Anyone who follows baseball, whose World Series ends this week, knows that the national pastime has changed more in the last five years than in the previous 50. It hardly seems the same game. Starting pitchers go five innings or less; they throw more curves and sliders; there are defensive shifts on almost every batter, and those batters are usually swinging for the fences, and succeeding more than ever before, or striking out when they don’t succeed; the stolen base is practically obsolete, as are, increasingly, the sacrifice bunt and the hit and run. Eventually, the ground ball may be too. This is a revolution, but it hasn’t been organic, arising from changes in the athletes. It has been imposed … by data.
What happened to baseball is one small example of something that has been happening for years across America. We are all now inundated by data, all of us living a data-driven life at the expense of what one might call a more human-driven one. In ways both obvious and subliminal, statistics are changing our institutions, including our politics — and they are changing us. It isn’t too much to say that they are becoming determinative.
Let’s start with baseball. When I was growing up and first became a fan many, many years ago, the game was something of an art. It was full of guesswork, intuition, tradition, idiosyncrasy and, of course, talent. Grizzled managers didn’t consult iPads, fielders didn’t pull cards out of their pockets for each batter to get defensive positioning, and batters didn’t watch a video of their last at-bat in the dugout — all things that managers and players routinely do now. The game was much looser than that, much more instinctive. It was all gut and nerve and hunch — a bunch of guys feeling the game. To quote the immortal Yogi Berra: “You can’t think and hit at the same time.”
That’s mostly gone now, so much so that when Houston Astros’ manager, AJ Hinch, sent his pitcher, Charlie Morton, out to the mound for the sixth inning Saturday night, going against the “book,” the announcers kept marveling that Hinch was “trusting what he sees,” “trusted his eyes,” rather than referring to the statistical tables that would have told him to remove the pitcher — the idea being that just about nobody trusts his eyes anymore. You trust the data.
This big revolution in baseball, maybe the biggest, which has spread to every other professional sport, is called “sabermetrics,” and it began back in the 1950s and then caught fire in the 1970s, when a few baseball diehards and stat nerds, most notably a former factory security guard and amateur statistician named Bill James, who has become something of a baseball deity, discovered that data was really a treasure trove if you knew how to interpret it and, more, use it.
John M. Ward of the New York Giants’ baseball card portrait.
Card set: Allen & Ginter World’s Champions (N28), 1887. (Image couresty of the Library of Congress)
Card set: Allen & Ginter World’s Champions (N28), 1887. (Image couresty of the Library of Congress)
James and his confreres didn’t believe in trusting their eyes. They felt that the statistics baseball generated often told a different story from the one you saw, and, more, that the traditional stats didn’t really tell you what you needed to know. Better stats could tell you who was really adding to the team’s victory total and who was detracting from it, who hit which pitcher best and which pitcher had a batter’s number, which batters were better at night and which during the day, and a thousand other things that would radically alter the practice of the game.
And they were so good at it, so right, that baseball team after baseball team began relying on new and sophisticated stats to determine just about everything they did. You didn’t play hunches anymore. You played the odds. Michael Lewis’ best-seller, Moneyball, and the movie based on it, told the story of how Oakland Athletics’ GM Billy Beane rebuffed the old-timers and swore by data instead of eyes for his decision-making. It’s a film in which sabermetrics is the real hero and tradition the real villain, whether we want to think of it that way or not. Nowadays, in order to be a general manager or even a manager you have to be a data freak.
No one, least of all me, is going to argue against statistics and scientific data. (Well, I suppose Republicans would argue against them.) The more we know, the better, and the more explanatory our data, the better. But there is nevertheless something to the quote Mark Twain attributed to British Prime Minister Benjamin Disraeli that there are three kinds of lies: “lies, damned lies and statistics.” It is not so much that statistics lie, or even that, as my late father, an accountant, used to say, “Figures don’t lie, but liars do figure.” And it isn’t even that statistics, taken at face value, are missing something essential, though I think they are — a human element that breathes life into them. If we didn’t have that human touch, AJ Hinch would have pulled Charlie Morton because the data shows that a pitcher is less effective the third time through the order, and Dodger manager Dave Roberts certainly wouldn’t have let Joc Pederson, a struggling hitter, bat in the ninth inning … when he hit a three-run homer.
There is something to the quote Mark Twain attributed to British Prime Minister Benjamin Disraeli that there are three kinds of lies: ‘lies, damned lies and statistics.’
The deeper danger, I think, is that statistics, as they are deployed so often now, aren’t just explanatory; they are dictatorial, meaning they impose themselves on whatever is being analyzed and tryrranize it. In baseball, for instance, the old stats — home runs, ERAs, RBIs, batting average, slugging percentage — were extrapolated from the game to give us some sense of value. We thought a player with a high batting average was a good player. Sabermetricians tell us otherwise. They use their stats not just to measure value, but to bend the game to conform to the statistics. Or put another way, they are less interested in what had happened than it what it portends.
Take “launch angle,” which is the hot baseball stat these days. The idea is that there is an optimum angle at which the bat strikes the ball in order to get a hit, or better yet, hit a home run. (And, yes, there is a comprehensive data set on the launch angles of just about every batted ball.) As a means of explanation, I suppose, there is some interest in knowing which players have the “best” launch angles. But that’s not what has happened when the data is mined. Instead, just about every player — and organization — has become obsessed with launch angles as a way of making sure that batters get that ball up over the wall, and they adjust accordingly. What used to be a very human endeavor — learning how to hit a ball — has become a mechanical one. It is as if data analysts were to tell a visual artist which colors to use or a writer which words work best.
Failure is a part of being human. It is a way we grow, develop, learn and ultimately gird ourselves for the inevitable defeats we will encounter. Failure isn’t necessarily to be avoided; it is to be understood. A life without failure is a life not fully lived.
And to what end? The end is to minimize failure, a pursuit which certainly has its place in many areas of our lives, but not in all of them. Not to get too philosophical about it, but failure is a part of being human. It is a way we grow, develop, learn and ultimately gird ourselves for the inevitable defeats we will encounter. Failure isn’t necessarily to be avoided; it is to be understood. A life without failure is a life not fully lived. A game in which failure is measured by the inability to adjust one’s launch angle is a game diminished.
We see this data-drive most clearly in sports, which are full of stats, but everywhere numbers crunchers are striving to find the formulas for what works and what doesn’t — which typically means, what gets us to do what someone wants us to do. This is the primary function of algorithms — formulas designed to let merchants know what we will want. Google operates on algorithms expressly designed to give us the information and the ads it has surmised we want. Facebook operates on algorithms to send us the news (and the ads) it has surmised we want. Amazon operates on algorithms to advertise books to us it has surmised we want. Netflix operates on algorithms to give us the TV programs and movies it has surmised we want. And increasingly, political organizations are searching for algorithms that will allow those organizations to give us the political positions it has surmised we want.
And so it goes.
There is a certain circularity to this: You take our data and then provide us with something customized to that data. But just as baseball statistics rob the game of some of its beautiful mess, algorithms rob us of the opportunity to be exposed to things we haven’t already endorsed. They place us within a prison of our own predetermined desires. We only get what we have wanted previously and nothing else. This is a limiting way to live. And like so much else in our culture, eg., social media, which I discussed last week, it tends to push us further within ourselves and away from others. Algorithms, like sports statistics, promise predictive success — in this case, the predictive success of being satisfied with what has always satisfied us in the past.
The cultural implications of living within our very own feedback loop are large. So are the political implications. Already, Americans barely extend themselves beyond their own political predispositions. We wall ourselves off from anything we don’t like. Once politicians find the political equivalent of launch angles or get the hang of creating algorithms that can push our buttons, we may lose the humanity of our political discourse too. We are clearly on that path now.
Maybe this is progress. Maybe this is just the direction of the world. Still, baseball, I think, used to be a better game before it got all computerized with analytics. But that’s just a game. A politics that figures out precisely how to incite us or pacify us or distract us is something else. And I am not so certain that something else is an advance.
No comments:
Post a Comment