back
2 comments
I focused mostly on the "Feet of clay: The CMM's fundamental misunderstanding of level 1 Organizations" section and I find a lot to disagree with.

The biggest issue, which I don't sense the author gets, is "repeatability". Level One companies are entirely dependent on "heroes" (or cowboys, take your pick :-) and their ability to follow their first success is not particularly predictable, except that it's likely to be poor. Look at Lotus, who's first product (the 1-2-3 spreadsheet etc.) was written by a team of 7 or so assembly language wizards for the first IBM-PC and that lost it's ability to develop software after that (1-2-3's mid-life kicker that saved the company (for a while) was an unauthorized project written by two programmer who left the company before the need for it was acknowledged).

I think we realize this less now because we're so much more leveraged and "powerful" today; a small team can do so much more and continue to do that if the company avoids Teamicide (http://www.amazon.com/Peopleware-Productive-Projects-Teams-S... , READ IT!!!) and doesn't have to grow its team (much).

Microsoft, on the other hand, by the 1994 date of this essay was most certainly not a Level One company. I've long said the secret to their success in the period that includes that year was that they retained the ability to write software that basically worked (compare this to all their Windows Office competitors who lost in the transition to Windows 3.x: with the exception of WordPerfect, the others couldn't get past the "crashes too often" stage, and WordPerfect's release was a sad joke. Sure, it didn't crash, but it also would frequently move all your figures and tables to the bottom of your document.)

It's No Accident that Microsoft was run during that period by a seriously good hacker (anyone who can successfully rewrite an assembly BASIC interpreter (without benefit of a computer) during a transcontinental flight earns my respect.)

I too prefer Gerald Wienberg's approach to this, and CMM can obviously be fetishized and at best promises little more than "we can continue write mediocre software", but there's something there that's worthwhile that it's trying to achieve.

That might be true, but this is funnier ....

  CIMM  http://en.wikipedia.org/wiki/Capability_Immaturity_Model