The proper use of WAR

It's a shame that the great baseball reporter Murray Chass is so devoted to being an internet troll, because there are lots of internet trolls and few great baseball reporters. Even so, being an internet troll doesn't make one wrong, and as he often is, Chass is more right than not in the final item of his latest blog entry, where he mocks the use of statistics like WAR in awards voting.

He's right partly because WAR and similar measures are just too imprecise to be useful in discriminating among a lot of good players even on a strictly mathematical level—that takes a more careful analysis—but also because award voting is interesting mainly for the way it provides an objective record of subjective opinion. Think of the annual Pazz & Jop Poll. It may be cringemaking to look at the 1985 results and see that Scarecrow finished above Rain Dogs and Dream of the Blue Turtles beat out Fear and Whiskey, but it also tells you something about taste and tastemakers 25 years ago. Of course baseball is better suited for objective analysis than music is, but that's all the more reason to want this kind of voting to not just be a matter of making an ordinal list of someone else's calculations.

The sad part of the idiot back and forth of which Chass and people like him are half is that both sides miss out on just what makes statistics like WAR so useful, which is that by putting all seasons ever on a single scale, you can answer questions in ways you otherwise couldn't.

To give an example, one of the great mysteries of baseball history is what exactly Ted Williams would have done if not for the wars. You don't need WAR to answer this; you can just give him credit for seasons like those he missed in the war, total up his numbers and come up with some sort of line. WAR helps bring some precision in, though. In the seasons before and after he left for World War II, Williams averaged 11.4 WAR; in the seasons before and after he left for Korea, 6.9. Total everything up and he seems to have lost about 45 WAR.

That doesn't mean anything in itself, but saying that the value of those lost years was about equivalent to Ralph Kiner's career puts it in context. Being able to say that if you credited Williams with five above-average years of three WAR and 10 average years of two WAR and then added that to what he lost to his combat years you'd essentially end up with the career of Reggie Jackson or Roberto Clemente does the same. The point isn't whether Williams was worth 11.8 or 11.3 in 1941, or how that compared to Joe DiMaggio, it's that by reducing seasons to rough numbers you can group and compare them, and, say, come out with a new appreciation of how unbelievably good Ted Williams was.

Some might say that this is a moron enterprise, and that the only things that matter are the smell of the grass, the hum of the fastball and the look on a player's face as he's taking off his uniform in late August. Fair enough, but I spend enough time in press boxes and clubhouses to call bullshit on anyone who tells you a sportswriter can tell you much about any of that. You can't consistently tell a ball from a strike from a seat in the press box, and even a diligent reporter is rarely going to get much of a tell on anything past how a player acts in highly artificial situations over the course of a year. Sportswriting, both good and bad, is built on supposition, intuition and trying to trick people trained not to say anything compelling into giving clues that hint at their true thoughts and character—it's an art, in other words, of rough approximation. If WAR et al. are of any use in this area—and they are, so far as they allow one to say that Roger Clemens was more or less as good after leaving the Red Sox as Sandy Koufax was in his whole career, the sort of broad point about which one could sustain a very good bar room argument or write a decent book—they're useful tools. And if they're useful why bring them in for grief? It isn't the hammer's fault if someone tries to use it to cut a log in half.

Comments

[Stevens]

Just semantically tickled by using WAR to determine Splinter's value while actually at war.