Flashes of Panic

Perfect record

The department staff forwarded around an email from a new grad student (starting in September) looking for housing. I was amused to discover that it was the student I’d convinced to apply last December on my otherwise-futile recruiting trip to Amherst and Northampton. Apparently if he was accepted elsewhere, they failed to make a better case for attending.

This also means I have a 100% recruiting success rate. Apparently if I talk to you about Computer Science grad school at the University, you’re as good as enrolled. I find this amusing particularly because a great deal of my argument was along the lines of, “Well, if you really want to do X, and you’re accepted at Y and the University, of course you should choose Y instead of us.”

Posted by pjm at 10:41 PM | Comments (0)

Lack of organization on your part does not constitute an emergency on my part

I think the title is all I should say about this one, actually.

Now Playing: Johnny Yen (Live) from Tomorrow by James

Posted by pjm at 3:48 PM | Comments (0)

Distributed responsibility

Thinking about graduations in the past few weeks, I also thought about everyone’s bête noir, the Class Notes, that quarterly litany of your peers’ successes and triumphs.

I imagined a commencement speaker…

“Those of you sitting before me will lead corporations and nations. You will invent, address the world’s great problems, save lives, and battle disease. You will peer beyond the edges of our world…”

Pause.

“You do understand that you don’t each have to do all of these things, right?”

Posted by pjm at 8:28 AM | Comments (0)

Missed a chance

Academic regalia always seems like a costly investment; Bog knows what’s going to happen to that cap/gown/hood now. I suppose there’s at least one possible future career path where it will come in handy.

I sort of envied the undergrads with their brown and blue tassels on their caps. (Not that I would trade my hood for one, of course.) I just had the anonymous black one, with an “07”, which I actually found faintly annoying—graduate students don’t really identify with a class-year cohort the way undergrads do. I wished, briefly, that I had thought to snag one of the tassels still hanging in my bedroom in my parents’ house. The “extra” one which I actually wore at high school graduation would’ve been a diplomatic faux pas at commencement, since only PhD graduates wear the gold tassels we had to indicate honors in high school, but the blue one I didn’t use at the time would have done just fine, even if it was the wrong blue.

And I probably would’ve been the only “92” there. My parents found a phone message about my 15th reunion when they got home.

Now Playing: I Turn My Camera On from Gimme Fiction by Spoon

Technorati Tags: gradschool

Posted by pjm at 9:42 PM | Comments (0)

Wrapped up

I haven’t yet written the triumphant “semester’s over” post I did for the last three semesters. That’s because this semester wasn’t one of those put-down-the-pencil-exams-are-over sorts of semesters.

My one class-for-grade was over on the last day of classes, and I expect to be hacking on my “masters project” for several days beyond Commencement. (Nobody is bothered by this, since I will technically continue to be a grad student here for at least a year of “leave.”)

On Friday, we wrapped up grading final exams for Programming Languages, a week after the exam was given. We poured the grades into a spreadsheet, generated semester numbers, translated those into letters, and put them in the online system. (I got a question about one of those within two days.)

We had two anomalies which I can safely discuss because I doubt I even have enough information to identify the students in question. First, one student who did all the homework and sat for the midterm, but didn’t take the final exam. I thought this was a bit unusual until we discovered that they weren’t listed on the class roster—not even as having dropped the class. As far as the registrar is concerned, this student was never associated with the course. We can’t figure out why they would bother doing as much work as they did.

Second, we had a student enrolled who didn’t hand in any work—no homework and neither exam. My guess is that they registered and somehow neglected to drop the course. There was no photo with the class roster, so we’re not sure if this student ever came to class.

Having willfully ignored all pleading from the registrar to register for courses next fall, I think I have only three interactions left with the University: commencement, my final paycheck (which may have already happened) and the completion of this project.

Now Playing: Deacon Blues from A Decade of Steely Dan by Steely Dan

Technorati Tags: grades, gradschool

Posted by pjm at 2:32 PM | Comments (0)

When it matters

Today I finally determined that I belong to the University’s School of Engineering. I think I probably could’ve reasoned this, but it’s not immediately obvious. The major divisions of the school appear to be between things like the med school, law school, vet school, sundry smaller institutions, and the “School of Arts, Science and Engineering,” which contains the majority of the local population of undergraduates, among others. I was pretty certain I fit in that, based on various bureaucratic clues.

The confusion comes when that subdivides into “Arts and Sciences,” which is basically the section most like my undergraduate experience, and “Engineering.” To further muddy the waters, Computer Science has undergraduate majors from both schools, unlike any other department on campus.

However, I confirmed that all our graduate students are under the school of engineering. (Otherwise, I might turn up too late to collect my degree.) Which is good, because as of this afternoon I am equipped for next Sunday’s ceremonies with an orange-trimmed hood. The interior is pale blue and brown, which does not look nearly as bad as one might expect.

Posted by pjm at 10:15 PM | Comments (2)

Working in translation

I think the reason I have put this project off as long as I have is that I am not comfortable working in Java.

One of the things I’ve learned from TAing the Programming Languages course this semester is how some languages require the user to think in a particular way. Prolog is perhaps the best example of this; students are used to thinking of functions as something which does something, and Prolog rules don’t really do anything. To write good Prolog, we needed to shift to thinking about conditions—X is true under the following conditions—rather than actions.

Java is not quite as dramatic as Prolog, but it does require the programmer to rearrange the way they think about the problem. I’ve spent more time in languages like PHP, or even C, where once I’d conceived of a means to solve the problem, the translation into code was fairly straightforward. Java’s object-orientedness forces the code into an organization I might not otherwise have used; beyond that, it makes it harder for me to read others’ code and make sense of how to use it. There have been times when the way I conceived of a problem made it easy to code up in Java, but not many of them.

This is not (necessarily) a shortcoming in a language. But it does mean I’d avoid Java in most cases. Maybe if I’d been taught Java in intro CS, the way the Shipwright was (in my day, it was taught in Pascal), I would think differently.

Now Playing: Everything’s Not Lost from Parachutes by Coldplay

Technorati Tags: java, programming, prolog

Posted by pjm at 2:02 PM | Comments (0)

I want to give you a good grade

An itch is a curious thing; it’s a little pain which drives you to inflict more pain on yourself. The things I’ve been itching to talk about lately are things I shouldn’t talk about here; little soap-opera dramas I’m only an observer in, or the travails of a TA at the end of a semester. I won’t even detail the grading dramas with pseudonyms; it’s easy enough to make the connection from who I am, to my class, to a pretty small set of students.

A TA in another department observed, recently, that she doesn’t think students know how much their TAs are pulling for them. It’s really easy for the students to see grading as something that’s adversarial, student vs. assignment, with the grader as a hostile judge. It’s also too easy for them to consider the purpose of grading as ranking, or somehow assessing a quality of the student.

It’s neither, of course; the TA knows that well-done assignments are easier to grade (“Happy families are all alike…”) and that well-written assignments lead to well-done assignments. The hours of office-hour help sessions and group reviews are, in that sense, an extension of the assignment sheet itself; they’re intended to clarify the assignment and help the students understand what’s expected of them, to better enable them to produce a good answer.

The entire teaching staff wants the students to do well, if only because it makes them look good; we want the sheer blinding force of our teaching skill to deliver knowledge like an electric spark to the students. (N.B. Having only lectured once, any discussion of my personal teaching skill takes place largely in the hypothetical plane.) Less dramatically, we want to pull them up the learning slope so they’re better able to engage in “interesting” discussions in higher-level classes.

But the students are conditioned to see class requirements as barriers and grades as battles. Their determination to emerge victorious is fine, but this conditioning can be counter-productive sometimes.

And that’s all I’m going to say about that.

Now Playing: There’s No Other Way from Leisure [US] by Blur

Technorati Tags: grades, gradschool, teaching

Posted by pjm at 12:38 PM | Comments (0)

This doesn't feel like an ending

I didn’t realize until it was nearly over that I probably sat through my last classes for quite a while today. I’d say “ever,” but that’s a dangerous statement. Assuming I am actually judged to have satisfied the requirements for the Masters degree, they’ll be handing me paper in less than a month. (I am actually turning up to graduation, though there was an option for me to just have the degree mailed to me.)

I am being automagically rolled into the Ph.D. program, but I have had my request for leave (i.e. time away that won’t be counted against my “time to complete the degree”) approved. There’s a chance that in the next year, I’ll decide that I really want to be back in a Ph.D. program in fall ‘08. There’s also a chance that I’ll be struck by lightning. I suppose I’m more likely to decide that I’m cut out for research than be struck by lightning, but not by much.

I’m still reluctant to rule out ever going back to school, but I think it’s more likely to be a part-time sort of thing than another two years of full-time grad school. My grandfather, as I’ve mentioned before, had three masters degrees, so I don’t feel any need to be constrained to a Ph.D. as my only future option.

N.B. There does seem to be some belief that just because I haven’t signed on with a corporation offering health insurance, that I don’t have a “real job” after graduation. This is not the case, of course. And, if you should happen to have a few thousand dollars that you can spare for an almost unquantifiably risky venture, drop me a line.

(What I learned in grad school: how to properly use e.g., i.e. and N.B..)

Now Playing: She’ll Come Back for You Tomorrow from Uninvited, Like the Clouds by The Church

Technorati Tags: computer science, gradschool

Posted by pjm at 10:38 PM | Comments (5)

Professor Coach

I’ve mentioned before how our building contains Computer Science, Electrical Engineering and Athletics. Professor β’s office has coaches for several doors on all sides, so last Sunday some of her grad students changed her job description. (I grabbed a phone-cam shot.)

Now Playing: Money Talks from Live From The Bowery Ballroom by Kathleen Edwards

Technorati Tags: aprilfool, athletics, computer science, identity

Posted by pjm at 9:20 PM | Comments (0)

Professor's revenge

• Get the first copy of the campus newspaper off the stack in the morning.
• Solve the sudoku.
• Post the answer as your first slide in each class.
• Your students without laptops now have nothing to do but pay attention.

(Posted, of course, as someone who was working on lecture slides in colloquium yesterday.)

Technorati Tags: inattention, laptop, distraction, teaching

Posted by pjm at 8:09 AM | Comments (0)

My very own lecture

Once I got used to it, I really like Keynote. I wish we’d been able to do the business plan presentation with it. It’s like all the cool stuff from Powerpoint, with all the cruft stripped away. I was able to run the lecture from my MacBook in two-screen mode, with the slides on the projector (one screen) and the “presenter view,” which shows the current slide, any notes, and the next slide, on my own screen, so I always knew what was coming next. Other than a few muffed transitions (I mis-programmed them) and some sections where I talked ahead of my own outline, things went pretty smoothly. I was easily able to click out of the presentation into demos online, some of which were actually running on my machine.

Except, of course, for the usual snoozers. 1:30 PM is a lousy time to have class; one of the women said last week that she found she needed to have her afternoon coffee early to get through this block, even when it’s not me lecturing. If I had time to re-do, I would hack more of my code samples into stuff they could easily download and try out on their own. I did show them how to switch on the web server on a Mac, and hinted at how it’s done for Linux. (Looking now at the default Ubuntu build I have in Parallels—didn’t I mention that my Mac now runs both XP and Linux?—I see that Apache isn’t installed on the standard Ubuntu, so maybe fewer students have a built-in Apache than I expected.) I also gave them the URL for my laptop (a DHCP URL only valid while I was jacked in to that ethernet cable) so they could run my demos on their own.

It turns out I was able to recycle some unused work as an example. I did this site over winter break (not the design, but the infrastructure) in Perl, then discovered that the host didn’t support Perl CGIs, so I redid it in PHP. (Pretty easy, actually; it’s a single HTML template, a CSS file, a couple images, and some plain text files. There’s not a whole lot of code involved.) I used the Perl version as a code example for the HTML::Template module, then the whole thing as a demo for the idea of using the filesystem as a simple database.

I also told them that one of the biggest sites built on PHP was one probably everyone with an open laptop had visited at least once during that class block. Several people guessed Google, but then someone guessed right: Facebook. Whereupon we got one denial… from a student who recently “friended” me on Facebook.

Standing in front of a class and talking requires a tremendous amount of mental energy; you have to be on all the time you’re up there. It’s like performing in that sense, I suppose. I feel burned out and unable to focus afterward.

Now Playing: Blue Pastures from Whiplash by James

Technorati Tags: computer science, facebook, gradschool, laptop, programming, teaching, websites

Posted by pjm at 10:01 PM | Comments (2)

Lecture

Professor Σ is away until Wednesday. This has been fairly common this semester, due to his long list of non-class responsibilities. In the class I TA, we had a visit from Career Services once, and a midterm review (which I ran) last time.

Tomorrow, I’m lecturing. Fortunately, not on our recent class topics (lambda calculus and denotational semantics) but on something a bit more practical: programming for the web.

I wrote up a brief outline, and now I’m hacking together slides in Keynote. I have to say, I’m hugely impressed with people who can lecture with slides twice a week. This is an incredible amount of work! I can only hope someone learns something.

Now Playing: Red Army * Blues (Song Of The Steppes / Red Army Blues) from A Pagan Place by The Waterboys

Technorati Tags: computer science, gradschool, teaching, websites

Posted by pjm at 9:34 PM | Comments (1)

How do you start learning about game design?

One of the undergrads I work with is an interesting case. He’s quiet, hard to draw out. He comes from one of the state’s desperately poor mill cities, and though he’d never say it, I think he’s still a little uncomfortable at the University, even after a few years. I bet if I described “impostor syndrome” to him, he’d be nodding before I was halfway done.

I think what he really wants to do is write games. I think that’s what drew him to CS, and I think that’s what keeps him at it—or, failing that, the unspoken promise of a well-paying job on graduation.

I’m not a gamer; I can play strategy games, but a few too many times I found that I’d blown a whole afternoon when there was something more important that I really should have been doing, so I just steer clear. As a result, I know next to nothing about the machinery of the games world. I know that graphics and rendering engines have a lot to do with it; I know there’s a lot of custom language development and language parsing that happens in games companies. That’s fine, I can steer him that direction.

But I also know there’s a whole branch of—sociology? anthropology? psychology?—focused on the study of games, what makes good ones, and why people play them. They call it ludology and it really is a serious academic specialty. I don’t think it’s worth steering this kid into that study, but I do think it would benefit him tremendously if we could find some kind of survey of the field so he’s aware that it’s out there; if he can develop an ability to apply their theories, that could help him land a job in games. Maybe.

So call this a sort of LazyWeb query. Does anyone know of a sort of survey of ludology?

Now Playing: Should I Stay Or Should I Go? from Combat Rock by The Clash

Technorati Tags: computer science, employment, games

Posted by pjm at 10:24 PM | Comments (3)

Snoozing students

The singular thing about graduate school may well be this: within a few hours, you can both doze off in class, and watch people doze off in a class you’re teaching.

With the professor out on Tuesday, I filled in to do a review for the midterm (today.) (If we had been a bit more forward-thinking, the midterm would’ve been Tuesday, but we weren’t.) Maybe it’s the chronically under-rested state of most undergraduates; maybe it’s the fact that class is from 1:30 to 3:00 in a chronically over-warmed classroom. I’ve dozed off in that room several times, myself. I didn’t count, but I can visualize at least four, maybe five of the students “resting their eyes” while I talked.

And I couldn’t really blame them, so I didn’t say anything.

Now Playing: Valentine from Pleased to Meet Me by The Replacements

Technorati Tags: gradschool, sleep, teaching

Posted by pjm at 9:37 AM | Comments (0)

The low-stress job hunt

It’s funny the kind of assumptions people make when they hear about my plans for next year. (My co-conspirator, who will need a pseud here soon, reports the same problem.) One is that we’re going to need jobs to pay the bills while we get things off the ground. It’s distinctly possible that we’ll need jobs in the not-too-distant future, possibly even by this time next year or sooner, but so far our optimistic plan is to go to work full-time for ourselves immediately following graduation.

Rather than explain this to everyone, it’s easier to just go to the recruiting presentations. It doesn’t hurt that they usually feed the attendees, and I’ve been doing pretty well this week. Julia’s company brought in a very nice lunch yesterday, for example, and this evening was pizza on Google. For a company which supposedly offers good food at all their offices, the pizza idea came off as a bit lazy… but if the turnout for yesterday’s presentation had been as large as it was for the Google crew, it could’ve turned quite expensive. And Google handed out t-shirts at the door, which they ran out of before I left.

Now Playing: Transcendental Sports Anthem by Devin Davis

Technorati Tags: business, computer science, employment, recruiting

Posted by pjm at 9:12 PM | Comments (0)

In which programmers in glass houses throw stones

Our home-grown grade-tracking software (written, according to department folklore, in some thousands of lines of Lisp,) was retired at the end of last semester in favor of a department installation of Moodle. (The perfectly rational reason for this was that the only person in the department who really understood the old system died last spring. Considering the time I and others spent wrestling with Sakai last year, there’s some irony in the choice of Moodle.)

We’ve decided that Moodle is too much solution for us in the class I’m TAing this spring, so we went looking for a simpler way to keep students up to date on their progress. (In any other field, handing back paper with written comments ought to be enough, but this is CS and most of the assignments never exist on paper in the first place.) It turns out that our widely-used perl utility for collecting assignment files, provide, has matching components for recording grades (profess) and displaying them to students (progress).

There’s also a utility written for checking to make sure configuration files, etc. are all set up properly. It’s referred to as a “sanity-checking” utility and is called, of course, prozac. From the manual page:

Like the real prozac, it makes provide happier in 95% of all situations, and otherwise becomes homicidal.

Now Playing: Nothing Like from God Fodder by Ned’s Atomic Dustbin

Technorati Tags: computer science, grades, gradschool, perl, teaching

Posted by pjm at 9:37 AM | Comments (0)

The Great Undergrad Empire

Writing assignments for this spring’s class is turning out to be the biggest brain drain. Grading them is tedious but not unmanageable; coming up with good questions is a colossal headache, particularly since one of my complaints about other undergrad courses in the department is the uninteresting and un-engaging quality of most of the exercises. (Really, who wants to write a business expense tracking system? Practically nobody.)

We’re throwing them into ML now (a few weeks ago it was Scheme) and I’m coming up with types and functions. I had the bright idea of basing this assignment around a text-adventure theme. What kinds of objects can be found and picked up in, say, Zork? What different qualities do they have?

The original Zork just has some kind of count limit on how many objects you can carry, be they matchbooks and keys or teapots full of water. Say we define a series of artifact types, e.g. containers, weapons, tools, treasure, and miscellaneous (there’s a use for a placemat, for example, but it’s not obvious.) What’s the distinction between them? They all have names, possible numeric weight and/or numeric value, but what qualities distinguish the groups—say, range of weapons, or uses of tools? Any ideas? How do you build the type structure of Zork? (And then, how do you describe the constraints to the students without simultaneously giving them the solution to the problem?)

Now Playing: Попробуй спеть вместе со мной from Группа Крови by Кино

Technorati Tags: computer science, programming, teaching

Posted by pjm at 9:53 PM | Comments (1)

How much do you really need to know?

The more I learn about computers, the more things I discover that I really don’t know. (This is related to the theorem that there’s always someone who knows more than you do.) But it seems like there’s really a pretty small core of tools a student needs in order to explore Computer Science; knowing them well (and being willing to apply oneself to learning) is probably 80% of doing relatively well in the field.

The tools are sometimes surprising. One of them, the Theory Tool, I didn’t really grasp until this past summer; it boils down to the idea of proof by induction.

Proof by induction and construction through recursion are the same process running in different directions; this is the means we use to take ones, zeros, and the concept of time, and build everything that can be done with machines and electrons. It’s a hairy topic; we’re taking the Programming Languages students through the “recursion” aspect of it now, and sometimes you can see their minds double-clutching.

I only recently bumped into an article suggesting that some people will simply never learn to program. It cites a paper arguing three things make up the primary hurdles in CS:

• assignment and sequence
• recursion/iteration
• concurrency

There’s induction/recursion in the second spot. The third, “concurrency,” is what my advisor describes as the “too much milk” problem: say you notice in the morning that you’re short on milk. On the way home from work, you stop at the store to pick some up. But wait: did your roommate just do the same thing? Buy the milk, and you may have twice as much as you can use before it goes bad. Don’t buy it, and you may have to go without. You have a concurrency issue. Modern humans invented cell phones as a solution for this problem; computer scientists have some tricks for it too, depending on the context (and it’s a major headache in some contexts.) Concurrency still gives me headaches, which is a bit of a problem considering that parallel processing fascinates me.

It’s the first hurdle which is sort of staggering. Assignment. It’s where you take a labeled container and put a value in it. The authors of the paper suggest that success in introductory computer science courses can be predicted by a simple test of a dozen questions or so. Here’s the first question:

Read the following statements and tick the box next to the correct answer.

int a = 10; 
int b = 20; 
a = b;

The new values of a and b are: 
[ ] a = 20 b = 0 
[ ] a = 20 b = 20 
[ ] a = 0 b = 10 
[ ] a = 10 b = 10 
[ ] a = 30 b = 20 
[ ] a = 30 b = 0 
[ ] a = 10 b = 30 
[ ] a = 0 b = 30 
[ ] a = 10 b = 20 
[ ] a = 20 b = 10

Easy, huh? Well, if you thought so, you may take to programming. If you didn’t, the second option (a = 20, b = 20) is the answer.

It would be cool if we could just teach those three things, then spend the rest of our time investigating the fun stuff, but there’s a lot of detail and ramification that needs covering as well. (I’ve heard it said that the goal of our entire Data Structures course is to make sure undergraduates understand the concept of a pointer.) We can let our machines build a lot with recursion, but we still need to pick the base cases and specify how to make them step, and doing that properly takes some care and practice which take time to learn.

But it is a little humbling, and perhaps inspiring, to think of all the work one can do just to fully understand those three ideas.

Now Playing: Orange from Come Down by The Dandy Warhols

Technorati Tags: computer science, programming

Posted by pjm at 9:50 PM | Comments (0)

Enough with the athletic metaphors

I am writing a homework assignment, and had to stop myself from labeling the questions “Warm up” and “Main set.”

Now Playing: Empty glass from I’m on my way by Rich Price

Technorati Tags: gradschool

Posted by pjm at 11:15 PM | Comments (0)

Making things deliberately difficult

This morning, after I spent an hour and a half or so sitting on stools in the stacks of the library with Professor Σ looking at textbooks (we seem to have finally found one) I took on the task of writing the first homework assignment.

We assume that the students know some C++, but probably not much else as far as programming goes. (That’s going to change in a hurry.) Professor Σ’s stated (if ambitious) goal for this course is to give the students a feeling for why one would choose one language over another in a given situation, and what makes particular languages good for particular tasks. So this assignment was to ask a language to do something it’s bad at—in this case, to ask C++ to parse a text file.

The assignment I came up with asks the students to write a C++ program which parses /etc/passwd, ignoring comment lines, and print the first, sixth, and seventh columns. (Columns in /etc/passwd are separated by “:” characters.) Then I tried doing it myself. I’m not an ace C++ hacker, but I was able to do it with Perl in eight lines… and in C++ I took 37 lines.

I suppose an optional way to make it tougher would be to require that only entries with valid shells be printed out. Perl, again, could handle that with maybe three or four more lines; I can’t even imagine how I’d do it with C++ (though, again, I’m sure it’s possible.)

It looks like I’ll be learning some Scheme. And I’m tinkering with Ruby anyway…

Now Playing: Too Fast For You from Hindsight by The Church

Technorati Tags: computer science, gradschool, programming, teaching, textbooks

Posted by pjm at 9:27 PM | Comments (1)

Scenes from the first day after break

Technically, it’s not “after break,” because classes don’t start for another two weeks (plus), and grad students aren’t required to be back until the week before classes start. So I was one of only four grad students in “the extension” this afternoon. (The extension has offices for sixteen of us.) The other three were various international students who do not, generally, go “home” over breaks. I saw two other grad students and only two faculty members all afternoon.

One of the four of us in the extension, who mentioned last month that I was “working much harder” this year, turned up while I was eating lunch. After sitting at her computer for ten or fifteen minutes, she turned back to me and exclaimed, “Classes don’t start until the 18th!”

I nodded in agreement.

“We don’t have to be here until next week!”

I nodded again.

“I thought classes started today!”

My best guess is that when you’re ABD and not actually taking any classes, this sort of scheduling detail isn’t always high on your radar screen. Actually, no, better guess: she has children, and for them classes probably did start today.

Now Playing: She’s On Fire from Drops of Jupiter by Train

Technorati Tags: gradschool, vacation

Posted by pjm at 9:46 PM | Comments (0)

Plans and decisions

I’ve been wishy-washy here for a year about what happens when I finish my MS at the end of the spring semester.

For the last month, it’s been increasingly clear that I’m not ready to move on to the Ph.D. For one thing, I haven’t lit on one area that sets me on fire, one thing I’m willing to devote three or four years of research to. Without that, I think going on is probably a bad idea for everyone. For another, the open doors have been closing; Professor β has decided I’m not such a great fit for her group (basically, my math skills are deficient, and I’d have to spend some time catching up,) and Professor Γ didn’t get the grant she wanted to fund me with. I could still try to work with Professor Σ, and I will be doing my MS work with him, but I’m starting way behind.

So it looks like I will take my paper in May and run. (Actually, I will be automatically rolled into the Ph.D. program, whereupon I will immediately go “on leave” for an indefinite period.) I’ve been talking with another student in the same situation, and he and I have been cooking ideas for a little website which we may try to turn into a going concern once we’re finished in May. I’ll post more as it becomes interesting. (Other than it being a website, it’s not an area I’ve worked in before, so let’s not get too excited yet. It’s the technology that interests me.) That will probably mean a lot of work, some of which is actually starting tomorrow.

But yes, of course this new project will have a weblog, too. Isn’t that the first thing after the business plan, nowadays?

Now Playing: Wings from Hello Starling by Josh Ritter

Posted by pjm at 10:19 AM | Comments (3)

Executive summary

My brain is toast. I have about five posts I want to write, but they’re all too long.

• Finals: It’s all over but the gradin’. I’ve been neck-deep since Sunday night; I’m short on sleep and haven’t been to the grocery store for so long that scurvy is starting to be a legitimate concern. Today I shaved and got a haircut so I’d look a bit less like a shipwreck survivor.

• Academics: I am, based on what my professors, an average student at best, and my math background is deficient. (This is not news.) However, I am in great demand as a TA; Professor γ was counting on having me another semester, but apparently while Professor β doesn’t want me in her research group, she does want me as a TA… and the department chair thinks I’ll be most useful with neither of them. (It looks like I will be both TAing and doing a Masters’ project in the spring with yet another professor, who I’ve mentioned before but I will now officially dub Professor Σ for brevity.)

• Apparently the University has had some small national notoriety in the past few days due to some so-called satire published in the campus conservative rag which some think crossed the racism line. I haven’t read the inflammatory text in question, and I think while there’s nothing wrong with holding the responsible authors and editors up to the ridicule of the University community—or, at the very least, explaining why their biases are wrong rather than simply chastising them for holding them—I also think that multiple public responses from the President’s office both overstates the importance of the publication in question, and lowers the President’s office. The editors in question are in a hole; let ‘em figure out for themselves when to stop digging.

• Racing: I will be at BU all weekend. Saturday morning I’m running a 3,000m on the track (I need to get out my old college logs and see if I even have a PR at that distance) and apparently that afternoon I’ll be in a relay or two over at the pool. (My team is looking for a good finish at the SCM meet.) Sunday I’m swimming 400m and 200m free, and more relays if I can still stand on the blocks without shaking at that point. Word is there’s wireless in the pool, too!

Now Playing: Boat from Let’s Cut the Crap and Hook Up Later On Tonight by Marah

Technorati Tags: computer science, grades, gradschool, medford, newspaper, race, relay, running, teaching

Posted by pjm at 4:51 PM | Comments (0)

Late policy

This semester was the first time I ever needed to think about stating a late policy, and I didn’t think of it until too late.

I’ve seen a few variations on the late work policy. Prof β this semester simply refused late work, because of the nature of the class. Last spring, her policy was 10% off for the first day an assignment was late, 50% for the second (and after that, why bother.) Another professor this fall gave us three “late days” to be used as we found necessary through the semester.

I’ve been liberal in my own grading. Labs tend to be graded on a ten point scale, but there are really only three places on that scale: 10 for excellent, 9 for good but not great, and 6 for incomplete (and, of course, 0 for nothing.) I accepted late labs for full credit until the last day of classes, yet there are still a lot of 0s in the grade book. The written assignments, graded on a 100-point scale, are trickier. Most students didn’t bother submitting them late, but when the first seriously late one came in, I made up a policy on the spot: five points off for each day late.

This turns out to be a bad decision. See, you can be a week late and still score (potentially) 65 points. (Nobody does, of course, because if you’re a week late you’re not turning in a perfect paper, either.) That’s a lot better than 0, so it doesn’t discourage the perpetually tardy terribly much.

What I need is a function over days late which starts small, increases by healthily large chunks per day and exceeds 100 somewhere around 5 days. 20 points per late day might be sufficient, but it’s too linear; I’d like some curve in there.

What would really be perfect would be if I could express the late policy as a function which requires greek-letter variables.

Now Playing: Starman by Dar Williams

Technorati Tags: grades, gradschool, teaching

Posted by pjm at 6:34 PM | Comments (2)

The effort shows

Since I’ve been a full-time TA this year, I’ve spent more time in my “office” at the department. (Last year, I did a lot of work at home, or in a particular basement lab.) Last week, one of my office-neighbors observed, “You’re working much harder this year.”

Now Playing: Be My Prayer from Seven by James

Technorati Tags: computer science, gradschool

Posted by pjm at 6:27 PM | Comments (0)

You have 38 days to finish and mail your application

As a result a series of emails that started mid-November and got somewhat confusing right around Thanksgiving, I returned to the College career center today to speak to any students potentially interested in CS graduate school, to pitch the University specifically and to answer other questions generally. Following my hour at the College, I had a tight connection over to Smif in the hopes that I’d get higher turnout from that institution if I didn’t ask them to take a 40-minute bus ride to the College.

The audience at the College was… small. One senior and one of the career center deans. The senior was a CS major who had already sent five or six applications but was still uncertain about what, exactly, he wanted to concentrate on. I managed to convince him to cut-and-paste from his existing applications into the University app (due January 15 for those planning to start in September ‘07!) so I suppose my yield on that meeting was 100%, which isn’t too bad, but if he’s accepted (and I suspect he will be,) he’s most likely to go to one of the other places that accepts him. The advisor had a lot of good questions, I was able to give her a good picture of why she might suggest the University to other students, and I left a folder full of University information there for anyone else.

And then I got caught in traffic on my way to Northampton, arriving at the designated auditorium ten or twelve minutes late to find it empty (and unsigned.) So I suppose I did get higher turnout by making that trip: many multiples of zero are still zero. I dropped off a few more folders at the department office, checked email at a public terminal in the campus center (which had another college’s webmail in its browser history) and headed home.

I’d planned to get in a run on one of my old Northampton routes, then maybe dinner in town, but after the disappointment of the non-meeting, I couldn’t get motivated to find a place to change and put in the energy, so I more or less went directly home. On the way back I thought, this is why I couldn’t work in sales. It’s not that I can’t sell; if I believe in what I’m selling, and I can be honestly positive about it, I do pretty well. It’s that I’m so keyed up for it that something like that empty auditorium makes me almost disoriented.

I find that have to remind myself of the small victory and not be overwhelmed by the subsequent failure. The most useful result to the University is probably the contacts I made. I can now write a short manual for contacting these institutions and arranging information sessions; the College (in the person of the dean) expressed interest in having a sort of panel of people from several departments at the University, though organizing things like that is way over my pay grade. I did put the University on some radar screens, and spread some seeds which may sprout much later. I suspect I added sentences to some future recommendation letters coming from my University by making this happen at all.

But all of this building karma for the future stuff is pretty tedious when you don’t know when—or even if—it’s going to pay off.

Now Playing: Exhuming McCarthy from Document by R.E.M.

Technorati Tags: amherst, computer science, gradschool

Posted by pjm at 9:13 PM | Comments (0)

Strategic use of leftovers

A few years ago—perhaps when I was a sysadmin and therefore a sort of de facto part of the support staff—I realized that it pays dividends to be on the good side of whatever support staff keeps the basics happening in the department I’m in. Applying to grad school really drove that home; it didn’t help me get accepted, but it did help me round up recommendations from absent-minded professors.

Today was the last meeting of the semester for the undergraduate group I mentor for, and I baked a batch of chocolate chip cookies from my mother’s recipe. (Actually, it’s Marge Standish, but it might as well be my mother’s. There are several things my mother does which I will never, as long as I live, be able to match, and two of them are chocolate chip cookies and apple pies.) Best batch of cookies I’ve made in years—maybe decades.

After our lunch meeting, I left about a dozen and a half extras in the department office. Subject line of the email I got from the Staff Assistant an hour or so later:

You get an A++!

Now Playing: Living It Up In The Garden from ‘Mousse by The Nields

Technorati Tags: cookies, gradschool

Posted by pjm at 5:26 PM | Comments (1)

Reinforcing cycles

It is an unavoidable coincidence of timing that the time in which a student is most busy—namely, the week or two leading to the end of the semester—is also the time in which a TA is consequently most busy.

Thus making me doubly busy.

There’s a payoff for this somewhere, right?

Now Playing: We’ll Inherit The Earth from Don’t Tell A Soul by The Replacements

Technorati Tags: gradschool

Posted by pjm at 9:56 PM | Comments (0)

Technology weirds language

I was a little surprised at how many students weren’t conscious of the fact that “software” is a plural noun with no singular.

It is possible to use “software” in such a way that it looks like a singular—the example one student gave was, “Software is hard”—but actually, it’s an object, not a subject in that case: “Selling software is hard.”

But companies can’t produce “a software.” They can produce “a software product” or “a software solution” or even “a software upgrade,” but not “a software,” just as you can’t walk into the nearest Home Depot and buy “a hardware.” (Nor, for that matter, can you buy “a linen” at Linens ‘n’ Things, as near as I can tell.)

I had previously thought this was a quirk of our students who aren’t native speakers of English, because they often have subject/verb number agreement difficulties, but it turns out some of our native speakers have this problem as well.

I also explained to them, last night, that pronouns are variables, and just like programs, their sentences will produce unexpected results if they aren’t careful about how they assign to those variables (or if the variables aren’t assigned before use.) I can never use that analogy again; I’ll never have another audience that will get it.

Now Playing: Released from Winter Pays For Summer by Glen Phillips

Technorati Tags: computer science, gradschool, writing

Posted by pjm at 10:35 AM | Comments (7)

I don't get out much

I was crossing the gym on my way to the pool, looking down at the basketball game against Springfield, when I made a discovery: the University has cheerleaders.

I don’t know why this is such a novel concept to me, but the thought that I’m a student at a university with cheerleaders has been amusing me all evening.

Posted by pjm at 10:18 PM | Comments (0)

Short strokes

A few weeks ago—I forget the context—I used the phrase “short strokes” and then had to explain what I meant. (It’s a golf metaphor, apparently, and not in wide use; now I can’t figure out where I picked it up.) But now that’s the best way to describe the semester. There are two weeks to go, we’ve drawn up our checklists of what needs to be taught/programmed/completed/graded in that time, and we’re just trying to get to the end of it.

When a former roommate and I were toying with learning to play golf, we used to go to a local driving range once or twice a week. We never, to my recollection, ever actually played golf; we just went to the driving range. I don’t think he even owned a full set of clubs, just three big drivers. Putting is what’s fundamentally frustrating about golf; everything else is whaling the skin off a little white ball, which is satisfying if you don’t slice like I do. So we’d get a medium bucket of balls (each) and try to smack them out of sight until our shoulders were sore.

The presentation went off today, I was barely prepared and took my lumps for it. (The draft I handed in a week before has not yet come back; I expect to take some lumps there, too, but I’m hoping to at least have a final paper that stands on its own.) I have a slew of coding and lab-sheet-writing and re-experiment-running to do over the next two weeks; I have my checklist written, the list of due dates lined up like wood that needs splitting, but no map of when to split it. There is no driving left; it is all putting.

I’m not the only grad student scratching at putting together my data and making it work. Scott is gathering data for his research, too, and while you can’t help me gather data (unless you ran Boston last spring, in which case you already have, thanks,) you can help him, particularly if you have a weblog. Read, understand, and give him a hand; we can’t push the ball into the hole, but at least we can give him a good lie.

Now Playing: Little Wing from Still in Hollywood by Concrete Blonde

Technorati Tags: gradschool, weblogs, research

Posted by pjm at 5:11 PM | Comments (0)

Presentation in progress

I ♥ gnuplot.

Now Playing: Song For The Asking from After Everything Now This by The Church

Posted by pjm at 9:14 PM | Comments (0)

254

That’s how many pages of double-spaced Software Engineering draft papers I have to read and mark up before being on vacation, counting title pages and references (from those who figured out BibTeX). So, nothing interesting to say right now. Lots of marking, though. These are engineers; they didn’t know they would be expected to write well.

Technorati Tags: computer science, gradschool, teaching

Posted by pjm at 2:04 PM | Comments (0)

Judgement calls

Is it fair to be biased against finding useful ideas in a paper when it contains a misspelling in the abstract?

s/influencial/influential/

I worry that their thinking may be as sloppy (and potentially hard to follow) as their spelling.

Now Playing: Car Underwater from What To Do When You Are Dead by Armor For Sleep

Technorati Tags: computer science, research

Posted by pjm at 5:49 PM | Comments (0)

A long run

I put together a script to run a decision-tree algorithm on all the various permutations of my data set (the 2006 Boston Marathon results.) Then I started the script on a timer, and went to do a workout.

When I came back, I found out the complete run had taken just under an hour. (56m40s, if you’re after precision.) An hour long test run! I almost feel like a real scientist.

Technorati Tags: computer science, gradschool, marathon

Posted by pjm at 9:03 PM | Comments (0)

I'm not usually early

It turns out that the degree sheet isn’t actually due until sometime in January. So instead of being ever-so-slightly late, which tends to be my usual state, I’m nearly two months early. The only explanation I have is that I thought I was late.

Now Playing: You Don’t Know How It Feels from Wildflowers by Tom Petty

Posted by pjm at 9:26 PM | Comments (0)

Anticipation

The nature of classes is that one tends to focus on the immediate future to the exclusion of the long-term. You worry about the problem set due tomorrow, and not the project due in three weeks, because the problem set is in your face.

So it came as a bit of a shock to me when I went in for course registration “advising” (irrelevant, because the only course I need to register for next spring is a Masters’ Thesis/Project) and my advisor said, “We need to fill out a degree sheet for you, don’t we.”

As a result, I spent part of Friday flitting around the department printing forms and trying to get signatures from people who weren’t present. I wound up with a sort of paper assertion that I intend to receive a paper and some letters from the University next May. Given that it has only been about fourteen months since I started here, it seems too soon to be planning departure, but the sheet confirms that I’m pretty close to done.

I checked the box saying I planned to continue for another degree, not because I positively have such plans, but because it seemed more prudent to leave that door open. However, the time for reaching a decision on that score is coming soon.

(Yes, I’ve barely been posting here. I’ve been busy… and scratching.)

Technorati Tags: gradschool

Posted by pjm at 2:33 PM | Comments (0)

Nobody likes a smartass

Plenty of people have pointed out that it may be difficult to predict DNFs purely from split data, given how many unmeasured variables affect the decision to drop out. I nodded to this in my project proposal, saying,

This [potential outcome] is what I think of as a “Tolstoy result”: successful runs are all alike, but unsuccessful races all fail in their own way.

Professor β asked for a citation on that, so it looks like I either drop the joke, or put Anna Karenina in my references.

Somehow the second course sounds like more fun to me.

Now Playing: Ghost of a Girl by Bluerunners

Posted by pjm at 9:27 PM | Comments (0)

Divergence

I read this on a job seeker’s website a few weeks ago:

This is the point of divergence. All of these potential futures become distinct universes in the coming weeks.

Distinct universes, as in choosing one excludes the others. Not something I’m looking forward to, myself.

Right now I’m (supposed to be) writing a proposal for a term project for machine learning. I’m going forward with the marathon split analysis, having been given the thumbs up by all involved, including two people who’ve done a little research in a similar area and are now interested in seeing my results. I turned down a few interesting projects to do so, including one which felt a little less… frivolous?

I also now have two “if I get this grant, I’ll be looking for grad student RAs, so be thinking about it,” offers. The one from Professor β, of course, is contingent on my doing well in this class (I’m interested, more so than I am by my other class, but the papers we’re reading are distressingly bewildering.) The good news—I think—is that the chair pointed out, in a discussion of the Ph.D. qualifying exams (not for me this time,) that everyone should come into quals with at least two faculty members who are willing to do research with them. So even if I’m otherwise completely unprepared, I could check that one off. But what if I like both projects? Choosing one excludes the others.

And, of course, there’s that big “so what are you doing next year?” question. I’m still chewing on that one. It increasingly comes down to the question of whether I like research, or at least whether I’m good at it. I’m beginning to get a feel for how it might be fun, but then, being a musician was fun, too. I just wasn’t good enough to do it for a living.

Now Playing: Seen Your Video from Let It Be by The Replacements

Posted by pjm at 9:21 AM | Comments (0)

The best demonstrations are accidental

The scene: a group of undergraduates, maybe around thirty of them, selected from “under-represented” populations in given fields. (So, majority female, more racial minorities than usual at this university.) We’re in a large first-floor lounge near the lobby of a dormitory.

The discussion is about social class, that subtle discrimination that we pretend doesn’t exist in this country. We’re discussing ways we’ve noticed class differences on campus, and one of the students points out in the lobby at a big stack of brightly colored cloth bags. “There’s one right there.”

“I was wondering about those,” says the facilitator. “What are they?”

Almost in unison, the students chorus, “Laundry bags.”

“Do you mean some students have their laundry done for them? Don’t all the dorms have laundry machines?”

In unison again, “Yes.”

The facilitator makes a face. I think they would have laughed, then, if it hadn’t been so sad.

Posted by pjm at 8:42 PM | Comments (1)

The real problem

The edition of Brooks that we’re using includes “No Silver Bullet,” a 1984(?) essay explaining why the problems facing software engineers were unlikely to go away quickly or soon. (The older the essay gets, the more right he looks.) Brooks’ argument centers on the idea that the dramatic improvements early in the computer era were achievable because they countered “accidental” difficulties, not “essential” ones. Essential difficulties are problems which are inherent to the process of communicating ideas in code; accidental difficulties are inefficiencies embedded in the available tools and the costs of hardware in the early days. Brooks’ contention is that the accidental difficulties were “low-hanging fruit” and were comparatively easy to solve; the remaining essential problems will be harder to solve, and the solutions will offer smaller incremental benefits.

The concept of accidental vs. essential problems is not simple for anyone, and I’m not surprised many of these literal-minded engineers trip over it. But I wonder how many of our people for whom English is not their first language easily understand the concept of a silver bullet?

Also, someone has tagged the book with my first name in Amazon. I don’t know who did it, but it has to stop.

Now Playing: T.B.D. from Throwing Copper by Live

Posted by pjm at 9:27 PM | Comments (0)

Media I/O

If Mark Will-Weber is to be believed, some time in the 70s Duncan MacDonald, a contemporary of Don Kardong at Stanford, was congratulated after a race by a fan who then asked, “I see you on television and I read about you in the papers. How do you do it?”

MacDonald answered, “I don’t watch television and I don’t read the papers.” Whether he meant it in modesty or irritation isn’t clear.

I feel a bit like that about graduate school, except a stellar grad school performance isn’t likely to get me on any television broadcasts. My weakness tends to be the newspapers, although I need to broaden that category to include “any non-academic reading in any format whatever.”

Posted by pjm at 9:35 AM | Comments (0)

The Facebook question

Do I or don’t I?

I’ve been eligible to create a Facebook account even before the recent opening-to-everyone. (I suppose I could’ve used my wsc.ma.edu address even before I came here.) I haven’t done it, mainly because I didn’t have much interest. I’ve avoided 90% of the so-called “social networking” sites (the other 10% being in response to a specific request for help; I’m now having difficulty remembering which site it was.)

I still have only limited interest, but the more I work with undergrads, the more it becomes useful to have some kind of visibility in their social space. I discussed it briefly last year with an undergraduate TA, and dismissed the idea with, “I’m too old for Facebook.” It does seem a bit sketchy—as though I’d be one of those adults who’s trying too hard to be “cool.” And I’ve heard about (and seen) too many cases of poor public-image management on the part of students to really want to be involved. (Begging the question: Is this site a case of poor public-image management on the part of this student?)

But I am seeing a point in putting up a face, so to speak. At least one other person I know through track is on, focusing on his work in his University’s athletic department. He’s making himself visible and accessible to the students he works without trying to compete with them in presentation—using the medium to present the image he wants to present.

So I wondered: as a TA, and as a “mentor”, should I have a presence on Facebook?

(This isn’t actually as important to me as this post makes it sound; I just thought I’d share the thought experiment.)

Technorati Tags: facebook, gradschool, websites

Posted by pjm at 6:36 PM | Comments (5)

Correction

I know you all, like my lab students, are comfortable with my infallibility, so you’ll be shocked to hear that I’ve mistakenly misled you all. Last spring, I contemplated the number of possible configurations of an Othello board and decided that the game was just too complex for the current state of computing. Yesterday, our colloquium speaker pointed out researchers from his group at the University of Alberta had written Othello-playing systems which were beating human champions nearly ten years ago.

As usual, there’s always more for me to learn. And I almost contemplated going back to Edmonton.

Technorati Tags: computer science, gradschool, programming, puzzles

Posted by pjm at 8:46 PM | Comments (0)

Graffle

Last night, I presented a paper in poster format to my Machine Learning class. Actually, about eight of us did; the class milled around the room, nodding distractedly as the poster presenters tried to distill eight- and ten-page blobs of theory into five minutes of talking with illustrations.

We didn’t have the facilities to print proper posters (one sheet, about three feet by four feet: think half a sheet of plywood.) Most people went by the alternate route of creating a bunch of slides (nine to twelve) in Powerpoint, printing the slides, and taping them up. I wanted a bit more consistency across my pages, plus I have a stated aversion to Powerpoint, so I made my “poster” in OmniGraffle.

OmniGraffle is a wonderful tool made mainly for building graphs; it comes with a bunch of “stencil” shapes for standard graphs like UML diagrams or network maps. In that way, it’s a lot more like Visio than Powerpoint, but in this case that’s its strength. It also allows the “canvas” for a single chart to spill over several pages, and that’s exactly what I did for this poster. It happens that I made the poster fit a series of discrete pages for ease of printing, but when I was working on it, I had all the pages sprawled out on my screen so I could see how they fit together.

Also, whenever I came in with a draft, the T.A. looked over the pages and said, “Can you use less text here? And here? Can you illustrate this?” I went back around two or three times adding color, trimming text, and generally cartoon-ifying the whole thing, until I felt like I was cutting important concepts.

I think I wound up with a very good-looking poster, and I thought it stood out among all the Powerpoint slides (though maybe I’ve spent more time looking at layouts than most CS grad students.) OmniGraffle wants to deal with layout elements, not text. It doesn’t lend itself to bulleted lists or large paragraphs. It’s possible that it took me too far in that direction, but I think it helped me.

(It has also been extraordinarily helpful in drawing DFAs and NFAs, first for my theory course and then for Compilers this fall.)

Now Playing: Rosebud from Cold Roses [Disc 2] by Ryan Adams & The Cardinals

Technorati Tags: computer science, gradschool, software

Posted by pjm at 9:51 AM | Comments (1)

University Holiday

Apparently the Trustees have declared today to be “Return the Algorithms Books You Borrowed From pjm Day.”

Unfortunately, that holiday doesn’t come with the “No classes” designation.

Now Playing: Honey Bee from Wildflowers by Tom Petty

Posted by pjm at 3:50 PM | Comments (0)

Refresher course

I have a theory—actually it’s more like a fear—that if I stay in this field, I’ll need to take Calculus biennially simply to understand the readings.

Technorati Tags: computer science, math

Posted by pjm at 4:00 PM | Comments (0)

In the homestretch

When I realized that half of my homework for one class was going to be almost trivially simple—it’s a review of some concepts from this summer’s class—it really began to sink in that this year is better than last year.

Not easier. I’m still pretty strapped for time, and assignments aren’t getting enough attention early in their life-spans to avoid stressful last-minute work. Some of my assignments still give me headaches.

But, somehow, better. Perhaps it’s that I have more than 2/3 of my classes done, and there’s a feeling of having the end of the M.S. in view. Being reminded that classes are additive, that what I’ve learned in the past year is helping me now, is a good feeling.

On Wednesday, I sat in on an information session for undergraduates considering graduate school in CS. At some point, the presenter said something like, “Most graduate students, if they can get through the first year, go on to finish.” And I thought, “Well, that’s good news.”

I may even have said it out loud.

Now Playing: All The Young Dudes from A Box Of Birds by The Church

Technorati Tags: gradschool

Posted by pjm at 8:14 PM | Comments (1)

That mindset thing again

I’ve mentioned that the Software Engineering course I’m TA for starts off by reading Brooks. The book is, for the most part, thirty years old, and shows its age in many ways. Mostly, the problem is technical details; the core principles still hold.

We asked students to pull out five terms they didn’t understand, and provide definitions. A lot of them were expected, but a few have taken me by surprise.

One of them: microfiche. If you’re my age, you remember hours in the library (libraries) doing literature research on microfiche, but for today’s undergrads, that’s apparently such ancient history they don’t even know the term.

Technorati Tags: computer science, gradschool, library, microfiche

Posted by pjm at 7:01 PM | Comments (3)

Learning from the data

I’m contemplating a course project based on “knowledge discovery” in marathon chip data. Not published results with 5k splits, but the raw, direct-from-the-mats chip data, the stuff the timing company uses to re-run the previous year’s race as a system test.

Knowledge discovery (usually called “machine learning” at the University, but also sometimes known as “data mining”) is an interesting field, because it’s implies the idea that there are patterns in data which are too subtle for us to see. One of the major tasks is classification, often used in medical applications to distinguish a set of symptoms as ill vs. not ill.

That’s not a simple task for marathon data; what are the classifications? Did the athlete beat their seed time? Did they finish? It might be intriguing simply to see if a program could predict, based only on chip data, the gender of the athlete wearing that chip.

The profusion of data with a high level of variance is a big problem for this hypothetical analysis, but another one is the mentality. We know there’s a huge number of variables in play, and at some point we discard the possibility that we could ever make sense of it all. But one of the strengths of machine learning is that the software decides which variables are actually relevant, and which are just noise.

It’s also approaching the problem of identifying which data are representative and which are outliers; our gut instinct is to suggest that we’re all outliers, but that’s clearly not the case, or there wouldn’t be thousands of runners crossing the line every hour.

So if you stop worrying about whether the answer can actually be found—that’s a question to be answered later—and just think about questions you might ask, what would you look for in marathon data?

Technorati Tags: computer science, data, marathon, running

Posted by pjm at 9:45 AM | Comments (5)

When "chair" isn't exactly the right word

I had a brief one-on-one meeting with the department chair yesterday, and while I think my initial apprehension had some basis in fact, I think I’ll get along well with her. If this woman had gone into the military, she’d be on her way to General Staff by now. She has clear ideas about the way she thinks things ought to go, and is talented at convincing everyone else to make it so without bulldozing them—more likely they’ll leave thinking it was their own idea in the first place.

She reminds me, in many ways, of the theater group director in my high school, though quieter. (At least two of you now know exactly the kind of person I’m talking about.) She’ll give you all the work, responsibility, and/or corresponding glory you can handle, but woe betide those who expect the rewards without the responsibility. I’ll do fine as long as I stay on her generally-good side, but I suspect everything I do for her will come with some low-grade, possibly-unjustified fear of not making the grade and being consequently cast into the abyss.

But I have to have respect for any manager willing to use the phrase, “read the riot act.”

Now Playing: Pizza Cutter from Wholesale Meats And Fish by Letters To Cleo

Posted by pjm at 11:08 AM | Comments (0)

You studied what?

My classmates never cease to amaze me. Today I discovered that one of the Ph.D. students I run with regularly has an undergraduate degree in Fine Art—drawing, specifically. And that before he got in to graduate school in CS, he was turned down (he says, thankfully,) by an MFA creative writing program, and an Ed.D. program. “So, yeah, I’ve got that Russian Lit degree of yours covered,” he said.

I also discovered that the department chair was a double major in English and wrote her undergraduate honors paper in literature. The next time someone drags out the myth about CS types having a narrow focus, I want to invite them to our department, to visit our little colony of escaped engineers, classicists, and lit majors, et alia.

Now Playing: Appalatia from Forget Yourself by The Church

Posted by pjm at 8:55 PM | Comments (0)

I need to work on my estimating skills

I was worried about my lab students finishing the first lab too quickly.

It’s now over an hour after the lab period ended, and there’s still two students working. Only one finished in anywhere close to the scheduled time. Apparently I was worried in the wrong direction.

Technorati Tags: computer science, gradschool, teaching

Posted by pjm at 12:53 PM | Comments (0)

Encouraging results

• Several years after the parking lot in the Bird Sanctuary, Zipcar is coming to the College.

• The female undergrad (sophomore) who wanted to know how many courses until she “caught up with all the boys who already know everything” last year is in Comp 20, the “multimedia programming” course that’s a forerunner of the “advanced web programming” course I took last spring. She asked me which course I was TA for, and acted disappointed when I named the Software Engineering course, which is only for grad students, seniors, and ambitious juniors.

• So far, I have been able to debug all the problems Professor γ is having with PHP and web forms. I am tempted to unilaterally declare myself course webmaster and fix all her code; I apparently have significantly more experience with PHP. (This may be true for most professors.)

Unrelated, but interesting: does anyone know of a PHP function for translating Perl POD documentation to HTML?

Now Playing: Riding on the Subway from The Fine Art Of Self Destruction by Jesse Malin

Posted by pjm at 9:59 PM | Comments (0)

Reoriented

Friday afternoon, I participated in a panel for new graduate assistants in our CS department. There were far more “experienced” GAs than new ones in the room; while the cohort joining the department with me last year was apparently relatively large, comparatively few (five or six, I think) are coming in this year.

One thing I learned was that the University has gone ballistic about plagiarism this year; supposedly that was the major topic of the University-wide new-TA orientation in the morning. Last year it was hardly mentioned. I was amused to see that the University has now contracted to use turnitin.com to help sniff out plagiarism; I’ve been seeing their bot in my server logs for years now. I can’t imagine what kind of trouble someone could find themselves in by copying indiscriminately from this site, even if the sourcing wasn’t detected.

I’ve also committed to help out on the mentoring project. I raised a few questions with the department chair, who had good answers for them. I expressed reservations about my qualifications, and she pointed out first that they wanted students with a range of experiences, so with the other grad student being post-quals and into dissertation work, they needed someone early in the process as well. (So I was selected for my inexperience—thanks, I think.) The fact that I’m not doing research with any of the project faculty (or any faculty, for that matter) means the students can ask me questions without betraying confusion or ignorance to people they’re trying to work with.

Finally, she pointed out, “We’re trying to help them through a successful graduate school application process. You’ve done that, right?”

I couldn’t really deny it.

Now Playing: Snow is Gone from Live at The Paradise Rock Club (Boston, 12-11-04) by Josh Ritter

Technorati Tags: computer science, gradschool

Posted by pjm at 10:25 AM | Comments (0)

Going way back

This weekend, at the wedding of a younger cousin, I talked with her other grandfather (i.e. not the one we shared.) Turns out that not only did he get his undergraduate and medical degrees at my current University—a feat now known as a “Double Jumbo” for reasons stemming from the University mascot—but he ran cross-country.

Class of 1936, or thereabouts. I wish the University posted all their old team photos the way the College did so I could see if I could find him.

The bride’s other grandparents were represented by a photo from their own wedding day; they were married in the same church about seventy years ago. The photo was taken at the reception, on the front lawn of the house I knew as theirs. In black and white, our grandfather looked even younger than than the ten or so years younger than this photo, with even more hair; our grandmother looked startlingly like the mother of this weekend’s bride.

My older niece was easily distracted, playing with her necklace; the younger, rapt. I whispered, “Promise you’ll invite me when it’s your turn?” She nodded solemnly.

Now Playing: Leave Them All Behind from Going Blank Again by Ride

Posted by pjm at 5:04 PM | Comments (0)

Negative data is still data

This afternoon, I passed through the campus bookstore to see if I could figure out what the text was for my Compilers class. (The sooner I know the text, the sooner I can try to order it online. Online bookstore fulfillment for textbooks can be slow this time of year.)

Both that class and the one I’m TAing—in fact, nearly the whole CS department—are listed as “No book order placed.” I know that’s not right about Software Engineering; I have “desk” copies of the two books.

I was passing through the CS building afterward, so I looked in on Professor γ and let her know about the missing books. Hmm, a puzzle, she was sure she’d made that order—but now she’s checking up on it. Better to find out today than on the first day of class.

Posted by pjm at 7:52 PM | Comments (0)

I'm already "old school"

Well, that didn’t take long.

This afternoon someone noticed my hat—a year and a half old and already faded to an indistinct blue-brown—and joked, “They’re gonna take that away from you; it doesn’t use the Official University Font.”

Seems you can’t get hats in this design anymore, or for that matter those traditional block-letter sweatshirts; they’re pushing visual brand identity now. So I’m holding on to this hat; it’s going to be a collector’s item, right? Scarcity drives up demand?

Now Playing: Guitar Song from Strangest Places by Abra Moore

Technorati Tags: branding, gradschool

Posted by pjm at 1:56 PM | Comments (0)

Advice welcome: mentoring

I continue to be surprised by the gap between my own assessment of my academic abilities, and the faculty’s apparent opinion of me. An example came in this morning’s email:

The position of “CUSP [Computing Undergraduate Scholars Program - ed.] grad mentor” is an “add-on” position to a fully-funded RA/TA/GA position. Although in general an RA/TA/GA cannot take on extra work, there is a small codicil that says that in the case where it is to the mutual benefit of [the] University and to the graduate students, one may take on up to an additional 5 hours/week at an hourly rate. [snip] Other hours would be for planning what happens at the lunch meetings or possibly meeting with CUSP students individually not in the lunch meeting. Overall, it is similar to CSEMS [Computer Science, Engineering, and Math Scholars - ed.], but with fewer students, with them involved in computing research projects, and with the goal to mentor these students to consider graduate school seriously, as well as summer research. …

… I hope you will consider being a part of the mentoring team. I have heard good things about what kind of energy and creativity you might bring to the program that would of course impact the students’ experience.

Now, “energy” and “creativity” are not words I would have used to describe my year in graduate school. “Tenacity” and “thickheadedness,” perhaps. (“Energy” comes between “empty-headed” and “failure” in my dictionary, not that that means anything.)

Still, my own cognitive dissonance isn’t the issue here. (Maybe later.) I need to figure out what to do about this mentoring program. Pros and cons:

• It represents a pay raise (upper bound of 20%, if I’m figuring correctly.) I’m not as hard up as most grad students, but I still want to have good reasons before I turn down a raise.
• This is an opportunity to do something good for the field. If I’m vaguely serious about sticking with it and becoming faculty someday, this is exactly the sort of thing I should be doing.
• It will give me greater exposure to real research in the field and presumably help me clarify whether I want to do that as a career.
• It’s another positive line in the recommendation letters people will be writing for me when I’m eventually trying to get a job at any level. In general, ve