Procrastination for fun and profit.

We start by noting that in any given project, there are some number of errors (possibly zero). Then for any given error, P(error has been found and eliminated before time t) is bounded above and is increasing in t.

So P converges to a limit K as t increases. Now K depends on many things about the specific project, but it does not depend on time t. It follows that procrastination for any finite length of time before beginning work on error checking (bugfixing, proofreading) has no effect on the final quality of the work!

There is no fallacy: the result is indeed true provided the project has an unbounded lifespan. Alas, most projects have small finite lifetimes instead. And for some reason I now feel like writing a short book called "Real Analysis: A Picture Book For Children".