Personally, I make homebrew because it's something to do.
I rarely enjoy games anymore.
I'll happily play GTA IV for a few hours at a time, maybe once a week.. But then.. that'd be about it for another week.
I much more enjoy making a silly game than I do play them, and more often than not that shows in my lack of ever completing a game.
Even GTAIV that I've been playing off and on for a good few months now (Bought it in November) is only showing a 57% completion on it!
Do I care?
Do I buggery, it's a fun game, and I'm enjoying playing about with it.
Could I MAKE GTA? Could I buggery! Unless I spent 8 years working on the exact same project, and then end up with something vaguely resembling GTA 1!
But 8 years is a long time..
My games take a week to make, and usually even less than that.
Sure, it shows, but I'm glad that I didn't spend 3 years on a game, go on some weird-ass rampage about trying to sell it, and then eventually fade away into obscurity.
And a lot of the homebrewers out there are in the same position.
It's not about "Teh Besterest Game EVaarrr!!", it's more about "OMG! I made this!!", and "I wonder what would happen if...."
Making a silly little game for yourself, and if anyone else likes it, then that's a plus point.
Freeform game design's much more fun than spending 18 months following a pre-written layed out gameplan.
The best rule that I can give to all homebrewers, though, is to stick to what you know.
If you overshoot your abilities, it really does show.
I can't count the number of games I've seen that have glaring mismatched graphical styles, with rendered 3D objects sat upon MSPaint-style backgrounds.
It's terrible when folk do that.
Pick a style, and stick with it. Don't overdo things, keep it simple, and focus on the way the game plays more than how it looks.