I love Patrick's insight here:

Stepping back a bit, problems feel a bit less important to me as time goes on. The world is awash in problems. Problems you care about are just so much more rewarding to work on, and they pull better work out of you. Problems experienced by people you care about quintuply so, because if you start a SaaS business you'll spend much more time talking to customers and getting in their head than you will be on e.g. modeling their workflow in Ruby or optimizing your AWS spend.

So I'd suggest most people asking this question instead ask "How do I find a user population that buys software that I would enjoy having in my life five days a week for 5~10 years?" My answer for that, after a lot of soulsearching, is "I enjoy helping software people a lot more than I enjoy helping undifferentiated professionals. Maybe I should just do that. Maybe no maybe." Your answer may vary.