) so now the pack has spare space. Call it the Tardis effect.
Another effect is curious gravitational attraction between the very bottom of the bag and the thing you need the most. Whatever it is you need next, guaranteed, it will be at the bottom of the bag.
Nowadays. all my travel is over airport waiting rooms. No carry dust from Thailand to Borneo (now its more like Pittsburgh to Arlington). Wheels on soul-less little trollies. Convenient little rolling things with long handles but terrible back rests. No "alley... oop!" and you drag the thing up to your shoulders first thing in the morning. No "ah" as you let is slide/fall off your shoulders at the end of the day.
Wed Aug 8 10:22:08 PDT 2007
Two new papers accepted to IEEE ASE 2007 (11 percent acceptance rate)
One paper accepted was "The Business Case for Automated Software Engineering".
The other one, "Nighthawk: A Two-Level Genetic-Random Unit Test Data Generator" was written by Jamie Andrews and I consulted on the random search stuff.
Tue Aug 7 08:53:31 PDT 2007
One new paper accepted to IEEE ISSRE 2007 (30 percent acceptance rate)
"Fault Prediction using Early Life cycle Data"
This was one written with Yue Jiang and Bojan Cukic. We show that combining information sources from across the life cycle increases predictor effectiveness.
Tue Aug 7 09:06:44 PDT 2007
The PROMISE repository- an "important initiative"
In a recent paper on the future of software engineering at ICSE'07, international effort estimation
guru Martin Shepperd
writes about the PROMISE data repository:
