Management by Baseball

The Intro to this series of posts exposes the generally-unknown fact that Major League manager One of the challenges of Agile product design & development is project management. The foundation of Agile is the drive to diminish overhead, and planning IS overhead. As I pointed out in the Introductory post to this series, the Agile MANifesto says

Responding to change over following a plan

But Agile and Lean mean moving quickly. And moving quickly without management is instrinsically risky...unless you do it it as they do in Baseball.

Tony La Russa and his mentors invented a form of project management that Agile Development has cloned almost exactly for its own purposes. As La Russa describes in his recent book, One Last Strike, the way to project manage in an Agile way is far more intensive and demanding and requires more skill than traditional plan-every-detail project management does.

Instead of front-loading all the individual details, the La Russa methods involve front-loading basic rules of action and working out every contingency and its tendencies in advance. That way, when you are in the moment of decision-making, you don't have to think through possibilities from scratch. It's very much like modular architecture (think Alexander's A Pattern Language: Towns, Buildings, Construction) or modular software development. Each game (project) has patterns you know in advance, probabilities that evolve rapidly during the execution. Each pattern consists of coherent clusters of interrelated pieces, like a set of 16 Lego blocks you would use to make a doorway-and-arch or a building corner. Each decision you make will be based not on perfection but on optimal utility in that moment, and that might involve riffing off one of the components that make up that pattern, to increase its "performance" probability in the specific instance of its application.

Figuring out which components, what might need to vary for contexts, and contingencies is a lot of hard work, fun for some of us, but challenging nevertheless. You are simplifying the work (by using patterns/clusters) in order to save energy to invest in the contingency efforts.

La Russa describes (page 206-7) a classic example of a super-critical project management planning/contingency-design cycle he went through in 2011, his St. Louis Cardinals' last World Series championship year. and his last season. The set-up is this: The Cards had to win the last game of the season to earn a spot in the playoffs. Weeks in advance of that game #162, he had arranged his pitching staff so that if that was a win-or-go-home moment, which it turned out to be, the team would have Chris Carpenter, easily the teams best pitcher, starting that game. That way, if the game was irrelevant to the playoffs, he would rest "Carp" & use him in the playoff opener; if the Cards were out of contention, he could start Carp or not.

He used Carpenter in the last game and that closed a few (Carp could not start the first playoff game as they would have had him do, ideally) and simultaneously opened a ton of possibilities/patterns he could use. They won their game, but two other games going on after theirs could affect the outcome by changing the standings. Look at his relentless (Agile) project management thinking.

My thinking about the rotation for the Phillies series began in the 45 minutes following our victory over Houston in game 162. While the guys were agonizing over those final innings bteween Atlanta and Philly, I was in the officegoing over stats and messing around with some ideas about how to get the most Carp for our buck. I knew one thing without a doubt: we had to have Carp pitch in two of those five games. Had to.

I came to that conclusion immediately, even when I was still in Houston before the champagne was uncorked. While we hadn't fully committed to having him start on three days' rest (note: less rest than normal, undermining of peak performance) I know that I first considered it back then. After we were sure we were in and headed back home on the plane, I looked at the rotation again. What we had found over the years about the best-of-five Division Series was that pitching your two best pitchers twice gives you the edge. Ideally, you'd like your number one go in games 1 and 4 and your number two in games 2 and 5. (note: ideal because then the number one gets an extra day of rest between series if you win, so can start earlier against the next opponent). We couldn't do that. Carp obviously wouldn't start the first game on Saturday because he had just pitched on Wednesday. That meant the soonest he could go was the second game on Sunday {snip} the fifth and deciding game. (note: not ideal, but a good fallback to have your best pitcher in a deciding game).

La Russa accepts this as his base decision. But he's not done. As with all Agile projects, there are people involved, and team members need both to buy in and deliver.

...we were on our way back to St. Louis. I was still sitting there with a pad and pencil taking notes about the possible rotation for the upcoming series. I got up and walked back to where Carp was. I held out the paper and tapped the space on the page where I'd penciled in his name.

Carp smiled and nodded. "I'm good to go."

"Not yet. We'll wait and see how you feel Friday (note: the second day after a start and the day of testing soreness/recovery)"

I walked back to my seat. Of course he'd want the ball. I already knew that. I just wanted to let him know that we were going to wait to see how he felt before fully committing.

That was a base plan for games 2 and 5, but what about the rest? It's somewhat simpler because La Russa is now building around a draft decision-made that makes for some givens.

His decision algebra is detailed in the book, but again, he's starting with known clusters, delivering small certainties, and then building draft plans for each decision.

Each of the decisions for this high-impact, multi-million dollar Agile project is something he crafts carefull but holds loosely. On game day while filling out the lineup card, he's not going to be krazy-glued to his initial draft; if he needs to riff to match evolving contexts, he will, using other patterns or, in La Russa's or other top-notch managers' cases, inventing soemthing on the fly.

La Russa is a pretty extraordinary practitioner of Agile project management, evevn for someone from Baseball. But as I've said before, the 25th-percentile Baseball manager can manage and lead rings around 85% of billion-dollar company CEOs.