CMM is obsolete. It's now officially replaced with CMMI, Capability Maturity Model Integrated. SEI is Carnegie Mellon's Software Engineering Institute.
"CMM Level 1 - Every company is already CMM 1, because there's nothing lower that you can be."
They haven't seen a couple of the places I've worked, obviously. One of the hallmarks of CMMI Level 1 is that it takes (per CM's own documents) individual heroics to make the program work. You know the type -- a Dilbert character who has worked several days in a row without sleep to meet a deadline.
Each successive step in the CMMI chain means that the organization is supposedly more and more capable of planning and executing software programs. One of the BIG benefits of CMMI is that it creates an institutional memory.
Anyone ever hear of a business or program that folds because one or two individuals leave and take institutional knowledge with them that can't be replaced? CMMI is a guard against that.
A lot of people poo poo CMMI, ISO and the other, smaller, process programs, mainly because they all include increasingly heavy burdens of document and record keeping.
Contrary to Broken Paw's assessment, CMMI's predecessors were developed at the behest of the Federal government, NOT because someone at Carnegie-Melon decided that the SEI would be a neat way of sucking in a lot more money, and NOT because ISO wasn't paying them.
At one time in the 1960s and 1970s upwards 70% of all money spent on software development programs was spent on failed programs because there were simply no standards, no institutional memory, no repeatability. Everything was heroics from front to back, and sometimes heros fail. The Feds wanted processes that would help prevent massive, expensive, failures.
The picture is MUCH different today. Even an organization that has CMMI lapses, such as the one I'm facing, is far more capable of executing a software program.
The data is there to support the overall efficacy of the CMMI process, but as with anything, the process is only as good as the people who are participating in it, and that's what I faced up against last night.
Yes, there are still some pretty spectacular program failures. Trilogy and Trailblazer come to mind. In part, those programs failed because of CMMI lapses on the part of the contractor, but they also failed because the government was unable or unwilling to provide clear, comprehensive, and consistent direction as to the goals of the program. Trilogy is a spectacular example of that. It was kind of funny, actually. The FBI gets up in front of congress and savages the main contractor (my employer) for failure to perform, and a short time later GAO issues a report that placed probably 90% of the blame not on my employer, but on FBI.
That said, program failures are now the exception to the rule, not the rule, and process definition, adherence, and improvement is largely the reason. It's also a major reason why the Federal government, more and more, is demanding, as part of its RFP process, that any organization that bids on Fed work be CMMI or other process method certified. No certification, no contract.
The refreshing part is that in an organization that isn't CMMI adherent, something like what happened last night could cause a cascade failure of part, or the entire, software delivery that we're working on now. I've seen it happen, and it's never pretty.
