The bottom line is to work on something as long as you're getting something out of it. Whether that is money, knowledge, 'cred', or *gasp* personal enjoyment that's all that matters. However you rank things like money, fun, and knowledge is how you should rank your projects.
I will say this though: Games are like children. They start as infants, so adorable and you have such high hopes for them. It's brain chemistry and instincts. Your first rigged character, the networking finally working correctly good enough... every day you're blown away by how much it's improved so fast. Then they get older... they start to become stubborn, they learn use the word 'no'. They become obnoxious and they no longer fit in your free bitbucket repo, they suck up your money and your time, and their code looks like ground zero of an explosion. You can't wait for them to move the fuck out and get their own job and start supporting you. Aaaaanyway, I'm sorry if that went a little too deep and off topic.
<edit: returning-to-topic>
I'm almost always working on at least two projects at once. One is usually to keep food on the table and one is because I genuinely want to see it made. (or I want to figure something out) Many times it's 3 and occasionally 4+. (not recommended)