Depends on what you can actually finish I suppose.

I have several game maker projects which are long dead and gone because I got bogged down trying to write nifty engines and crap for them. In one case, I spent probably weeks working on my own custom (and, to be fair, only barely passable) physics engine for a game which can now be accomplished in a few lines now that game maker studio comes with a physics engine.

I can only imagine how deep the rabbithole would go if I ever started to learn game design in C++ or something (which, of course, I plan to) and then added language design on top of that. Would probably be really fun though.

In line with the other commenter, I have spent many years tinkering with just game engines in my spare time, and indeed it is a rabbit hole.

For that reason I think things like indie game jams (Global Game Jam & Ludum dare) are a good idea because you will write crappy code, but you will actually make a game.

I guess this is why we have 100 HTML5 game engines and 0 HTML5 smash-hit games.