CMM SEI and how to organize discussion

On the Space Generation Talk mailing list, Jose Aban and Jim Volp raise a perennial question:

When you have a whole lot of people who have a whole lot to talk about over periods of years, how do organize that?

Jose sounds more like a top-down design kind of guy on that thread, Jim a little more bottom-up. I actually sympathize with both positions. If you've got a good organizational template, why not reuse it? But as Jim Volp points out, imposing organizational templates can be like doing a skin graft -- there's the possibility of rejection.

I wish I had the answer. But maybe I have a clue.

Back in the days when people talked about the software crisis as if it might be solved someday, software methodologists saw the first glimmerings of the idea for an organization metric eventually called the Capability Maturity Model (CMM). As formulated by the Software Engineering Institute (SEI) associated with Carnegie Mellon University, the CMM had 5 levels:

1. Initial
2. Repeatable
3. Defined
4. Managed
5. Optimizing

Some have proposed a Level 0: Incomplete/Negligent. And, not entirely tongue in cheek, three negative levels: Obstructive, Contemptuous, and worst of all, Undermining.

How do you get to a target level? Unfortunately for companies that would like to be graded at Level 4 or 5, despite little attention to process, it's like every other learning curve in this world. You have to crawl, then walk, then run and then fly before you can reach space, much less orbit. In short, what they learned was this:

You have to start at Level 1

You can't skip steps.

I'd like to make two observations about this.

(1) An organization that has reached a high organizational process level may not be very prepared to host new activities, because there's so often a focus on how to fit any new work into the process. (Correllary: the template from a highly organized process will likely result in skin-graft rejection when applied to a less-developed organization.)

(2) Even in high-process-level organizations, there is seldom a process applied organization-wide for estimating what level a new activity should start at, and making sure it doesn't sink after it's started. (Remember, even from Level 1, you can sink. Who hasn't seen it?) Some projects don't need to start at Level 1 because they are hived off smoothly from a Level 2 or higher project. Perhaps most do, however.

Some large aerospace companies have worked around the first problem, by inaugurating a skunkworks for new, risky, innovative projects. Often, they were able to justify departures from mainstream corporate practices because of the ultrasecret nature of the project -- the extra secrecy was itself an added process burden, they might have argued, so why add "business as usual" as a drag force? They didn't expose their fresh, young, energetic talent to the corporation's whole mature development process. To some extent, they actually shielded them from it.

As for the second problem, it's quite common to have "intrapreneurs" in large organizations, well-connected innovators (or at least, innovation-sniffers) near the top, marshalling resources informally when they see a good idea. However, when you lose them, you lose the process that's captive in the brains. That loss is especially damaging when the process is owned by two top executives and you lose them both in the same year, as happened to the instrument company Tektronix at one point.

The good news here: I think you can form renewable organs even within a Level 1 organization (which themselves have to start from Level 1 and hang there for a while) that perform two functions:

(1) Assessing the level at which a proposed new activity might best start
(2) Pushing activities to the highest maturity levels they need to reach in their project lifetimes in order to be effective.

These organs within the organization need not have any formal authority. It would probably be enough for them simply to operate, evaluate, and educate. I would make some suggestions about how they ought to be organized, but doing so would imply an arrogant assumption that I know what they should look like at Level 2 ("Repeatable"), when I don't even know what they should look like at Level 1 ("Initial").