I've had problems like that. I will start a map, then spend a year adding finishing toutches. I think that if I'm working on a map that I want to be good, it is hard to let it go. I think the hardest part is when to say enough is enough and release it.
To be honest, this is due to lousy planning. If you don't think anything out it becomes very obscure and almost impossible to create anything decent.
Even having a simple list will help, but if you want to look at it more closely:
There's an apparent cycle in almost all hobbies that have this type of process, or any development for that matter; what's important in EVERY development is working the idea to death in the earliest stage to limit the actual creation time (and to make your ideas concrete). Granted, this is a "professional" point of view, but it should help in almost guaranteeing completion of a decent map(novel/screenplay/program/etc).
I'd suggest looking at the development cycle for software development (taking that it's the closest relating cycle I can think of, it's more adaptable to mapping then anything else) I don't suggest looking at it from an architectural standpoint because that process is ridiculously long and it honestly isn't needed when doing this. However, you'd be able to create something amazing doing it this way.
The cycle is fairly simple:
Requirements
Specifications
Coding
Testing
Maintenance
That for mapping can be translated to:
Requirements (Outline)
Specifications (Layout/Design)
Map
Test
Maintenance
There are certain versions of the cycle in which the mapping and testing are inter-mingled until there's final testing, which you would call beta-testing (after doing an initial alpha on your own, of course)
More Clarification (stop reading if you don't care):
Create an outline of the rough ideas you have for your map, in words. Like themes you want throughout the map (i.e. rounded corners, space station, rusty metal, etc.) Some other outlines you'll want to make is layout (connectivity) and game-play style (i.e. vertical) and then think of situations that it would play well in. Another thing you may want to start planning here are entities and interactivity.
Take the outline you created and start drawing out "rooms". Paper is a good choice for this stage, nothing super detailed of course, but a very nice guide. By the time you finish this step you'll end up having the entire conceptualization for your map (until you come up with something new.)
Map like crazy. Just start out by creating, in your editor, what you've created on paper and in your head. Then start going through and adding more detail to each "room". You'll probably come up with more ideas here, whatever you change here go and make changes in the requirements step and make a note of it in the specifications. That way, when you pick the map up again in four months you'll know what the hell you were thinking.
This is the final testing (seeing how something looks in game is part of the mapping stage), which of course, is best done on servers.
I think you can figure out the rest.
But, if you really think about it, this same process is used for most things, at least professionally. (Writing a novel usually follows this type of process. Obviously, planning software follows it, but it is more strict.)
Sorry for the drawn out post. I hope it helps though.