http://web.cs.ucdavis.edu/~defigued/index_files/trustdavis.pdf Trust is problematic with strong privacy and anonymity. Maybe this is a workaround.
Interesting, so it's about reputation in online shopping or so which can't be easily manipulated by many of "spam" review.
As written, this is about real-life stuff. From page 2: "Security deposit" = money. In order to have an identity, you must deposit money with an existing member, who puts their deposit at risk by vouching for you. Through the use of anonymized Bitcoins etc, this could readily be extended to pseudonyms. However, basing trust on money doesn't work except for relationships that are based primarily on money. Better would be a system that used trust itself as the deposit. But how can one quantify trust? I don't know. It's a tough problem.
It is a tough one, and to an extent money is the only trade-able counter that might work in this context. I think there are other forms of affirmation and linkage to online identity (which has a value in its own right if persistent and authentic). So I'm reminded in the latter case of the keybase.io approach which does link some kind of trust-in-the-persona together with their online presence. Of course, there are major downsides to assertions of trust and reputation (think Othello!), because they can be gamed by the unscrupulous and by governments (who certainly have the ability to manipulate and destroy reputation outside the rule of law). And that aspect is what is one of the most damaging aspects of the loss of privacy and illegal surveillance that has happened (whether corporate or government) - a lack of trust, and a lack of confidence in the marketplace. And I do not think that schemes which attempt to put sticking plaster over that wound will ever be really satisfactory. We need better governance and consumer protection.
Well, tho I haven't read entire paper, it is a mathematical model which is hinted from real-life relationships. For this system work to address trust problem in anonymity, each related players have to have known his relatives previously (tho new player can wait til newer player to come). E.g. now this reputation system was introduced in a P2P network and I entered the network with certain HN. I previously exchanged PM with you here in Wilders (...I don't know if Wilders PM is secure. lol) that I'll enter the network with certain HN and also about your HN (possibly also password, unless there's sth like keybase). I entered and find you and both of us accept liability each other. Once either you or me violated someone's trust then partner have to compensate it in some way. But as you say, it will only work on money-related network or sth like that, tho we can just imagine sth, say, trust itself is somehow quantified and all its info are shared in all P2P players just like Bitcoin's Block chain, but to be honest I can't imagine its actual form.