It's not just anonymity, it's anonymity coupled with a lack of consequences.
The ideal solution would be to make users sign in through their Facebook page, but then let them comment under whatever pseudonym they wanted. That way commenters could have anonymity, but abusive users could still be permanently banned.
Minimum account age, minimum number of friends, need to have your account vetted the first time you post, need to be friends with certain people or within two degrees of certain people, etc.
Essentially it's extremely difficult to create a new Facebook persona and have it look at all authentic. Not impossible, but at least 250x more difficult than creating a new GMail account.
I have encountered numerous fake profiles that have survived for months/years. Many of these represent businesses that should really be pages. Others are accounts that are used on hate groups/pages exclusively and have highly offensive fake names.
The worst example I've seen was a vicious cyberbullying profile that Facebook allowed to accumulate hundreds of friends, under an obviously fake name, despite a large number of abuse reports.
There are a number of accounts we created for inanimate objects while at uni which are still around. They all have a certain amount of friends and photos though.
We also used to create fake people using photos from stock art libraries and try to create realistic social lives for them, with our aim being that our real friends would eventually start enquiring after the fake people as if they knew them.
I don't see why you should decide to never have a social account. Personally, I don't trust Facebook, so I don't use it. Twitter, on the other hand, I've found useful both to follow people I know or know of in the context of a hobby &c., and as a microblog for myself.
That does seem to be an excellent middle ground. The pseudonymity would eliminate the element self-censorship that the Techcrunch article seems to be hinting at, while the threat of banning combined with the high cost of creating a new Facebook account would (hopefully) still suffice to keep the discussion reasonably civil.
As a cryptography researcher I've come across numerous cumbersome anonymous reputation systems, but it didn't occur to me that Facebook is in a position to roll one out right now with a minor change if they chose to. The caveat, of course, is that it is centralized and controlled by FB. Whether they have the incentive to implement it is an entirely different question.
"As a cryptography researcher I've come across numerous cumbersome anonymous reputation systems"
What's the best one you've come across? The only way I was ever able to dream up was basically requiring an iris scan to create a new account, but then not connecting the iris data to the account, just using it to ensure the user couldn't create a second account. This may actually be feasible now that every laptop has a built in camera and the patents have finally expired, but at the time I wanted to create this it wasn't yet possible from a business perspective.
The ideal solution would be to make users sign in through their Facebook page, but then let them comment under whatever pseudonym they wanted. That way commenters could have anonymity, but abusive users could still be permanently banned.