This. Nobody tells you this part. I've read the blogs of several successful side-projectors and independently came to this conclusion a couple of years ago. Even Tom Preston-Werner had a bunch of side-projects that went nowhere (he wrote about it a while back) before GitHub. So also Dennis Crowley (FourSquare) wrote about his side-projects, incl. Dodgeball (college project?), which was a pre-cursor to FourSquare.
EDIT: Found an AMA that Tom did here on HN in 2010. Some interesting questions about doing side-projects while being employed full-time.
=> Ask Tom Preston-Werner, cofounder of GitHub, anything Today, Mon 18 Oct 2010. https://news.ycombinator.com/item?id=1804443
This x100000. This is fundamentally it; just like how we love to talk about 10,000 hours and targeted practice, so too can we extend this concept to startups. You just have to throw (and attempt) a bunch of ideas at the wall until one of them sticks, and learn along the way.
I used to get so bent out of shape that my side project wasn't a massive success or that I'm not a young prototypical YC founder straight out of college on the fast track to entrepreneurship. When I took a step back, realized that most founders (even outside of the tech bubble) are older than 35, and have tried (and failed) at a lot of different startups...you then embrace that it's just all about executing and learning from as many projects as possible. It takes out a lot of the stress and anxiety about succeeding, and instead using your side projects as practice to get you to that great idea in your future.
People who are succesful did 99 mediocre side projects before they hit the jackpot. But it still took them tons of time and energy. Very often a overlooked by-product which was developed over night and benefitted from all prior experiences will be the cash cow while the main project which took 12 months still hasn't won any users.
Most stop at some point because it's just tough, especially the mental part.