WoT can't be effectively spammed, unless those spam accounts are trusted by people you trust. The spam accounts can vouch for each other all they want (a la Twitter), but you control whose trust you value.
I was thinking about this the other day. From an information perspective, it shouldn't be impossible to design the system you describe (including the implied nuances) because spam usage patterns do and must look different than normal usage patterns under any system that penalizes new accounts.
Hypothetically, the worst you could do would be astroturf. (aka the US/Chinese military style "slightly biased posts from a large number of centrally controlled but seeming unrelated accounts")
However, the idea of slight bias over longer periods is somewhat antithetical to the idea of a spam. In that it might influence you to buy Sparkle towels (honestly, with Amazon prices that low and shipping that easy... [meta :p]) over a competitor, but isn't going to convince you to navigate to {insert sketch get-rich-quick spam scheme here}.
Weeding out astroturf is an entirely more interesting problem though...
Yeah, that's the problem. Unless I personally build a trust score for every member in the web, all I can do is rely on a score based on friend-of-friend rankings. Eventually a friend and I will disagree on a friend-of-friend ranking. Do I lower the trust of the friend, or the friend-of-friend? Is that even possible? And if it is, how much time do I really want to spend pruning the web?
Good example: I want to see everything my Aunt Susan is doing in her personal life on Facebook. I do not want to see anything ever for any reason that has to do with her Zynga games.
A single trust score doesn't really encapsulate that relationship and it's very possible she would effectively breach the WoT by allowing Zynga to send me messages, notices or e-mail in exchange for her to get a shiny new Farmville tractor or something.
I stopped using Facebook because of this kind of crap. I don't have the time, energy or interest to deal with people I do know sending me crap I don't want, and more importantly Facebook's flexible definition of privacy and customer service. I had Facebook change my settings away from their desired state more than once as part of an "policy" or "feature" update.
So I guess the meta discussion is about whether you trust the holder of the trust. LOL.
It only takes one break in the chain to compromise the entire web of trust. With such wide-spread connections across the planet these days, the chance of someone you trust three times removed accidentally breaking that chain is quite real.
Let's say you see some spam. You could have the software tell you which part of the web has made the spam trusted, and then you could manually mark that part of the web as untrusted. If there are only a few breaks in the chain like that, it'd be a workable solution.
Hypothetically, the worst you could do would be astroturf. (aka the US/Chinese military style "slightly biased posts from a large number of centrally controlled but seeming unrelated accounts")
However, the idea of slight bias over longer periods is somewhat antithetical to the idea of a spam. In that it might influence you to buy Sparkle towels (honestly, with Amazon prices that low and shipping that easy... [meta :p]) over a competitor, but isn't going to convince you to navigate to {insert sketch get-rich-quick spam scheme here}.
Weeding out astroturf is an entirely more interesting problem though...