Fair Science

I was a judge at the Science and Engineering Fair of Houston on Friday. Sounds grand, but it's pretty simple, really: you go around with a clipboard, talk to the kids, try to ask some meaningful questions, then you rate their projects according to a bunch of boilerplate questions that don't sound all that different, and where you have to put a mark between 0 and 10, even though you'd be hard-pressed to explain the difference between a 6 or a 7. Then you add up your marks, normalize, combine with the other judges... and then the fun starts: you're discussing why you thought project A was better than B, or the opposite, and you learn from the other judges and come to a consensus. In reality, most of the projects are good, and you end up wishing that you could make everyone win.

I saw twelve computer science projects, all done by senior students individually (for those of you not used to the US school system, this means last-year students, so they would be about 18 years old at this point).

One guy built an FTIR multi-touch interaction table, just like the one some people working for me built in the summer of 2006 at our lab in Boston. "My guys" are about 25, but this kid is 18 and in high school, and did pretty much the same thing. True, he mostly focused on the construction of the device rather than on the computer part... but that's exactly what we had found to be the challenge ourselves.

Another guy (there were a couple of young women in the group, but only a couple, as usual most of the engineering or computer science projects were presented by male students) had done a simulation of the growth of a colony of cells — inspired by Stephen Wolfram's book that claims that all complex systems in nature can be described accurately by simple computational systems. This was of course reminiscent of Conway's "Game of Life" from a Scientific American article published in 1970, which those of us who learned computer science in the 1970s were all crazy about.

Another young man presented as his science project something he's worked on for two years: a chess-playing program in which he combines several sophisticated algorithms to prune search trees (starting with the α-β procedure I learned 34 years ago...) and got his program to apparently play at grand master level.

And a fourth candidate simulated the flow of traffic through a city grid, including the fact that some drivers are more aggressive than others, comparing what happens when traffic lights are on the same cycle, as opposed to the "green wave" where the lights turn progressively green along some routes, so that in theory traffic can flow uninterrupted once released from a red light. He found that, counter-intuitively, the "green wave" method is usually not effective, unless the timing of the greens is within a very narrow range, and that it is very sensitive to how heavy traffic is: if traffic is heavy, for example, people may be slowed down just enough that they get to the next light just as it is turning red, and this repeats itself, so in fact you hit all red lights instead of all green ones.

Altogether, this was rather impressive. What fascinates me about any sort of poster sessions, science fairs, and the like, is that they ask me to judge, when in fact I feel that I am learning from the students. It's humbling to see so much talent at such an early age.