It's not that these ideas are necessarily bad. But they need to convince me that they're not being done purely out of the fear of getting out and talking to potential customers.

Make a product you would use yourself is all well and good, but there are two ways to get there: limit your products to the domains you know, or go out and learn new domains.

I believe the second strategy has a dramatically higher chance of success, because you can pick from a near-infinite choice of domains, many of which are woefully lacking in people who understand software.