I'm not convinced you understood what I was saying in the way I intended it to be understood, so let me clarify.

PGP's web of trust is designed to answer the question "is john@example.org who he says he is". A user is "valid" (in the above sense) if you have marked it as such, or if a sufficient number of users of sound judgement vouch for it. (There are multiple levels of "vouching", and the meanings of "of sound judgement" and "sufficient number" are user-configurable; and "users" are really "keys".)

By default, PGP believes that users who are vouched for as in the above model are valid but not of sound judgement (i.e. we don't let this users vouch for others), but you could easily change that.

The point I was trying to make is that, if you replace "PGP" by "Outer Circle" and define "valid" to be "worth reading", the above sounds a lot like what you want to achieve. It has also been deployed in the real world for quite a while, so you can see to what extent it has worked (it's rather succesful in a very small part of the population).

Additionally, PGP is effectively decentralized, even though day-to-day operation uses keyservers. This model is a lot easier than full peer-to-peer (which, in the presence of NAT, is probably not possible without centralized servers anyway), and a combination of clever crypto and fallback options means that PGP does not rely on keyservers. Look at the documentation for more details.

Notwithstanding the above, which is honestly intended to be good advice, I don't see how your system is ever going to become sufficiently widespread to be worth learning for the average user, and I don't really see the value in keeping a medium that's 99% nonsense usable for the last few erudite users - just let them migrate.

Thank you for expanding on that. In trying see the relevance of PGP's web of trust to Outer Circle, two things throw me off the scent.

First, identity in Outer Circle is simply a public key used to decode each signed post. When an identity is highly regarded, others will wish to impersonate it, in order to get their writing or advertisements read, but they are unable to do so because they do not know the private key.

Prevention of passing off is the whole of identity in Outer Circle. We do not care if users have multiple accounts nor do we care whether group accounts are acknowledged or pretend to be an individual. Circle jerks, be they conspiracies or sock puppets, do not undermine the Outer Circle system because an identity, even in good standing, has no right of recommendation. If some-one on your direct list starts reading crap you "shallow" him or drop him, and that is part of the normal operation of the system as tastes change (his or yours) or time pressures squeeze.

So it is confusing to me to read about web of trust because Outer Circle has no notion of who somebody really is. Ossian is welcome.

Second is Outer Circles core concept of the coexistence of incompatible through invisibility. Whether somebody is who he says he is is a valid concept that is either true or false. I see the idea that a post is worth reading or not worth reading is a trap. There is no truth of the matter and that is the heart of the reason that traditional forums with moderation and meta-moderation collapse. They are trying to forge a consensus when they should be facilitating disagreement.

You doubt the value in keeping a medium that is 99% nonsense usable for the last few erudite users. That foregrounds the issue of scaling. I picture Outer Circle growing from a thousand elite users to million users, 99% hoi polloi 1% elite. So the proportion of erudite users falls, but the absolute number rises tenfold. I see the endless migration of the nomadic elite as an important problem, preventing them reaching critical mass.

Do you still think that PGP's web of trust is relevant? How do I relate "is john@example.org who he says he is", which has type boolean, to "john is worth readng", which has type function from users to boolean?

Hmm, you had not previously expanded on the prevention of impersonation, although that is indeed desirable. In fact, piggybacking this whole thing on PGP (with special "is interesting" messages) may work. Somewhat.

I don't see why the web of trust-approach would fail. Certainly, I admit, "is worth reading" is not (intended to be) a global value, while "is really that person" is.

But ultimately the web of trust answers the question "do I trust that person to be who (s)he says (s)he is". For instance, someone who trusts a lot of people is likely able to verify that key D5327CB9 belongs to Wietse Venema, while someone who has just created a key but doesn't yet trust anyone cannot. Therefore, while this key belongs to Wietse or does not belong to Wietse irrespective of who asks this question, the answer may differ.

There are obvious parallels to "do I trust this person to produce interesting content/tag appropriately".

I can understand your desire to bring the elite together, but what is the big added value over, say, Twitter or blogs? These are fairly stable, fairly reliable identifiers of people, with connections between them. Filtering out the nonsense is so easy that it's hard to notice that we're even doing it, and "clusters" of smart people form more or less automatically. (Which is not to say that either Twitter or blogs are the best possible communication medium, especially for small messages like this, but they seem to get this mostly right.)

What is the big added value? I was prodded into making another attempt to bring coherence to the swirling mass of ideas I call Outer Circle by a recent thread here on Hacker News: http://news.ycombinator.com/item?id=916448 Other people consider the forum problem important.

When Kuro5hin died, I started reading The Edge, http://www.edge.org/. However The Edge is what I call an Inner Circle site. Outsiders get to spectate, but there is no way for outsiders to ask questions or raise objections. The Edge seems introverted to the point of sterility. My dream for Outer Circle is that it should impose a wholesome discipline on outsiders. They are not confined to spectating but can ask questions and raise objection. On the other hand outsiders have to work for access, playing nicely, and earning a place on insider's white lists.

The big win happens when you have people you don't know making unexpected positive contributions. The hard question is how to permit that without suffering the usual forum problem.

I understand that people believe this to be important, and I see the value in interesting communities. It's just that I don't consider the forum model something to emulate, but rather a regression from earlier models - I much prefer mailing lists or Usenet groups, both of which allow a user much more control via their client, and both of which seem to, on average, suffer less from "the idiot problem" than their web-related counterparts.

Part of the reason may have been that many web fora try to appeal to the broadest possible audience, while many mailing lists and Usenet groups have a formal or informal charter.