Martin Nowak's work on the evolution of cooperation imply that large communities are likely to be destroyed by free riders without mechanisms for creating and advertising personal reputation. This is a result based on solid mathematics and validated by simulation experiments. I regard these kind of results as a lot more robust than listening to the opinions of the day. (Anyone who is interested should read Nowak's book, Supercooperators, for a review of his and others work in this field.)
In practice, the negative effect of the lack of a reputation process is clearly present on the net. We only need compare the average quality and civility of anonymous versus named commentators on blogs for a vivid demonstration of this. Outfits like eBay use reputation to weed out crooks and incompetents and we have probably all seen the strenuous attempts of eBay sellers to protect their reputation. This is the theory of cooperation in living detail.
Google's policy seems to me to be fundamentally good for the future of the Internet. I expect to see a time where groups will only allow interactions with people who have a good online reputation, for example, discussion groups only open to people with a G+ or similar validated identity and history. In future, I see systems evolving that allow a reputation to receive negative and positive points and for those attributions to be sourced to known individuals, so that an attribution is rated by the reputation of the giver. The maths and the systems are complex and will need to evolve but the potential benefits are enormous, and probably essential. The members of a traditional village can learn each other's reputation and protect and extend themselves appropriately for the interaction; a global village requires more sophisticated mechanisms.
As for the supposed attack on liberty, I'd say this: Liberty won't survive without good systems to protect it. Free riders will always have payoffs to rort large open systems unless they are detected and advertised. Google's real name policy does benefit Google but in the long run these approaches are critical for us punters too. I'm happy for the monitoring and limiting of Google's use of our identities (and, despite the target size, I'd generally rate Google's behaviour at the pretty good) but mechanisms of reputation are intrinsic to the survival of liberty. We all want the abilty to free ride when it suits us, and generosity is required for cooperation to persist, but it should be given by choice, not taken at will.
[repost of comment to The Register; http://www.theregister.co.uk/2011/08/30/google_plus_anonymity_ban/]