Not So Super
I always feel weird when I go to an event that's supposed to benefit some kind of grassroots community, only to discover that it's mobbed with harassed-looking middle-class people covered in press badges and bearing large cameras. I'm not under the indie rock illusion that an event has lost its purity if the media are there -- otherwise I'd have to kick myself out of my own parties. But I must confess to a substantial feeling of disappointment when I arrived at the much hyped "supercomputer flash mob" at the University of San Francisco April 3 and discovered the press practically outnumbered the participants.
The idea for an ad hoc supercomputer built out of 1,000 separate networked computers brought in by volunteers was terrific -- it combined the nerdy passion for flash-mobbing (what happens when a whole bunch of people in electronic communication with each other converge in one location for a short period of time, do something preplanned, then disperse) and crazily fast processing power. It seemed inevitable the thing would work. After reading preview articles about the event in the New York Times, New Scientist, and (of course) on Slashdot, I was worried I wouldn't be able to penetrate the hoards of geeks converging on the Koret Center, eager to hook their boxes up to a little piece of technohistory.
If the supercomputer flash mob worked, it would mean massive calculations involving things like weather patterns, astronomical events, and genomes could be done outside fancy labs and rich corporations. You'd call up a few hundred of your closest friends, network your boxes and hack the mouse genome for a day. Or create a complex simulation of a hurricane. I imagined high school students creating supercomputers on the fly and doing science projects that might compete with whatever stupid thing Craig Venter has been doing lately.
But, sadly, the USF supercomputer was a classic case of "what if you had a flash mob and no mobs came?" The cavernous gym where the supercomputer was to be housed was full of long, empty tables covered in fat bunches of cable ready to be plugged into boxes that never arrived. Most of the computers present were hardly the result of eager flash-mobbers, but instead had been donated by USF labs, Hewlett-Packard and eLoan.
As the media swarmed over the organizers, I sidled to the back of the room to talk to a group of students who were sitting in a circle playing cards. They seemed more excited about the free pizza they were about to eat than about the prospect of building a supercomputer. Most of them admitted they were there because they were promised extra-credit points for attending by their computer science professors.
Where were the true nerds? I found a small group of them outside the supercomputer room, slouching over their laptops and looking bored. One was complaining about how he couldn't get his computer's built-in camera to work under Linux. "Why aren't you in there?" I asked him.
"Got here too late, man," he said with a shrug. "Eleven a.m. is too early for a flash mob." Actually, the "mob" had started at 8 a.m.; the supercomputer was supposed to benchmark at 11. I had to admit the hour did seem punishingly early, especially for the sort of people who had probably spent most of the night before running intensive port scans on SCO.
Despite their inability to attract a mob to their project, the USF supercomputer geeks should be given credit for starting a very cool media meme. Several other groups have already announced plans to attempt similar experiments with ad hoc supercomputing. So the supercomputer flash mob will happen -- it just won't be in USF's gymnasium. I suspect the Chaos Computer Camp, an outdoor hacker event held every four years in Germany, would be a better place to stage such an experiment. Here's hoping for a supercomputer flash mob at CCC 2007.
In the meantime, we should all be learning more about mob theory so that next time our friends try to create a flash mob, they won't be forced to entertain themselves by talking to a bunch of dweebs from TechTV. How can we translate the online mobs that swarm our favorite social networks into real-life mobs? Certainly, Howard Rheingold has some ideas in his latest book, Smart Mobs.
But why read a bunch of well-researched sociological analysis when you could just check out the latest findings on mob theory as featured in the Higher-Lower Mathematics Review? Among other things, you'll discover mob theory allows you to claim a proof is true simply by threatening to incite a riot. Included in this fine publication are also instructions on how to do several other excellent things with math, such as prove that "every horse has an infinite number of legs" and that "everything is the same color." Armed with this kind of logic, it should be no trouble for you to get 1,000 hackers to wake up at 8 a.m. and network their 1.3 GHz or higher machines in a college gymnasium without promising them sex, doughnuts, or free iPods.
Annalee Newitz (firstname.lastname@example.org) is a surly media nerd who has figured out a way to possess an infinite number of cats who eat a finite amount of food. Her column also appears in Metro, Silicon Valley's weekly newspaper.