Take an interest you currently have that isn't tech related and start solving it. In my case, sports league websites suck (or at least the volleyball leagues I've played in), and I got fed up enough to do something about it.

There's always some underserved niche to be found, and you don't necessarily have to seek it out. Eventually, if you keep your eyes open, something will annoy you enough to change.

So what if it takes a bit more research or even a lot more research? If you want to go into business to make money by building domain-specific software, not knowing the intricacies is not only inexcusable but should very much be frowned upon as downright lazy.

Most startups fail because the serial entrepreneur build stuff they think people need but would never use themselves. It's not as easy as just researching what potential customers may need. If YOU are not that potential customer, then you won't have the intuition to make something better than the competition when they pop up. Also, if YOU're not the target customer, your passion for the project will fade way before you actually make money from it.

The above is true for both projects large and small. I still think Google+ will ultimately fail b/c its top level execs don't use the platform and therefore won't have any insights about where to take it.