Just create an account with "John or Jane Smith" or any other believable combination of common names.

That doesn't work long term. Sooner or later, your social and activity graphs overlap enough to warrant flagging.

Care to elaborate? Preferably with proofs that if activity and social graphs of two accounts ovelap at some threshold level user accounts are flagged.

I don't work at Google, so you won't see any proofs out of me (nor of course you'd see any if I did work at Google).

But sooner or later, the browsing habits out of the two active accounts are bound to converge pretty well. Remember you've been tracked not just when you post comments, but on every page out there that has a "+1" button. Then, there will be some close friends who'd be connected on both accounts.

In addition both your real and fake account are very likely to be accessed from the same set of IPs or subnets, locations and time slices.

This may sound complicated, but this sort of mining is just an extension of spam filtering techniques (to which at this point Google invested many millions), and requires zero human intervention. And since it can be done and there is an incentive for Google to do it, it will be done.

So no proofs, got it.

btw, I agree that Google will try to correlate account data. That has obvious business value. But it is not obvious at all that Google will flag accounts based on this correlation. That was your statement I was arguing against, not practical possibility of matching various accounts to the same owner.

Why even have a real name account if you don't trust Google? Just always use it under your pseudonym.

This is fine, assuming you don't use Gmail or other Google services as primary - and many people do.