Lots of people who are happy to "be themselves" in safe settings, around likeminded people, will "clam up" or take on a persona they don't like immediately if the alternative is that anyone can link something to their name.
Political beliefs, religious beliefs, sexual orientation etc. Even gender - there are plenty of women online in male oriented environments that are uneasy about using their real name in certain settings because of bad experiences with sexism.
If they'd bothered to listen to any of the people they affected, they'd have learned this on day one. But it ought to have been blatantly obvious: In real life we don't go around with name tags and hand out dossiers on ourselves to random strangers, which is the equivalent effect of using full real names online.
I don't know what it says about the culture of Google that this was such a hard realisation for them.
Quoting Yonatan (tech lead for G+):
> There were quite a few things we were trying to achieve, some of which turned out to be key and some of which didn't. Greater authenticity was actually one of the big successes: it set the culture to be one where people, by default, act as themselves rather than as a persona. That makes a huge difference to how people behave, and now the culture is stable enough that it's OK to relax that. Reducing trolling, on the other hand, turned out not to be as tied to names as anyone thought. We had lots of reasons to believe that might be the case at first -- look at YouTube, 4chan, and Facebook comments back then -- but in retrospect, the dynamics which control trolling seem to be somewhat different. Or to put it a little more bluntly, people are quite willing to be assholes under their real names, too.