[fishtoaster]

There are a lot of properties of names that go against the basic "John Smith" format. How many of those properties you want to support probably depends on how much of the population you need to service.

For a bootstrap proof-of-concept project, I might expect names to be a pair of strings with no spaces in them. This'd cover (at a guess), 95-98% of US users.

For a startup, I might expect two strings, one of which can be blank, each of which contains 0-50 unicode characters. This at least adequately support (at a guess) 99-99.99%

For a government organization, maybe a single string, 1k unicode characters, and a way to snail-mail in a different name if yours doesn't fit that definition? That might be a bit extreme for a local government, but something like the IRS, where you have no choice but to use their service, has to push the limits of permissiveness.