Personally, I hope it's not all full of Facebook and web 2.0 crap. That's about as far from anonymity as you can get.
To me, the most basic concept of an anonymous board is that a message cannot be traced back to its poster. I think even storing an IP address is more than what the ideal anonymous board should be doing. The only legitimate use for an IP is to ban abusive posters (and the only two things that should really qualify as abuse are spam and child porn), so on my own boards I run a script that scrubs IPs from any post that's more than a week old.
I think Linking with Facebook/Twitter/Youtube and all that crap is completely missing the point, and registering usernames even more so. Anything that persists state information between one message and another ought to be theoretically minimized or removed. With current anonymous boards the only useful exception to that is the poster ID, but I think that concept is also flawed: Kareha uses the IP address to generate the poster ID, so it's still technically possible to do long-term user tracking with it (if you know the salt which is used to hash the ID; and with sufficient computing power this could be theoretically determined).
Honestly I have a very dismal outlook on anonymous boards. Not very many people see any practical use or benefit to them, and those who do aren't usually interested in them for actual reasons of anonymity. What I would like to see is a system where no one post can be traced back to its author, or correlated with other posts, but where it is still possible for an author to (willingly) self-identify as a previous poster on the thread. I think the most obvious implementation of this would be with a public-key system, where new key pairs are generated for each message unless one wants to deliberately identify as the author of a previous post.
This could even be handled with existing board software with a few modifications (and likely also some education about public-key authentication and message signing to the prospective users), but I think if it were implemented in a streamlined manner, it would be far more groundbreaking than any of the ideas presented in >>1. (and a great number of >>1's ideas have in fact been implemented in various forms on anonymous boards already)