Sunday, December 14, 2008

Review - Blindsided

I have, in the past, admitted a weakness for the Big Theory, the grandiose claims of a writer who believes that he or she has found a key to understanding that has eluded all others who have come before. Back in March, I admired Amy Chua as she attempted, however ultimately unconvincingly, to provide a new look at the rise and fall of empires.

At the same time, if you get carried away with self-praise (as in The Black Swan), to me, it undercuts any hope of connecting; no matter how much of Naseem Taleb's book I liked, there was so much chest-thumping that I found the book more irritating than convincing.

Sports seems to bring this kind of thinking to the fore, as someone invents a new statistic and becomes fixated on their creation as the linchpin of a new science, rather than as a (possible) addition to the accretion of knowledge. Earlier this year, I reviewed a book called Stat One that I found particularly obnoxious in its claims to revolution, especially as it ignored everything that had previously been done in the field called sabermetrics (the attempt to use statistics to gain insight into sports).

In that review, I spoke of Bill James, the father of modern sabermetrics, as the model to which people seem to aspire; he was not the first to apply statistics and computers to the study of baseball, but he did ask the most interesting questions and wrote in a marvelously witty style. All who have come after are inevitably compared to James, and many would like to see their work placed next to his.

And this is what KC Joyner, ESPN's so-called "Football Scientist," tries to do in his book Blindsided: Why the Left Tackle Is Overrated and Other Contrarian Football Thoughts (2008). He invokes James as his model, telling us that he is going to use statistics to explore football questions that have never been looked at before. In doing so, he undercuts a pretty entertaining book with expectations that it cannot, and does not, live up to.

Let me address his most important conclusion first, one that has nothing whatsoever to do with stats. Joyner contends that the NFL should open up their game films to researchers, allowing regular people to do what Joyner somehow gets to do: pore over actual footage to examine various questions. His analysis and reasoning here is dead on, that it would cement ties to fans in a way that would be profitable and favorable in the long term.

It is the rest of the book that, for all its virtues, seems disappointing. Disappointing, I think, because Joyner oversells what he's doing. For example, he chooses the best defense of all time. So you expect him to come up with some kind of statistical framework for this, but he doesn't. He chooses three defenses, the Steelers of the '70s, the 1985 Bears, and the 2000 Baltimore Ravens. He looks at them in three ways, ascribing the method to Bill James: peak performance, performance over a longer period of time, and position-by-position match-ups. He quickly discards the Ravens, then compares the other two. The Steelers are given the clear lead going into the positional comparisons, and they then blow away the Bears in this head-to-head contest. That I agree with the conclusion that the '70s Steelers should be considered greater (though there might be some other teams on the short list) is less important than the reality that Joyner doesn't use any real facts to determine this.

This is a pretty consistent pattern. Who should be in the Pro Football Hall of Fame? That is a fine question to ask, and Joyner's writing is incisive and entertaining. But none of it is based on statistical analysis, just on the number of times a player was selected to an All-Pro team at the end of the season. This may be the best one can do, especially for the early players, but it's not really what we were promised. Concluding that Jerry Rice is the best receiver of all time is hardly "contrarian," and the use of stats is pretty basic.

The problem, of course, is that football just doesn't lend itself to the kind of analysis that Bill James (and, now, many others) has done for baseball. Baseball is a series of discrete, well-recorded acts in which intent is pretty clear. Football (and most other sports) is a collection of continuous acts, most of which go unrecorded, and the intent of the players on an average play is obscure.

To illustrate what I mean, we generally know exactly what a baseball player is trying to do. Hit, bunt, catch a ball, and so forth, these are all actions that lend themselves to being counted and evaluated. (It is the "other" acts that are more difficult; it's hard to know exactly what a pitcher's plan is, so we don't really know if they executed properly and were unlucky, or if they threw a pitch in the wrong spot.)

In football, on any given play, there are 22 people running hither and yon, we don't know what their intentions are, so we can't know if they're doing well or not. The end gets around the tackle and sacks the quarterback, and we blame the tackle; we don't know that the fullback was supposed to stay in and help with the block, so the tackle was running a slip block and intending to run upfield.

As a result, football is inherently less amenable to statistical analysis than is baseball. It's possible that film study and other techniques may be able to assist with some questions (and James has always understood that there are boundaries between the knowable and unknowable in baseball), but, sadly, always relatively fewer than in baseball. Blindsided pretty much ignores that distinction, and it weakens what is otherwise an entertaining book by a fellow who knows football. Joyner either needed to claim less, or dig a little harder to determine which questions might yield to the tools he does have.

[I'd be remiss if I didn't point out that Joyner is not exactly pioneering here, either. Bob Carroll, Pete Palmer, and John Thorn tried to do just this kind of analysis some time ago, most notably in The Hidden Game of Football. It was a little more successful, I think, but only a little, due to the previously-discussed intractability of the game itself.]

No comments:

Clicky Web Analytics