Tuesday, January 1, 2008

Can GP2.8 be satisfied without process metrics?

I see organizations usually implement GP 2.8 with product and process metrics. But, now I am working in an organization that says they have talked with lead appraisers who told them that it is possible to implement GP 2.8 without any metrics, just systematically review the process execution, generate a qualitative report, identify issues, take corrective actions and follow them upon closure without mention any metric. I think this is not the intention of the practice. Am I wrong?

I love questions that are "hot topics" anytime a bunch a Lead Appraiser get together - and this is one of them! Yes, unfortunately we are geeks and we argue a lot about process. There are those LA's who would agree with you and say that the intention of the practice is to use a metric. I know one at the SEI in particular who insists that this is required. But, there are those who say, no, that isn't always the case. I say "it depends." So, how to decide?

The Generic Practices are frustrating because little or no guidance is given (compared to SP's) as to what is really expected. But the authors do provide the answer, albeit indirectly. So I cracked open my “signed by the author” copy of the book and checked it out for myself. In order to understand the GPs, we need to consider the reason why the GP is there in the first place, and then map back to the PA from which it came. In the case of GP2.8, it clearly maps to PMC, not to MA, because we want to know "how is it going?"

Can the Specific Goals in PMC be satisfied without a metric? While metrics are often presented as part of the objective evidence for PMC during an appraisal, there are many parts of PMC that do not require metrics to be successful. As a matter of fact, the SP’s don’t speak to metrics at all! However, there is another clue – and it’s right in your original question (great questions often have the answer in them don’t they?).

In your question you say the client has told you that they "systematically review the process execution, generate a qualitative report, identify issues, take corrective actions and follow them upon closure." That reads a LOT like the SPs in PMC and all of those could certainly be performed without a metric couldn't they? As a matter of fact, the SG’s in PMC could be satisfied without a formal metric (unless you consider schedule and budget part of a metric . . . which it kind-of sort-of is). Now, as to whether they actually DID all of this or not is another question, and a metric about this would be great evidence that this actually occurred. Darn! That CMMI is so circular!

So, using Socratic reasoning as my guide, GP2.8 is derived from PMC, and PMC is largely about “systematically review the . . . “ and not about “metrics” THEREFORE, it’s logical to assume that yes, it is possible for GP2.8 to be satisfied without a metric ASSUMING they are really doing all these things they claim to be doing (which may require a metric show!).

Question answered? Not so fast!

Now, let’s fast forward to ML4/5 where we all start wishing we attached a metric to GP2.8 back when we were ML2. Why? Because that data gathered from GP2.8 happens to be the very same data we desperately need in order to perform OPP as we develop models and baselines in order to begin QPM.

This is where the real power of the CMMI becomes evident and it is a shame that many organizations get to this point and are stopped dead in their tracks because they have no historical data about the performance of the process.

So there you go. The analysis tells us we don’t HAVE to have a metric, but that if we don’t, we’ll be sorry. Go figure.

Thanks for the fun question.


No comments: