Before committing to a rewrite

A rewrite is attractive partly because I know every irritation in the current system and none of the ones in its replacement. That's a fairly uneven comparison.

Before proposing one, I'd try replacing a small part behind an existing boundary. Find out what is difficult to move and what behaviour people depend on. If a larger rewrite is still justified, that work gives us a better estimate and a route to release it in pieces. I want an exit plan for the old code too.