In my experience, the trap big projects frequently fall into is a desire/expectation that a comprehensive specification has to be completed before any part of the project can proceed.  In a large project, this usually leads to endless cycles of project specification meetings which produce lots of paper (not environmentally sound) and little product.  I prefer to establish a loose framework and a set of standards for the overall project, and then focus on smaller parts of the project for specification/coding, and even then I will admit to being light on the specification side, preferring a RAD (prototype/review/modify/implement) approach.  Projects that have all the deliverables at the end tend to never make it to the end, and in an open source effort, I fear that without some steady flow of useful development, interest in the project would gradually fade away.
Mark

Reply via email to