Stats
Once fans see a handful of shows, they start to notice differences. They ask about what's rare, and talk about things that seem novel to them. Fans who've seen dozens, scores, or even hundreds of Phish shows relish an entire realm of this sort of discussion. The idea is of course not unique to Phish fans. Depth of investigation in any area conduces interest in quantitative variations. And the Phish.Net fostered that kind of discussion early on. In 1993, for instance, Andrew Cluley (then <drew@eskimo.com>) posted "Phish-Stats" for the Summer tour -- something he had done previously, and did for a few tours following. And others followed with their own efforts at producing "statistics&quit; about setlist variation. However, the posts were simple frequency distributions (such as indicating how many times was each song played, and at which show); none included even basic univariate summary statistics (such as mean or standard deviation), and bivariate statistics were mocked.
David "ZZYZX" Steinberg (aka "The Timer") <zzyzx@seanet.com> has put together an enormously fabulous web site, Phish Stats, that'll put facts and figures behind those crazy assertions you've been making on rec.music.phish. (While still essentially univariate, they include probability estimations, likelihood coefficients, and temporally weighted frequencies.) Find out how many shows did have one set of songs but none of another set, cross-analyze song orders by days of the week, and more. Or look at output Dave's already produced, such as for the current (or most recent) tour.
David also has a mailing list. For each show that happens, ZZYZX sends an email with a table showing: every song played, the number of times since it was last played, the last time it was played and the number of shows it wasn't played before that, the last time it was played before the last time and the number of shows it wasn't played before that. So, for example, set 2 of 12/28/96 looks like this:
If you think you're sharp at predicting what Phish will play, based on statistics or your own intuition, try your hand at Phantasy Phish, a game that ranks players based on how well they predict what Phish plays each tour.
"It's kind of silly when [fans are] are making pie graphs about set list openers. But then, I always liked a good graph.""
-- Mike Gordon, Detroit Free Press, 12/5/97"