I was actually working fulltime at Powerset and doing GitHub on the side. I spent a lot of time coding, but it's what I love. Working on GitHub was how I expressed the creativity that I couldn't when I was doing tool support at Powerset. It was something I could call my own, and it was a way to unwind after a long day in meetings. Money wasn't a problem. I was getting a good salary at Powerset.
I found my cofounders via local Ruby meetups. I was by myself when I was doing Gravatar and it sucked. Having two cofounders with GitHub has been one of the best decisions I've ever made. They provided additional manpower in the early days when velocity of development matters most, and they acted as a safeguard against bad decisions. It's extremely motivating to have other people that are depending on you to produce great work. I would highly recommend against single founder startups for all of these reasons and a thousand more.
Gravatar did become hard to run towards the end. I was spending a few hundred dollars a month on servers and there was no good way to make money with it. It seemed a bit hopeless. I never gave up though. I put in the hours to make it work and eventually sold it to Automattic. To me, that was a success. I created something of value and sold it to someone that had the resources to take it to the next level.
It's hard to explain how I judge my startup ideas. It's a lot of intuition based on years of experience with the internet and with a lot of examples of things that I love and ton of examples of things that I hate. All this adds up to a feeling that something will work out. Once I have that feeling about an idea, I jump in 100%.