So I read Stewart Brand's "How Buildings Learn" with great curiosity:
It's devoted to precisely this subject: how buildings evolve from their original form to something new, informed by actual daily use, which is often subtly or significantly at odds with the builder's or architect's original intent, and how buildings continue to adapt as they are summoned to all sorts of new purposes for which they were never designed. (A house becomes a store. A warehouse becomes apartments.) It's a manifesto of sorts, calling for the building trades—and architecture in particular—to accept and embrace, not fight, the fact that buildings change over time and arguing that part of the job of creating a new building is to ensure that it will adapt gracefully to unforeseen demands in the unknowable future (something which many high-concept high-budget constructions do rather poorly but which many time-worn building vernaculars do quite well). It's also a highly idiosyncratic and opinionated book, full of quirky examples and curmudgeonly advice. He probably gets some things wrong—Brand himself acknowledges that the it's largely a collection of hypotheses that should be verified or rejected by empirical study—but I suspect he gets the main thrust right.
Having nearly finished it, I now look at old buildings differently. I'm more aware of what I'll call their "strata", the modifications that have accreted over time to result in the present-day structure. (New Orleans offers particularly choice opportunities for this because of all the twin-houses—or triplet-, quadruplet-, etc.—that started life identically but diverged until the original shared lineage is scarcely recognizable.*) It emphasizes that buildings (and the neighborhoods and cities they compose) are never merely static constructs but are, of their essence, dynamic entities moving through and changing in time.
* In fact, the cover of the book shows precisely such an example, a before-and-after of a pair of buildings on lower St. Charles Ave.