Has anyone else noticed how nervous developers get when more testers are hired? Or how nasty the media gets about buggy software after testers are laid off? I think there's a pattern... Testing Is Important.
Nothing has diminished the effectiveness of CMM. CMM was developed as a way to improve your software process and to evaluate yourself on how well you're doing. It was never intended as a marketing tool. As mentioned in the article, "We weren't chartered to be policemenówe're a research and development group". The same thing would apply to any certification that isn't conducted in an almost cheat-proof environment such as that of SATs or MS certifications. Its sad that a lot of great honest companies get weeded out based upon being level 3 or 4. 3 is still a really good rating and I would consider such an organization over a supposed 5 if they had a proven history of customer satisfaction. You wouldn't buy a car based upon its spec sheet. Caveat emptor.
Were you able to see any INDEPENDENT confirmation of a CMM efficiency?
4 years ago I put a lot of efforts and failed to find such data.
Only people, who were "selling" CMM, provided "data" that supposedly proved CMM usefullness.
People, not affiliated with CMM (e.g. Capers Jones), found that some other simple methodologies were more efficient than CMM.
I wouldn't question any methodology over another and an organization could implement as little of the CMM as needed or even twist it way out of shape to suit them. The absolute wrong way to do CMM is to attempt to implement too much too fast. Building up a CMM level 5 software shop to design a simple e-commerce portal is probably a huge waste of money and highly inefficient because you don't need *that* much assurance of quality. However, I would expect someone building guidance systems for Tomahawk cruise missiles to have such a rating or equivalent. I wouldn't call CMM or any related process 'efficient'. You have a LOT of time doing documentation that your people could be spending churning out code. The key here is predictable and repeatable. Your customers are buying your service because they are reassured that you'll deliver a quality product on-time and on-budget. Twice I've seen CMM fail and this was due to lack of training and not much support from upper management. The execs told the mid-levels to do it and then went about their business. The mid-levels got together and came up with a plan and told their people what to do. The problem was that all the workers (dev/test/doc) didn't see the big picture due to lack of training and there was no intra-office coordination. The place I have seen it work was in the Air Force. As a tester, I was required to attend *all* of the CMM courses including code estimation, project planning, and many others I didn't really have a direct hand in. However, learning what everyone else in the organization was doing was a great help. It VERY much improved the level of teamwork. I can't advocate CMM over any other process because I've always worked in CMM shops. I just don't think you can blame a process for the marketing problems in the original post or an organization's failure to succeed using that process. Careful planning has to be made for what your organization needs. [uggg, ok. this is too long] .. To sum up, trying to force your software house into a CMM rating is probably going to cause huge problems and failure.