Why does CMMI come in two flavors: staged vs. continuous representation?
There is a religious war in the SPI community on this question, much like Windows vs. Linux. A staged improvement model a la CMM with its levels 1 – 5 gives clear, unambiguous guidance for what to concentrate on next in an SPI program. Critics say, that this is very inflexible, and applying the same approach to a 10 person software house writing embedded software and to a 1000 person organization developing banking software cannot work. They say, the thing to do is to identify the weaknesses of the organization, prioritize them and work on the ones with the highest impact. While this is indeed a good and strong argument, practice has shown that this approach is very difficult to implement for many organizations. There are other points, which are discussed at length and heatedly whenever two or more SPI experts meet. SEI has taken the right decision and provides both versions of its new model, to a large extent with the same contents, and so everyone can choose which one to use.
