Simple positive trust scheme with threshholds

Update: I nowadays think that voting down is useful, but only for protection against spam and intentional disruption of communication. Essentially a distributed function to report spam.

I don’t see a reason for negative reputation schemes — voting down is in my view a flawed concept.

The rest of this article is written for freetalk inside freenet [1], and also posted there with my nonanonymous ID.

That just allows for community censorship, which I see as incompatible with the goals of freenet.

Would it be possible to change that to use only positive votes and a threshhold?

• If I like what some people write, I give them positive votes.
• If I get too much spam, I increase the threshhold for all people.
• Effective positive votes get added. It suffices that some people I trust also trust someone else and I’ll see the messages.
• Effective trust is my trust (0..1) · the trust of the next in the chain (0..1) · …

Usecase:

• Zwister trusts Alice and Bob.
• Alice trusts Lilith.
• Bob hates Lilith.

In the current scheme (as I understand it), zwister wouldn’t see posts from Lilith.

In a pure positive scheme, zwister would see the posts. If zwister wants to avoid seeing the posts from Lilith, he has to untrust Alice or ask Alice to untrust Lilith. Add to that a personal (and not propagating) blocking option which allows me to “never see anything from Lilith again”.

Bob should not be able to interfere with me seeing the messages from Lilith, when Alice trusts Lilith.

If zwisters trust for Alice (0..1) multiplied with Alices trust for Lilith (0..1) is lower than zwisters threshhold, zwister doesn’t see the messages.

PS: somehow adapted from Credence [2], which would have brought community spam control to Gnutella, if Limewire had adopted it.

PPS: And adaption for news voting: You give positive votes on news which show up. Negative votes assign a private threshhold to the author of the news, so you then only see news from that author which enough people vote for.