PBG 2005

Well, after two more days in Long Island than I would have preferred (read: two days), the 2nd Annual Eurographics Symposium on Point-Based Graphics is over. Didn’t get to see much of SUNY Stony Brook, but the Wang Center where the conference was held was pretty nice.

There were some interesting papers, but with only 30 submissions for 15 slots and a pretty dismal overall turnout (I counted 41 people in the room during my talk), I have to wonder if point-based graphics really needs its own conference. The good theoretical results would be perfectly suited for SODA, the good animation and simulation papers should almost certainly appear in SCA (separating simulation papers into distinct conferences by the underlying framework seems exceptionally silly), and new rendering results that matter (if there are any…after two days, I remain unconvinced) would be better off in EGSR or SIGGRAPH. Instead, it seems as if several of the more prolific authors in the field are trying to (ahem) force elliptical pegs down triangular holes, and publishing results that aren’t actually all that interesting or important but find a home because they use points instead of grids or meshes. As a result of this, the conference had a slightly shady feel to it, and it was quite galling to see several accepted papers being presented by people manifestly unfamiliar with the material since none of the (three plus!) authors could be bothered to attend.

The proceedings were a total disaster: many of the mathematical symbols in equations and embedded figures didn’t show up in what the printers produced, rendering several of the more theoretical papers completely useless (well, even more completely useless than most theory-oriented graphics papers). My paper came out fine, but I suppose that will prevent me from subsequently claiming that its salience was mitigated by lackluster reproduction. Oh well, it’s not like scientists read dead tree anymore anyways.

The highlight of the conference was a very good keynote talk by Nina Amenta, which contained the first compelling explanation I’ve seen of why the projection procedures originally described by Levin and Pauly for Moving Least-Squares surfaces don’t actually produce points on the MLS surface. [I think I'll eventually type up some notes on this subject, since the appendix to the Defining Point Set Surfaces paper, while certainly technically correct, isn't particularly instructive.] The main focus of her talk boiled down to the assertion that it’s senseless to keep producing algorithms that process points and then asking ourselves what sampling criteria are necessary for these algorithms to produce good results. Instead, we should first define the sort of sampling we’d like to have (which she called a “super-sampling”), and subsequently develop algorithms that take advantage of all its super properties. Of course, when every sampling is super…none of them will be.

This was definitely food for thought, but I’m inclined to believe that both approaches are wrong: it seems more sensible to me to ask ourselves what sort of samplings we’re likely to encounter in real world applications and then design our algorithms appropriately. Sampling criteria which are predicated on measurements like Hausdorff distance with the medial axis may make for nice theorem-proving frameworks, but it’s sort of silly to describe how faithfully we can reconstruct a surface by requiring measurements we can only make if we know the surface a priori. I understand the role these formulations play in the larger scheme of things, but I remain unconvinced that it’s the “right” way to be thinking about these sorts of problems. Of course, the “right” way to think about these sorts of problems involves a hot tub, a bottle of Jose Cuervo Reserva de la Familia, and more Playboy models than I’ve had the privilege to know thus far.

Prompted by some things I saw (and didn’t see) at PBG, I’d like to share some notes on Answering Questions at Conferences (aka “How Not To Be An Asshole”):

1. The correct initial response to a question asked after you’ve given a talk is “That’s an excellent question!” Any of “Uh, I think I just answered that,” or “Your question doesn’t make any sense,” or “That question’s so vacuous my Grandmother could answer it…and she’s been dead for two years” make you look like an asshole, regardless of whether or not you happen to be correct (or an asshole).
2. If someone asks you a question you can’t answer, immediately state that this is the case and then verbalize your thought process for the next 30 seconds. If, at the end of that time, you still haven’t produced a satisfactory answer, remark again on what an excellent question you were asked and offer to discuss the matter further with the questioner offline. Answering a similar (but much easier) question to which you do know the answer instead just makes you look like an asshole. Making up an answer not only makes you look like an asshole, but may also cause new graduate students who previously idolized you because of your publication record to realize that you are mortal and occasionally full of shit.
3. If someone asks you a question that really is trivial[1], answer it quickly and clearly and then use the opportunity to segue smoothly into a restatement of some aspect of your talk that is novel and interesting. This has the net effect of satisfying the initial inquiry while simultaneously making the person who made it appear insightful and knowledgeable. It’s quite likely that this person will end up reviewing one of your papers one day, and if you blow them off they will remember.
4. If someone asks you a question that you truly don’t understand, restate what you think they’re trying to ask and then say “Is that more or less your question?” Lather, rinse, and repeat until convergence. Staring blankly at the questioner, making them repeat themselves verbatim, and then saying “I don’t know what you mean” makes you look like an asshole.

[1] Beware of non-trivial questions which seem trivial. If you emphatically and derisively take a position, then immediately reverse yourself, and then reverse yourself again, you look like an asshole.

Things To Do…

1. Get into top-ranked Ph.D. program.
2. Move to California.
3. Purchase Sex on Wheels.
4. Pass quals.
5. Write dissertation.
6. Land professorship.
7. Get tenure.
8. Torture students Make the world a better place.

Crap! I think I forgot “Make obnoxiously self-indulgent blog post” between 3 and 4.

Edited (6/23/05): Non-standards compliant browsers suck.

To the Nerdery!

Just so I don’t forget about this: if you’re ever forced to write a paper using ACM’s incredibly ugly LaTeX style file, you must download Jeff Erickson’s fixacm. For bonus points, you can imminentize the eschaton…but that might be more of a commitment than some of us are willing to make.

And you may ask yourself…

Like most male middle-class American children born in the early 1980s, I became enamored with video games shortly after leaving the womb. In elementary school, I can vividly remember accepting play-date invitations offered by children I would have otherwise avoided like syphilis in order to secure a few precious minutes of frenetic button-mashing. Christmas of 1988 saw my father finally succumbing to my incessant whines of “But Dad, everybody else has one!” (are we detecting a trend here?) and buying me my very own Nintendo Entertainment System, along with two games: Defender II, and Ninja Gaiden. The games looked so cool: I was instantly hooked, and the rest, as they say, is history.

Except, of course, it isn’t. Even if history is “just one damned thing after another”, being only twenty-three and of modest talent (if not motivation), I have yet to actually do that many damned things. Life thus far has consisted almost exclusively of progressing dutifully through the lower echelons of the American educational system, and for the most part this journey has been only moderately satisfactory. Elementary school is idyllic and enjoyable largely due to its utter lack of purpose, but also, like most aspects of youth, totally wasted on the young. Middle school is both awkward and exciting simultaneously, but belies the utter horror that awaits on the horizon so cunningly that it leaves its matriculants feeling somehow double-crossed. High school, of course, is a torturous exercise to which we are sentenced in partial repayment of our original sin and from which we derive no actual benefit. College, for those of us with greater ambitions than the reasonable expectation of food on the table and a warm place to sleep each night, is simply an affirmation of what we’d hesitantly suspected all along: namely, we don’t know a goddamned thing about anything.

Graduate school, then, has the potential to be either a continuation of this rather pointless certification scheme or an opportunity to make a unique contribution to the wealth of knowledge of mankind. The latter is a much more attractive proposition: finally, after all these years of intellectual masturbation, a chance to do something! The trick, it appears, is figuring out how to go about it.

In particular, it seems as if there are many different ways to be a bad graduate student and very few ways to be a good one. Most strikingly, being a good or bad graduate student doesn’t seem to be strongly correlated to one’s intelligence or diligence. Certainly you must be reasonably smart and willing to work hard, but the road to industry (read: hell) is paved with bright and earnest graduate students. Too many factors can influence a student’s success for the process to even be deterministic: funding opportunities, grant limitations, personality conflicts, tenure decisions, or even an unlucky research approach can turn a promising young Ph.D. candidate into just another M.S.C.S.

One of the most daunting problems a graduate student faces is picking a dissertation topic that people will care about. It’s (fairly) easy to write papers and publish them, but a good dissertation should be more than a hodgepodge of all the ideas one managed to convince journals and conferences to accept. In the same vein, students often mistakenly believe that if their research is very difficult and complex, it must be meaningful. In fact, the converse is almost true. Richard Feynman once said that if you can’t tell a small child what you’re doing in less than fifteen seconds, you’re probably not doing anything interesting. The most successful and sought-after Ph.D. students wrote dissertations to answer a fundamental question or solve a very general problem.

In some ways, picking a dissertation topic is a more challenging task than doing the research contained therein. In part, this is because most intellectually active people have a wide range of interests, and picking a small subset of these to focus on exclusively for four or five years seems limiting and mentally claustrophobic. Of course, this is also largely due to the fact that most students good enough to be accepted to competitive graduate programs (as virtually all of them are) are also smart enough to tackle most problems set in front of them. Eighty percent of success may well be showing up, but that first step out the door can be a real doozy.

Switching to a new school, a new advisor, and a new sub-field after my Master’s degree has granted me what amounts to a one-year reprieve before I’m faced with the dissertation decision. Nonetheless, it’s never to early to start thinking about the future.

The Saga Begins…

So, bowing to the ever-increasing pressure from the world at large and in a (probably futile) attempt to emulate some people who do it well, I’ve decided to start blogging.

A little.

Generically speaking, I’m not a big fan of blogs. As with most forms of electronic communication, people are apt to blog things they would never even think of writing down on paper and nailing to the door of Castle Church, and this can be a bad thing. More to the point, though, blogging is a domain to which Sturgeon’s Law most definitely applies, and it can be hard to separate the wheat from the chaff.

The very first time I catch myself bitching about my personal life on here, I’m going to delete the entire blog and defenestrate my computer.

All that being said, I see two primary benefits to establishing a blog. My job for the next twenty or so years is going to involve rather a lot of writing, and one striking commonality of all the good writers I know is that they write a lot. So, first and foremost, I’m hoping that having a blog around that I feel mildly guilty neglecting will force me to write more, and that doing so will, in turn, make me a better writer. This may be a pipe dream, but I don’t see how it can hurt.

The second benefit is related to the fact that recently I’ve begun receiving a lot of email asking various research-related questions. This probably has something to do with the relative prominence of my webpage. In any event, a large percentage of this kind of email merits a response of some kind, but as I am unwilling to spend more than a few minutes a day answering mail, most of it just gets deleted. The plan, therefore, is to talk about some of the more common topics here in an attempt to cut down on the email I receive and progress towards finding happiness. This is almost certainly a pipe dream, but, again, it’s unlikely it will make things any worse.

So, in the days to come, I’ll be occasionally updating this space. Hopefully it won’t suck too much.