I think every project passes a threshold after which the time invested will only be considered worthwhile if you ship it.

Before this threshold, you're learning new technologies, trying new techniques, and generally developing skills that could be applicable elsewhere. After this threshold, you're becoming an expert in the project itself, which will all be for naught if you don't finish it.

So my advice would be give up on projects that are near that threshold, and only continue beyond if you intend to finish it.