Wolfram Speaks!

The O'Reilly Bioinformatics conference was a happy old pot luck. Proteins over here, starches over there, a vector-space model or two in the soup tureen. There were biologists just getting into computing, and programmers just getting in to biology, and a very smart few who had been doing both for many years. And there were winter refugees like me, grasping at any excuse to visit San Diego in February, reading frantically on the plane to try and catch up on my genetics. My closest claim to biological expertise is through my microbiologist uncle. But maybe heredity means nothing there. I don't know, I haven't read that far yet.

The biologists were a fascinating crowd. It is a mirror-world. Ordinarily I'm used to people from non-computer disciplines being quite shy about programming. Software to them means Outlook and Microsoft Office, running on Windows. Open source and scripting languages don't get a lot of respect - the real prestige is in 'enterprise computing' and languages like Java.

The bio people are totally different - they come from Unix backgrounds, they're used to dealing with nasty command-line tools, they've had bad experiences with belly-up closed-source software that leaves their data stranded. Perl is a serious language for them, because they have a thousand file formats to parse and interconvert. Programming is just another skill to pick up, like titrations or using the microscope. I guess a 'for' loop doesn't look that intimidating when you've done five years of post-doctorate research in protein synthesis.

To make it even more exciting, there is the electrifying sense of urgency. Every problem we face on the general Internet comes up in more acute form in bioinformatics. Imagine having Google searches that routinely take three hours, or having to cut and paste spam out of the body of your email messages, and you'll have some idea of what the biologists put up with. The data isn' t clean, or reliable, just voluminous. What they desperately want is tools to help them sift through enormous amounts of it, and make it intelligible. After dealing with educators and humanities types for so long, it's quite exhilirating to find people who aren't cowed by computers. These people deal with biochemical feedback loops, they know from complexity.

A big deal at the conference was Steven Wolfram, who gave one of the keynotes. That this man is not on a private island somewhere, perfecting a mind-control laser with which to subdue the world, says a lot about the rising cost of Pacific island strongholds. I was particularly eager to see his talk because he is such a cultural iconoclast in the research community. We don't normally think about cultural norms in science, because it is supposed to be a dispassionate pursuit of knowledge. And after all, many famous scientists were also famous misanthropes. But just because the regular social rules don't apply to scientists doesn't mean they don't have their own set of norms, equally strong. Some of them are strong enough to be taboos. You always cite other people's work. You publish results in scientific journals, and adopt an impersonal written style. You attend conferences and referee others' work.

A lot of this functions as a kind of sanity check. The rule is, if you stick to the conventions, we'll vouch for your reputation even if your actual results get wild and crazy. It's a mechanism form distinguishing the true genius from the incomprehensible kook. If you think about it, there are only a very few people who can judge theoreticians at the most rarefied levels on the merits, and their attention is limited. The rest of us have to take it on appearances, and on the opinions of qualified people we trust. So we depend on scientists to be rigid conformists in some areas, and to discuss their work with other scientists, oftentimes to teach. It all serves as proof that they are serious and coherent enough to be worth listening to, and lets other people evaluate them personally and professionally, like in any web of trust.

Wolfram is a problem becuase he has holed himself up for years, and done much of his work behind the barrier of a private software company. He huffed and puffed for twenty years before publishing a gorgeous book in the twenty-pound range, full of pretty pictures and of footnotes - but the footnotes all cite Wolfram. He took a lot of credit and made grand claims, many of them requiring years of work to corroborate. So now no one knows how much of it is the genius, and how much is the cook. His book is very dense, some of his ideas are very beautiful - but you can't help but feel suspicious.

And Wolfram isn't helping much. His talk was an attempt to condense the thousand pages in his book into forty minutes, but as time ran out it morphed into a product pitch for his software, Mathematica. He seemed annoyed that few people in the audience used it. Wolfram has a soft, lovely speaking voice, and a striking resemblance to George Costanza from "Seinfeld". It's pretty riveting. Wolfram's central thesis was that we should study his book and use Mathematica before doing any further work in biology - it was kind of like hearing a pitch for Mao Ze-Dong Thought, except that the Little Red Book is much easier to wave in one hand. We almost lost the presentation when he managed to spill a copious amount of water into his laptop keyboard. Four hundred biologists and one imposter sat quietly while the father of a New Kind of Science stood there draining his Dell. He seemed like a very nice man, not creepy at all, just a little bit lost in the aether.