Saturday, February 16, 2013

the sports guy's sports narratives

Arguably the most prominent writer at America's most prominent sports site, ESPN's Bill Simmons frequently promotes an idea he calls the Ewing Theory:
Q: What's the Ewing Theory? Where did it come from? 
The theory was created in the mid-'90s by Dave Cirilli, a friend of mine who was convinced that Patrick Ewing's teams (both at Georgetown and with New York) inexplicably played better when Ewing was either injured or missing extended stretches because of foul trouble.
First of all, let me point out that this is not a theory but an observation. It might be the seed for a theory and maybe some more investigating would be warranted, if one cares about such things, but it's not a theory and it doesn't become one later in the article either. Furthermore, he's been promoting this idea since 1999 and is currently the editor of ESPN's Grantland, which he fathered into existence (with sex!) and which currently  includes as staff a guy named Bill Barnwell who is actually good at this stuff and who repeatedly points out the problems with Simmons' bread and butter -- the post-hoc narrative, of which this is a perfect example. Simmons is using Alex Jones' methodology and getting handsomely rewarded for it!

Instead of getting Barnwell on the case, Simmons discusses a bunch of arbitrarily chosen candidates -- Barry Sanders, Nomar Garciaparra, Drew Bledsoe -- and tests his "theory" against them using arbitrary, shifting standards. How far did they go in the playoffs? How good were they (based good Simmons thinks they were)?

The right way to proceed is to choose parameters and then run through the data. The parameters will necessarily have somewhat arbitrary endpoints but it'll still tell you what it tells you, according to those parameters. For example, with basketball, you could say that any player with a PER (player efficiency rating) higher than 22 over the 150 games prior to injury qualifies and for baseball, anyone with a WAR (wins above replacement) over 3.5...and so on. Then check the before/after win-loss records or some other team results-based criteria and make your data set as large as possible, and so on. So you'll know if, generally, losing a great player has a positive impact on a team (I won't ruin the surprise for you). You might discover some common features in those cases where a team performed better after losing a star. If so, you can come up with a new theory and run the data again.

Now Simmons knows he has Barnwell, and is at least somewhat aware that he himself doesn't know what he's talking about but doesn't let that stop him. He even mentions Barnwell in the piece:
[Dan] Marino's career postseason stats: 18 games, 10 losses, 4,510 passing yards, 32 TDs, 24 picks, 77.1 QB rating. I'd keep going, but I don't want to provoke Bill Barnwell into writing a 6,500-word piece about how it's stupid to judge QBs by their playoff stats. 
Again, this is one of Simmons' pet ideas, and a typical one at that, and he's the most prominent writer at the most prominent American sports site on the internet and has no concern for using sound methodology even when it's readily available to him. And his readers don't care much either. Check Barnwell's facebook recommends against Simmons'. It's safe to assume that Grantland readers are, at best, OK with choosing to be entertained by nonsense and at worst, choosing to be fooled by it. Even when there's no political agenda directly involved. I'm not even criticizing them here, just describing what I see. And, mea culpa, I like reading Bill Simmons articles! He regularly has me slamming my head into my desk but, like Alex Jones, he's entertaining, though I'd much rather read Barnwell.

(This post feels like it doesn't quite get to a point but hopefully I'll get around to a follow-up comparing political media, sports media, and fantasy sports media; the first is the worst, the last is the best, and there are reasons.)

No comments: