Mike Masnick compiles arguments against the push by Facebook and Google Plus to require real names. From Danah Boyd's contribution:

What’s at stake is people’s right to protect themselves, their right to actually maintain a form of control that gives them safety. If companies like Facebook and Google are actually committed to the safety of its users, they need to take these complaints seriously. Not everyone is safer by giving out their real name. Quite the opposite; many people are far LESS safe when they are identifiable. And those who are least safe are often those who are most vulnerable.

Alexis Madrigal applies the theory to our interactions offline:

[I]n real life, we expect very few statements to be public, persistent, and attached to your real identity. Basically, only people talking on television or to the media can expect such treatment. And even then, the vast majority of their statements don't become part of the searchable Internet. Online, Google and Facebook require an inversion of this assumed norm. … On the boulevards and town squares of Facebook, you can't just say, "Down with the government," with the knowledge that only a small percentage of the people who hear you could connect your statement to you.  But the information is still being recorded, presumably in perpetuity. That means that if a government or human resources researcher or plain old enemy wants to get a hold of it, it is possible.