What passes in some software projects for "analysis".

PC here is the root axiom of PrematureGeneralization, PrematureMarshalling, PrematureOptimization, PrematureConcurrency, PrematureSerialization, PrematureRobustness, and all the other ways to brag about wasting time before actually getting anything done.

The cure is to design in increments. See NotesOnTheSynthesisOfForm.

-----

''I smell a pro-XP bias above. For a counter-point, see XpIsForBadPlanners.''

I smell industry experience suffering a CombinatorialExplosion of bugs caused by colleagues performing each of the above as a BestPractice. If that has lead me to ExtremeProgramming, even as a stick to beat them with, so be it.

Until your defect rate drops below, say, one bug per month escaping the lab, or 3 hours flight time in the debugger per month, please don't try to debate which practices are best here ;-)
-------
See Also: YagNi
----
CategoryComplexity