April 26, 2013
The Hawk Trap
I don’t know how we got to this point, but the long-awaited grudge match between White Sox color commentator Hawk Harrelson and MLB Network broadcaster Brian Kenny (with occasional contributions from Harold Reynolds) took place last night. Everyone was polite, nobody got sent to the hospital, and Hawk launched a thousand indignant tweets. You can see the whole thing through the miracle of YouTube, if you have ten minutes to spare for Hawk to say five minutes’ worth of sentences twice:
So what we have is a discussion between someone who's practically a caricature of the old guard of baseball and someone in Kenny who is trying to immerse himself as much as possible in new analysis. And, of course, it started with Hawk asking, “Did any you guys see the movie Moneyball?”
Hawk had seen Moneyball, and boy, did it get his goat. He felt like the movie missed the point: “When you got Mulder and Zito and Hudson, you can write any book you want to. Because this is a game of defense, and pitching is the first line of defense.”
This is as good a time to mention this, not that I haven’t said it before and not that I won’t have to say it again. But let’s quickly examine the official rules of baseball. Don’t worry, we’ll stick to the very first page:
That’s baseball in a nutshell—you win by scoring more runs than your opponent, or put another way, more runs than you give up. You can do this two ways:
Except that good teams tend to be good at both. And the key point is that each run scored is a run allowed for the other team—there’s a symmetry in baseball between scoring and allowing runs. Hawk can focus all he wants to on one side of the ledger, but that just means he misses out on the other half, not that he’s found some key insight. He also seems to miss out on the point of Moneyball, which can be explained by this delightful exchange:
Ignorance of the other side’s point is the best way to prove yourself right. (To yourself, at least.)
And now we get to the crux of Hawk’s argument, doing the best job I can of capturing his unique manner of speaking:
I also said there’s a place in baseball for numbers. But, I also said that it’s the most overrated issue to come into baseball in the last 10, 15 years. And I stand by that, because it’s not ready yet. Down the road 40 or 50 years… when you can put some of those categories, you know you got your oh-bee-pee-ess, and all that, and the vee-oh-arr-pees. When they put in tee-dubya-tee-dubya and then interface those numbers with tee-dubya-tee-dubya, then you might have something cooking. And that’s the will to win.
Let’s get the minor objections out of the way first—there is no such thing as OBPS (and yes, I listened to that section of the clip a good dozen times to transcribe it for y’all, that is what he said), and while we at Baseball Prospectus are always thrilled to have a mention of VORP, it should be noted that it hasn’t been our primary replacement-level metric for, oh, about six or seven years now.
But that’s incidental to the thrust of his argument—that these things just aren’t ready yet, dadgum it. It turns out, though, that real baseball teams are using this stuff to win baseball games. And not just the Moneyball A’s and the Too Much Moneyball Yankees and Red Sox. It’s the San Francisco Giants, who just won a World Series and who have been the pilot team for a lot of new technologies like FIELDf/x. It’s the St. Louis Cardinals, who have quietly built one of the most impressive analytics operations in all of baseball. Numerous division rivals, even the tradition-oriented Minnesota Twins, are getting in on the act. Hawk may be surprised to find that even the White Sox themselves are not above using a number now and then, and have a general manager who can talk meaningfully about the concepts of sample size and regression to the mean.
And why are all these teams using sabermetrics, even though there are still gaps in our understanding? Because there are still gaps in our understanding! That’s not a belief contrary to doing good sabermetrics, it’s one that is central to it—Bill James got his start precisely because he was willing to accept that there were baseball questions that had yet to be answered. So there are things we can’t measure, and there are things we measure incorrectly. Yet people still use sabermetrics, and it still has value. How is that?
Think of how our knowledge of our own solar system has grown and changed over history. We started off with convoluted geocentric models that, while they did have predictive power, managed to get such things as “the Earth orbits around the Sun” completely backwards. Then Nicolaus Copernicus gave us the first complete heliocentric model, which managed to get the Sun in the right position and improved upon the geocentric models for predictive power, but made such basic errors as assuming orbits were circular. Johannes Kepler came along and gave us elliptical orbits and his laws of planetary motion, further improving the model. Later works introduced the idea that the sun itself is not stationary but also moving throughout a larger galaxy. And Albert Einstein’s theory of relativity has further increased our understanding of how our solar system works.
What do all of these models of planetary motion have in common? Well, one common feature is that all of them were wrong (give Einstein’s theories time, and eventually they will probably be supplanted as well). But each in succession has improved our understanding of how the universe works and our place in it, as well as our ability to make predictions that astronomers (and interested observers) find useful.
Meanwhile, you have some people who still think the whole shebang is a giant disk sitting on the back of an enormous tortoise and that this whole globe idea is nothing more than a conspiracy of cartographers, and that’s where you’ll find Hawk Harrelson.
Hawk is at least two kinds of extra special wrong. One is what Isaac Asimov described as “wronger than wrong,” where you’re clinging to outmoded beliefs and defending yourself by claiming that better ideas than yours are also incorrect and refusing to address your own deficiencies. Sabermetrics’ failure to be perfect is not a blanket justification to ignore all of it.
The other kind of extra special wrong Hawk stumbles into (in the same paragraph, no less!) is what Wolfgang Pauli called “not even wrong.” It’s when you make claims that can’t be refuted, and in doing so make claims that aren’t worth refuting. Hawk talks about “the will to win” (and believe me, he keeps repeating this) being the most important thing in baseball, and apparently it’s judged by how many wins you have. If you’ve won a lot, you had the will; if you didn’t, well, you didn’t.
People like Hawk will always roll out this line of thinking because it can’t be disproven, so they never reach a moment where they’re refuted and forced to actually quit. So what they never notice is that it’s also totally meaningless; you can only ever figure out who had the will to win after the fact, at which point it’s too late to do anything about it.
Sabermetrics, meanwhile, works the way science does: by making predictions (scientists call these hypotheses) that you can test. In testing them, you can find evidence for or against your prediction. And over time, as you come up with more hypotheses and do more testing, you inch closer to a better understanding of what it is you’re studying. One side effect of this is that you end up making predictions that have been tested, which turns out to be useful to people who have to make decisions about the future. So by trying to do baseball science, sabermetricians found themselves doing things that general managers could adopt and use to their own purposes. Hawk has noticed this, but it hasn’t exactly made him happy. In fact, it has made him kind of angry, which prompted a wonderful rant.
“It’s gotten people fired,” Hawk said. “It’s gotten managers fired because they had to manage a game they didn’t know. Sabermetrics disdains defense, disdains speed, bunting—bunting!—hit-and-run… A friend of mine who was a scout, God rest his soul, told me there was a manager who had to call up in the late innings of a ballgame to a GM’s booth and ask him if he could bunt, because it was a close ballgame.”
The three of them debated this for a bit until Kenny asked point blank, “Wouldn’t you want to know what the percentages were?” And Reynolds and Harrelson both in unison shouted “no!” At this point Reynolds took over for Hawk, asking the question, “But are the percentages going to tell you what’s going to happen at that exact moment?”
No, of course they’re not. Nor is the manager’s gut. What they are going to tell you is the likelihood of certain things happening, with a certain amount of accuracy (and the more and better inputs you have, the more accuracy you can get).
Let’s take the intentional walk as an example. If you intentionally walk a guy with two outs, typically the next batter you face gets out and the intentional walk “works.” On the other hand, if you decide not to issue the intentional walk and face the current batter with two outs, typically he gets out and challenging the hitter “works.” This is because even the very best batters (aside from a four-year stretch by Barry Bonds) make outs more than half the time.
So if you look at one “to IBB or not to IBB” decision in isolation, either way it works most of the time. But one decision is more right than the other—maybe not by a lot, probably never by as much as you might guess from the more impassioned reactions of some armchair managers. And you won’t notice the difference just looking at this one instance. But if you step back and look at a whole lot of them all at once, then you can see how the small difference accumulates over time.
That’s why the percentages matter—they cannot tell you exactly what will happen right now, but they will help you make the right decision often enough that over time you’ll accumulate more wins than you would have otherwise.
That is the game managers know, and it works towards the goals that managers have—winning ballgames. It can be difficult for managers to incorporate new information and new learning and process it in real-time in a ballgame. And yes, some of the responsibility falls on the sabermetricians and the front-office types to make the information manageable and accessible and convenient to use. But some of it falls on the manager to adapt and incorporate every advantage he can get his hands on. Some managers won’t be able to bring themselves to do what it takes to bridge that gap, and yes, they will be left behind. Because there will be other managers who take it upon themselves to learn these things and take advantage of them.
I guess you could say it’s because they have the will to win.