Which is like, never
On of the big structured proponenets - DeMarco or Humphreys I think, came clean and admitted that all of his earlier writings on structured analysis and software process were hopelessly naive because he failed to take into consideration the idea that design decisions are often political and not technical. IOW, fuzzy specs often arise out of unresolved disagreements between parties and simply sitting down with pen and paper isn't going to get those issues cleared up.
IOW, in typical engineer fashion he completely failed to grasp the political/social implications of collaborative construction of software.
Iteration, OTOH seems to provide the negotiating basis for reaching consensus.
