Monday, May 28, 2007

Computational doping

As a fan of road cycling, I've been following the Floyd Landis trial and recent doping revelations (mostly via velonews). The whole affair feels a bit like a witch-hunt at a wizarding convention, in that doping sounds sufficiently commom back in the late 90s that lots of riders probably could make confessions. The question remains whether they will be burned at the stake fueled by collective desire to do something about this practice now.

At the same time, getting into a new research project reawakened my envy at my husband's memory for research results. He's very good at recalling the gists of projects, who worked on them, and how they relate to other projects. He builds mental roadmaps of research areas in his head as he reads, and can recall and access those models easily. My memory is far better for birthdays and people's stories, which isn't usually as useful for professional computer science.

As this skill I desire is all about managing information, though, I should be able to imagine a software system to help me build, manage, and access a store of professional knowledge. The technical challenges to this notwithstanding, I found myself pond ring whether using such a tool would be analogous to doping in sport.

There are clearly some similarities: person uses artificial substance to enhance their ability to perform; person may get more money and more prestige as a result of better performance; person isn't competing solely on their natural talents. There are also some strong differences: sport has one winner (individual or team) whereas research competition usually has several; elite sporting is more financially lucrative than elite research; the public gets emotionally caught up in sport.

High-profile awards such as a Nobel or a Turing (the analogue for computer science) are exceptional in having single winners and non-trivial financial rewards (though not sport level). Imagine a Turing-award winner attributing success partly to a program used to help manage idea infrastructure. The awardee would probably get bonus points for having written the program, but even if someone else wrote it, I can't imagine the computing press calling to revoke the reward.

Is this because what prompts a research award is the ideas instead of the person (the person gets credit, but the idea is what captures the imagination). The same emotion arises in sport: I don't want to let the doping allegations taint Floyd's stage 17 ride in the 2006 Tour de France because it was so frigging _beautiful_. I love cycling for those moments of explosive power where the person disappears and only an instrument highly tuned to its environment, internal condition, and condition of its competitors remains. At that moment, I don't care whether the riders are aided by technology: the joy is my drug, and I just want to experience it.

But when the ride is over and the podium gone, doping takes us past the performance and focuses on the person. One standard argument against doping is that it is unfair, disadvantaging those who choose not to dope. The usual counterargument is that doping can help some riders be more competitive with those with more natural abilities. My memory-enhancing software would fall into the latter, but an already top person could use the same software to widen the gap again. Unlike drugs, my software would cause me no physical harm. In an ideal world, nobody would have to choose between winning and harming themselves, but winning often requires risk and compromise. How is sport truly different?

In the end, research and sport have different value systems, different notions of what makes a heroic performance. Both value results. In research, purity of results comes from replicability; collaboration is both welcome and expected to be visible (via attribution and citation). In individual sport, purity manifests as an unaided athlete achieving the improbable. Collaboration is rampant (top cyclists use a myriad of bike designers, clothing designers, wind tunnel simulations, etc), but expected to be invisible. Doping destroys invisibility and becomes an affront to purity. If our culture viewed scientists as heroes on par with athletes, would my software become distasteful? I suspect not, because we still view research as "hard" and athletics as inborn. If our athletes are simply trying to live up to our expectations of them, perhaps we need to ask whether it is us, not them, who are really on drugs.

No comments: