Management Lessons

Phil Garner, manager of the Houston Astros, gave a press conference the day after the epic 18-inning game that the Astros won last week over the Atlanta Braves. During the course of the questions, he was asked about his managerial style, and in particular whether his making of particular decisions was "a seat of the pants kind of thing . . . a gut feeling." Garner's response is instructive, both because of the ground that it covers and because of the place where it ends up. I think that there are some lessons to be learned here, lessons with applicability beyond baseball -- dare I say it, lessons about what social science is for and how it relates to social practice.

"Without knowing exactly which moves you're talking about it'd be hard to say," he began, noting from the outset that particular moves have particular rationales and implying that no single kind of decision-making captures or explains all of the choices that one makes in the course of managing a baseball game.

"I do -- sometimes by the gut, but most everything's calculated." Here Garner introduces two specific grounds for making particular decisions: gut instinct, and rational calculation. He is honest about the fact that some moves that he makes are simply the product of instinct; presumably this instinct stems from experience, so it's not entirely random or mysterious, but it's not directly capturable in rules or procedures. Contrast this to rational calculation, which is a type of decision-making based explicitly on rules and procedures -- what works in situation X and what doesn't.

If everything stemmed from rational calculation, then there'd be no point in having a manager at all; a book of procedures would suffice, and if there were someone called a "manager" involved at all his job would just be to look up the situation in the book and apply the appropriate response. Any agency that the manager has would disappear, as he would become merely the throughput for a set of factors over which he exercised no influence or control. Indeed, in such a situation the manager's only effective exercise of agency would involve ripping up the book, going off course -- acting on instinct rather than according to the rational plan.

The conventional Social Science Project, I think, involves making as much as possible calculable, eliminating the element of uncertainty that comes from not really knowing what the result of some move or decision will be. Garner's abstract comments to this point remain within that conventional project: social science would be the "calculating" part, leaving only the manager's "gut instinct" to serve as a ground for going off the planned course.

But when Garner starts to describe concrete particular moves that he made during the course of the game, this absolute opposition between instinct and calculation vanishes in favor of something else: Garner's specific knowledge of what his team can and cannot do.

We have interchangeable parts, moreso, I think, than most of the teams in the playoffs. We have a guy -- Bruntlett -- that can play every position, so if I take any one guy out, and I want to maneuver around the lineup in some way, I have him that I can just keep moving to any position on the field. And he'll play a great defense. And he's actually probably saved three games for us because he's made phenomenal defensive plays. Not to mention the couple he's won with his bat for us in the course of the season.

Here we see a subtle combination of calculation and instinct at work, as Garner is able to derive possibilities from his experience with particular players on the team and to formulate different strategies as the game unfolds and the situation changes. The relevant basis here is not rule-based knowledge of which move belongs where, but the kind of practical wisdom that allows the expert manager to grasp conditions of possibility and then work to actualize them. And doing this doesn't mean that one abandons calculation or advance preparation, as Garner's discussion of one of his other moves makes clear:

The other day in our 18-inning game, [Brad] Asmus goes out from behind the plate because I wanted to keep both catchers in because, in case we get into a drawn-out affair and I had to bring Clemens in I want Asmus to catch Clemens, so it predicated a move to keep him in the ballgame at first base and put Chavez in and then flip-flop the two when Clemens came in. But that's not a difficult decision. I'm confident with both of them playing out there because both of them do a good job, and they practice it all the time: they take ground balls, all year long. And I even put Brad at shortstop one game, and first base -- or second base one game I believe it was -- in preparation for something just like this. So they're mostly all calculated.

"Calculated" has subtly shifted its meaning by the end of this example, and now means "based on experience" rather than "in accord with an abstract specification of rules." Because Asmus and Chavez have practiced taking ground balls at first base, Garner knows that he can put either one of them there and expect that they'll do a good job, but there's no necessary line from this fact about the two of them to the observed outcome. Instead, something else intervenes: Garner's experience-based sense of what he wants to accomplish and how he might best accomplish it, given the resources available to him.

Could this decision-process be automated, and replicated by a machine -- or by an abstract decision-making procedure, in accord with the conventional Social Science Project? It could probably be simulated, certainly. And we could probably build a baseball-managing machine that would apply formal rules to situations and arrive at outcomes similar to those that Garner arrived at. But I'm not at all certain that doing so would prove that Garner was really making rational calculations all the time that he was on the field. Just because I can retroactively narrate some situation in particular terms doesn't prove that the situation was really that way; it just demonstrates that we can describe that situation in certain ways. Period. We could describe Garner's decisions as entirely calculated, as he seems wont to do himself, but in so doing we'd miss a lot of the way that those decisions are being produced -- and the subtle ways that Garner keeps redefining "calculated" as he discusses situations.

This is nowhere clearer than in his conclusion, in which he explicitly denies having put Chris Burke in the game in order to hit the game-winning home run that he did hit in the bottom of the 18th inning:

I would say that there are some things that I feel good about, some things that I don't feel good about but I feel like I have to do -- I didn't want to take Berkmann out of the game the other night, but figure this one out for me: Everybody in the world would say, why would you take Berkmann out 'cause he's the guy that's gonna win the ballgame for you, and the guy I replace him with is the guy that hits a home run to win the ballgame! Now I didn't figure that was gonna happen. But I figured, based on Atlanta's outfield, if we get a base hit there, they're aggressive, they all have good arms and they're accurate, and if we don't take a chance to score I'm not gonna like myself very much the next day, so you gotta do things, in my opinion, that way. So most of it's calculated.

Note the sudden emergence of a third ground on which to place decisions: a logic of identity, based neither on instinct nor on calculation. "Because I am X, or because I want to be X -- because I think of myself as X and want to continue doing so tomorrow -- I need to make certain moves, take certain chances, go in certain directions." Here we have a kind of active self-crafting, in which Garner basically produces his identity and his team's identity by going in one direction rather than another. The removal of Berkmann from the game flies in the face of rational calculation, but it isn't an instinctual move -- it is instead based on Garner's sense of who he is and who he wants to be. "I'm not gonna like myself very much the next day" isn't a preference held by a fully-formed rational actor; it's what John Shotter would call "knowing from within," sensing the potentials inherent in a situation and acting on those potentials in such a way as to craft a certain sense of oneself. I get the feeling that if the Astros had lost, Garner would be defending the decision in very similar terms: it may not have worked out quite right but at least we gave it a shot, played our kind of baseball, kept our integrity intact.

So: three grounds for making decisions, and no simple formula for integrating the three in specific situations. In the gap between these different logics we have agency, understood as the capacity of the manager to have done otherwise than he in fact did. We have responsibility, because Garner can't deflect what he did onto any more solid or objective grounds than his own actions. And we have the subtle combination of art and science that makes baseball such fun to watch.

In this case, what is true of baseball is most likely true of other areas of human social action too. Rules, instincts, and identities concatenate in unique ways in every instance to produce particular decisions; reducing a decision to any one of these three deprives the actor in question of effective agency, as well as making it impossible to meaningfully attribute responsibility to her. If one is going to analyze decisions -- which is not something that I usually do in my work -- or if one is going to make decisions -- which is something that I like everyone else do all the time -- one should keep Garner's comments in mind, and work to develop all three of these capacities.