Getting back to the subject of scientific method and its possible application to programming, readers may be interested in the observation of Thorngate, who offers that a theory/model cannot simultaneously be general, accurate and simple. It's like a one-armed clock: it can point to any of these three or in between them, but no more.

I see this a applicable to the classification of a requirement: the collection of them cannot simultaneously be general, accurate, and simple.

Does this help?

Thorngate, W. (1976). "In general" vs. "It depends": some comments of the Gergen-Schlenker debate. Personality & Social Psychology Bulletin, 2, 404-410.