This should strike as a suspicious idea, one defeated multiple times in history: just think of the history of mathematics, and the representation of numbers.

Furthermore, this idea (that one might define processes first, and select tools to implement these next) is itself an idea borne in a certain historical context: this of a limited complexity, be it this of the domain to manage, or of the tools available to manage it —both being software. In this context, the tools were only ancillary, assisting to perform tasks which used to be undertaken previously without their help. The tools themselves were simple enough not to need any management: they were supposed to be transparent. This is what I have proposed elsewhere to designate as CM.

SCM proper concerns a context radically more complex, where (as stated in Understanding Computers and Cognition, by Winograd & Flores) there is often no way to predict how a program will act, short of running it. This is where techniques such as testing and auditing (recording) become relevant and necessary.

-- MarcGirod - 23 Dec 2006