Sunday, February 10, 2008

Can't We All Just Make Sense?

School is in and I’m on the thesis trail with a purpose. The story so far is this: A guy previously in my lab worked out a theory and a computer program for solving a previously unsolvable type of optimization problem. It works but it has a lot of short comings. That’s where I come in. I’m supposed to improve the method so it can be used on bigger, more complicated problems. Well, while I’m away in Germany some dude from an unnamed university releases what appears to be a landmark paper in exactly the type of problems I’m supposed to work on. It solves almost every issue with my labs previous method, and this paper makes a major point of demonstrating this new methods superiority over ours. So I spend a couple of weeks trying to understand what they did, and their theory is gorgeous. Things look pretty dire for our method. But then last week, about three weeks into this whole episode, my advisor and I finally find what we think is the Achilles heel of this new method: it doesn’t scale well. Vaguely what this means is that as the size of the problem you’re trying to solve gets bigger (more variables), every computer program will take longer to solve the problem. In this case, their method gets worse a lot faster than ours. We think. It will take me 2 months of grueling coding to find out whether that’s actually true.

Points here are these: First, the dude who came up with this new method must know that his method scales poorly, but he doesn’t mention it in his publication. He solves a series of relatively small problems, shows how much better his method is, and completely ignores the fact that he’d get smoked on any realistically sized problem. Second, he purposely obfuscates half the shit in his method. Like he’ll mention some crucial step in passing but not actually present the details, so that when I try to implement it, I have to spend weeks reinventing his method from what little clues he gives and snippets of a hundred other papers on marginally related topics. Basically, I think this guy is being a bit of bitch here. Healthy competition is probably responsible for a huge amount of development in science, but making grand claims and then withholding information is a punk move. I’m going to figure out what he did, it’s just a matter of whether or not I think he’s a complete asshole by the time I do. Meanwhile, the rest of the optimization community who hasn’t pined over this paper for three weeks basically thinks this whole subset of problems is solved. What’s the appropriate anamonopia for a scoff?

Yesterday I was at this party and I basically got called uncouth for wearing a white T and rocking a bottle of beer in my front jeans pocket. Fuck you. It’s hot and my hands are full. I’m sick of taking heat for exercising logic in my everyday life. And who talks shit on white T’s? Everyone knows they look good.

0 Comments:

Post a Comment

<< Home