I am ready to coin a new rule, and I like to call it the 10% rule. This rule says that for any problem you solve, you should spend an additional 10% of the time you spent finding the solution to improve your solution. This can be applied to most things that come about from thinking, including something you write, proving a math theorem, or designing a house (since we all do all of those things). I decided on the rule based on my work in developing software. In software development, there are usually numerous ways to solve a problem, and it stands to reason that the first solution you come up with probably won't be the best. However, once you do know a solution, you have a better perspective on the problem than when you started. So while it probably isn't worth it to go back and overhaul your solution, it probably will benefit you to find areas where you could make small improvements and performance tweaks.
I came up with this after recognizing my own methods of developing. I don't spend a ton of time going back over my work once it works, but it is worthwhile to spend about 10% of my time trying to improve it. It is like in anything else, including brewing beer, packing a car, or mowing a lawn... once you know what you are doing, you can stand to do it better and you know how to do that quickly. (If the changes aren't quick, your returns start to diminish, hence stopping at 10%.)