Could Your Username Be Used For Fraud Detection?

If you go to just about any social website and search for user “hunter walk,” there’s a reasonable chance you’ll find me. It just always seemed easiest to employ my real name and its uniqueness meant it was often available. Any site I register with would also see it’s the name in my email address and that my identity corresponded to a number of other sites where there exists proxy data suggesting I’m a good egg (eg in good standing on LinkedIn). So basically I should be trusted on any new site since it’s a very low probability that I’m a fraudster, at least if past behavior is indicative of future. So why does each site start off not knowing a thing about me?

I’m not talking about a Facebook-style federated identify but wondering if there’s a “Username Analysis as a Service” component to fraud prevention, or at least asshole detection. Can a username be valuable in predictive analysis about behavior? For example, do users who purposefully misspell words in their screen name behave better or worse? Are nonsense usernames more likely to be spammers? Can someone who seemingly uses a real name (“Hunter Walk,” “Larry Goodstein,” etc) be trusted because they’re using their name, even though the system doesn’t require it?

Cursory Google search didn’t turn up any academic research on this. Can anyone point me to relevant studies?