Friday, August 20, 2010
In my experience with consulting and development, project teams (both Agile and traditional) invariably face issues with availability of team members. In one (rather long and messy) talk on craftsmanship Robert Martin came up with an interesting idea that I'd like to explore a bit more. It might actually solve this problem if used in the right way in a large organization. The idea is simple: organize into long term teams of upto 12 people and accept work from different projects as a team.
This is actually the way small development companies have worked for ages, and they have done quite well on it. The problem is that when projects and clients get bigger they tend to turn things upside down and instead of pushing work to teams that can handle it, they start pulling resources (eww I hate that word when applied to human beings) into the project. This process is called staffing.
There is a problem with staffing. People are not resources. The difference between people and resources (think coal, plated steel, money) is that they change based on changes in their environment. This ability is defined by Stephen Hawking as intelligence, and I like that definition. The big upside of intelligence is that it can actually solve problems for you and that is something that typical resources cannot do. If intelligence is the stuff that allows you to come up with solutions, resources are the stuff that you consume to implement them. A profound difference here is that coming up with a solution actually increases the amount of intelligence, whereas implementing a solution always decreases the amount of resources.
If managers pretend that people are resources and projects are black boxes that turn resources into implemented solutions they will get two things: shortage of intelligence and shortcomings in implemented solutions. The shortage of intelligence is caused by the fact that a project starts being about finding a solution and invariably more and more becomes about implementing it. The former makes you smarter (increasing intelligence) the latter dumbs you down (decreasing intelligence). If you only implement solutions you lose your ability to adapt to change, but if all you ever do is adapt to change you might go crazy. There is an optimum between the amount of change a person has to handle and the amount of time he needs to regain his footing, and this is different per person.
This is all well and fine, but how will we get those essential team members available then? I'm getting to it, don't worry.
First we put a team together thinking about a certain context, not a project. Then we let them establish a healthy environment for them to adapt to and increase their intelligence. This causes team members to be available to each other by default. The we let the project managers sell the work to a team, possibly multiple teams. Each team takes on what they are comfortable with but they stick together. Now this chunk of work might be too little or they might get bored with it. Then they take on some more work, possibly from another project, but they keep together as a team.
If a team member wants to leave this is a career move, not a ephemeral thing you can do for a couple of hours on friday if you feel like it. You can't be part time on one team and part time on another. If you want to take on some work you have to convince your team, or make your career move and be gone totally. Now the question of when person X is available to person Y is extremely simple. When they're in the same team they can count on each other. If they are not on the same team they might have a meeting some time.
This should up a lot of things, I'm looking forward to try it. If anyone already has experience with this approach on a large scale or moving a large development organization into this mode I would love to hear some war stories.