I haven't though about this carefully but I wonder what the shortcomings of the following would be: when a domain 'owner' wants to update their DNS entry, they just push out a PGP-signed (or similar) message of their new IP onto a P2P swarm, which will only propagate your message if their message verifies with their public key, from which you can then grab the entry (and verify it again for extra security). Then, registering your domain would just consist of getting the swarm to accept your public key. Of course, this could be first-come first-serve, or it could be regulated in some way. This way you could have multiple swarms, and each swarm is only governed by their own decisions.

And I suppose on a smaller scale, TPB could create their own swarm right away, and have people use it parallel to existing DNS infrastructure. Then other censored domains could jump onto their swarm.

You're right, it doesn't look like the project advertised 2 years ago got past the idea stage.

Assuming someone actually tried, however, I don't think I'd be as worried about tampering as you have suggested... Firstly, SSL and whatever certification system used in the future will still exist to prevent man-in-the-middle hijacking, meaning even if trivial to do (like defacing Wikipedia), I do not think it will occur substantially unless as a form of DDoS (as in defacing Wikipedia), and second... by implementing something like the possibility of manual pinning and upstream authority/record-source blacklisting (a common technique in P2P apps, with an anology in Wikipedia to marking a page/record "frozen" for a period due to contention), it would be possible to preserve consensus, overriding/responding to the current level of state censorship for the forseeable future. If governments more broadly dictate filters, that will change the nature of things (and of Wikipedia... They will be able to change history), but we are not there yet nor do I think we will be there on the order of 5-10 years (and something like this might slow that timeframe). I also think people are getting more vocal about the negative aspect of imaginary property enforcement. ( https://news.ycombinator.com/item?id=5606921 )

Forgive my ignorance, but how did they secure the .bit TLD in the first place? Is this part of the whole slew of new TLDs that are coming out? They must have had a benefactor to make this happen.

Er no it doesn't even seem to be on the list. http://www.newtlds.com/applications

Namecoin isn't ICANN. You install the software and whenever your machine does a lookup for the .bit TLD it uses their lookup system instead of sending it to the DNS resolver.

Namecoin is a very interesting project, but I prefer PPC coin's proof-of-stake algorithm to bitcoin's proof-of-work for long-term stability and robustness.