Yeah I don't agree with "bulldozer code" bit. Splitting a long function into smaller parts doesn't have to be about reuse. It's useful to do it for clarity. As a single function grows, the number of local variable often grows as well and eventually you end up juggling tons of variables and the slightest change ripples three hundred lines further down. Break it down into a handful of steps with a number of arguments you can count on your fingers and it suddenly becomes so simple. Same about deep nesting of loops/conditions.

Rule of thumb is a function needs to fit in my head - so if I stick my face on the screen and the function's code can still be partly seen, it's time to break it down.