Apart from the excellent advice jhancock and others gave (pick something small and finish it), I wanted to say that reading your post I realized there is anothe

Apart from the excellent advice jhancock and others gave (pick something small and finish it), I wanted to say that reading your post I realized there is another practical solution to the same problem: in the work I do, finishing doesn't make sense. I write custom software for a few clients. They're developed continuously, without any hint of them ever being done. The catch is that since very very early they're "working". "Working" is a concept much more useful to me then "finished". Working means it does useful stuff for the people using it. Will there be more development? Sure. Does it matter? Nope. The only such project I can consider finished is one in which the client company was bought by a bigger company, which came with its own software solution. All development stopped (eventually), but the software is still up and running, used occasionally for its database.