Consider a familiar, mundane, urban situation: You walk into an ATM vestibule in a bank. Your arrival has been preceded by other customers. No queue exists. But a ‘queue’ forms nevertheless and it deploys a simple algorithm: You simply wait till everyone that was there before you takes his or her turn. You don’t care in which order the rest of the ‘crowd’ uses the ATM; it is sufficient that you not use the machine before anyone else was there. You presume–and you would be right–that everyone else in the vestibule is following the same principle. If there are any ‘jumpers’ they will–in all probability–get called out; you don’t need to do this enforcement for those that were before you, they will take care of it. You might however, need to do this for someone who shows up after you and tries to use the machine before you. There is no conversation, no explicit agreement among the users. Rather, an implicit convention, the contours of which are familiar to the ATM users from a host of other situations requiring an efficient ‘crowd procedure’ to solve a problem, is put into place and followed. An extraterrestrial watching this scene would notice that a gaggle of people standing around in various positions in this enclosed space had somehow agreed upon an ordering in which to use the machine without negotiating among themselves.
Crowds solve co-ordination problems like this all the time, without explicit instruction (and in doing so, they solve problems that in their abstract form are not amenable to tractable mathematical modeling or solution). A classic well-known example is, of course, the sports fans who file out of a stadium and empty it quickly, and efficiently. As lines of fans exit into the aisles, they quickly fall into an alternating scheme that lets every other person go ahead into the exit. Later, on the exit ramps, things can become chaotic (and sometimes even go terribly wrong when stampedes occur). But the preliminary filing out into the exit ramps is almost always orderly. Or consider conference attendees asking questions of a speaker using a two-mike arrangement (one in each aisle of the auditorium): the questioners take alternate turns at each mike, a convention that does not need to be enforced. (There is a slight wrinkle here; sometimes one line grows quicker on one side than the other, and then, the questioners on the other side need to recognize this and grant priority accordingly. But by and large, the alternation works).
Because I live in a large populous city, I have ample opportunity to observe successes and failures of this kind of social software. On 34th Street, at the Herald Square station, crowds persistently fail to solve a simple co-ordination problem. Sometimes only two turnstiles will be available at one entrance for users to exit and enter. Bizarrely, the simplest, most efficient solution is never arrived at: one line should use one turnstile for exiting, the other should use the other turnstile for entering. Instead, an undignified hustle to try to sneak in before another user can exit takes place (and vice-versa). Watching this is mortifying: it’s like witnessing a demonstration of collective stupidity. It’s puzzling too: why can’t the crowd figure it out? The absence of an explicit shaming mechanism has something to do with it, of course. Getting called out for queue-jumping is one thing; being accused of simply trying to get ahead in a free-for-all isn’t. Conventions and mores like these–at some level–underlie the failures and successes of almost all crowd solutions to co-ordination problems.
These sorts of co-ordination mechanisms have been studied by many different disciplines–economics, computer science, philosophy, sociology for instance–precisely because they manifest themselves in so many ways. (Think about language for instance, in the way that Wittgenstein analyzed it in the Philosophical Investigations.) And they remain endlessly fascinating because the kinds of co-ordination problems that social interaction can throw up promise to remain diverse, unpredictable, and imprecisely amenable to the kinds of solutions and failures noted above.
Samir Chopra 