Everybody talks about how the New England Patriots Super Bowl win last year was a team effort. Whether it was the backup quarterback imitating Peyton Manning on the scout team, the statisticians, the players, the coach, or the personnel guy, everybody contributed.

Teamwork, of course, is one the perennial topics du jour in the software world. Demarco and Lister’s classic Peopleware talks about it, introducing the concept of a “jelled” team. All the variants of Extreme Programming rave about it.

But what makes a team good? It is difficult to have a good discussion about software teams because there are not enough public concrete case studies. Could sports teams provide a basis for such a discussion?

Peter Drucker, the management expert, thought so. In a Wall Street Journal essay written several years ago, he discussed three paradigmatic teams: baseball, football and tennis.

What type of team does your organization have? As Drucker makes clear in his essay these teams are distinct alternatives. They have unique strengths and weaknesses, but attempts to combine parts of each are a recipe for disaster.

I bet a lot of traditional software organizations have baseball style teams. You do not play as baseball team, you are a member of the team. The third baseman never pitches, the tester does not do development. Designers do not have much interaction with developers. This is the old-style Detroit assembly line. The big advantage to this team is that it makes it easy to train and evaluate personnel. Everybody can be a "star" no matter how difficult they are to get along with. On many plays, certain players are not important. The left fielder does not do much on a routine ground ball to the second baseman. Symphony orchestras are like this as well.

This approach works well when the task is well understood and can be reduced to "routine". Here one can understand the drive to outsource and offshore software tasks. If things are well defined, and competence exists elsewhere, then price drives all.

The problem comes when you need to innovate quickly.

Football teams are more flexible. There is no equivalent to the halfback option pass in baseball. On almost every play, every player is necessary, if for no other reason then to prevent some other player from getting to the ball carrier, or the quarterback. Everybody works in parallel. Unlike baseball, everybody has to follow the coach's orders or else the team will not win. You have to train together to be effective.

How do you evaluate and train people? An individual's value is often related to how they complement the rest of the team, not only on their individual strengths and weaknesses. Play without a linebacker or a safety, and you will fail just as if you played without a quarterback. But why do quarterbacks get more money? Why do they both get more money than teachers? This is the old diamond-water paradox in economics. Here you have to reward people based on their marginal value to your team, not on some absolute scale. It is much harder to outsource, much less offshore if you live in this world.

You still need, as Drucker points out, a score to evaluate how well the team is doing. Though as any football fan will tell you, the score does not relate well to how an individual player is doing. That is why you have to "watch film" to evaluate players, something baseball sabermetricians ( http://www-math.bgsu.edu/~albert/papers/saber.html ) do not have to do. Software managers in this world really have to understand what is going on in order to evaluate and train their people.

Finally there are tennis doubles teams (or jazz combos). There are no clear players, only roles to be filled by different team members at different times. This is the ultimate in flexibility and adoptability to changing circumstances. But there really has to be a fit here. How do you train and compensate in such a world where you can succeed or fail, but there is no score. It is hard to relate the end result to the individual. You certainly cannot outsource or offshore here.

As Drucker says, teams are tools, and you have to understand your environment and pick the appropriate approach. This is what management is all about.