I think programmers need to solve very similar problems to chip designers, in that what you can fit in the local cache (ie the active bits of the brain) is limited, so you want layers of larger caches to minimise the time required to get each bit of information. The next layer is those windows physically open in front of you, the one after that is the windows just a keystroke ("click" for Windows/Mac users) away. And I guess the one after that is the shell that lets you specify and open up a file that doesn't have an edit session open yet.

The layer of "windows open in front of me" is the one with quickest access time, and the bigger I can make that the better.

Hugo