Do you think your project will need new features regularly? Pick a project whose problem domain changes rarely, and get out of the features = value mindset because it is outrageously false.

Do you think you will have to be constantly selling? Don't pick something with high touch sales processes for a side project.

Do you think your project will require support regularly? Do something less critical, so you have less time pressure on support. Do something simpler. Give people self-help resources and push them. Automate as much of your support burden as reasonably possible. (e.g. "Hey Patrick I need a receipt" takes me about 10 seconds to answer. It used to take a few minutes, but I get asked it frequently enough to justify 30 minutes of dev to make it quick.)

Do you think you will need to play sysadmin frequently? Become a better engineer. (I have a couple articles about this.)

I am not the sharpest knife in the HN drawer by a long shot. Currently I'm travelling internationally doing some networking at a Silicon Valley startup and, on the weekdays, consulting in NYC. I profoundly hate flying right now. I also have two software businesses and, from their perspective, this state of affairs is exactly identical to me just sleeping in back at my house. It is rather unlikely the businesses will cause significant work for me.