Terrific slidecast about trust and reputation by Rentathing

By Rentathing. In a single slidecast, they explain clearly a possible and reasonable future society in which interactions are based on trust and reputation (someone would call it “whuffie”). And it even shows examples of ridesharing (letting other people benefit from your car) and couchsurfing (letting other people benefit from your house). A slidecast is worth thousands words.
UPDATE: As I was suggesting in the comments, read “Down and out in the magic kingdom” (Creative commons released so you can download it for free and much more) or just the Whuffie page in wikipedia: “Whuffie is the ephemeral, reputation-based currency of Cory Doctorow’s sci-fi novel, Down and Out in the Magic Kingdom.” BUT if you intend to read the book (suggested!!!), don’t read the wiki page!
And also check Ripple (really blowminding!) and its white paper “Money as IOUs in Social Trust Networks & A Proposal for a Decentralized Currency Network Protocol”.

Created with Admarket’s flickrSLiDR.

10 thoughts on “Terrific slidecast about trust and reputation by Rentathing

  1. Folletto Malefico

    Doesn’t exist a bit issue in the reputatio-and-trust-based systems, as I was thinking the other day about Digg?

    In other words, I think that a system like this needs more work on the weak points than on the strong and useful points: the trust system is quite easily hijackable if you have malicious intents.

    So you’ll get high reputation, and you’ll get more since many will use your “services”, also there is the problem of closed loops: reputations within a group already exists and are hardly converted into numbers…

  2. paolo Post author

    > I think that a system like this needs more work on the weak points than on the strong and useful points
    yes, but it is not very wise to try to sell your points (i.e. convince people that this is a great possible society) by highlighting the (existing) weak point, no? ;-)

    Anyway, trust metrics can be attack-resistant, the first step is to make them very local, i.e. there is no global reputation of identity “bill_gates” but only your trust score in “bill_gates” and this depends only (or mainly) on what your trusted users (friends) trust.
    To put it in Digg perspective, the reputation of digg founder is not the same for everyone and so stories promoted by him don’t necessarily hit the homepage, simply because there will not be one homepage but a different homepage for every different user: the stories that hit YOUR homepage will depend on the users YOU trust. You are free to trust “bill_gates” or “richard_stallman” or “osama” or “bush” … it is entirely up to you. Then the system is already much more robust, isn’t it? ;-)

  3. Folletto Malefico

    Solving the hard points in my opinion could “sell” better than just showing the positive ones. :D

    At a first glance, that kind of system seems more robust, sure. :)
    I wonder how it could be implemented practically. :)

  4. paolo Post author

    Practically it is not hard to implement, it is very requiring from a computational point of view (you don’t have to predict the trust of all the n users just once from a global point of view but n times, from the point of view of very single user) but it can be done, offline and continuously.
    It remain to be seen if users will like it, most of the online systems use a global reputation and this seems to work fine for most of the users. Somehow I wrote about this in this paper http://www.gnuband.org/papers/a_survey_of_trust_use_and_modeling_in_current_real_systems/
    and partially in http://www.gnuband.org/papers/reputation_is_in_the_eye_of_the_beholder_on_subjectivity_and_objectivity_of_trust_statements/
    and http://www.gnuband.org/papers/trust_metrics_on_controversial_users_balancing_between_tyranny_of_the_majority_and_echo_chambers-2/
    but I don’t suggest reading them ;-)

  5. Folletto Malefico

    As it’s one of my actual points of view, I think that to work *well* it could require heavy investments in an infrastructure. I see two problems:
    . ubiquitous computing (infrastructure issue)
    . computational issue (could be addressed, but… needs centralization? or p2p? everyone needs somewhere where to put its “personal trust server”)

    I’ll look at those… but I have to find some time. :D

  6. paolo Post author

    Not sure I got your point about ubiquitous computing.
    About computational issue, running everything centralized is always easier, if there is no specific reason (usually there isn’t) making it centralized is much more likely to turn your project into a success than making it decentralized which is much much more complicated.

    But some things have to be decentralized! EX: your identity (openid servers) and your currency account ( check http://ripple.sf.net , THIS is interesting!)

  7. Marco

    Not my field, so I may be completely off-base :-)

    One thing that surprises me when reading about trust and reputation system is that nobody seems to mention the one, big, trusted (no pun intended) reputation system: the credit system.

    In the credit system, credit worthiness is one’s trust; various metrics exist to quantify how trustworthy one is (e.g., FICO in the US); friends (read: parents) can vouch for somebody else (e.g., by co-signing their child’s mortgage); sites exist that collect, manage, and update trust values (credit report companies); trust drammatically affect one person’s life (e.g., how much money he/she can borrow and at what rate), etc.

    So, I guess my question is: what are the main differences between the “web2.0 trust systems” and the credit system (life1.0 trust system, if you want)? Identities cannot be faked to some extent is one. Centrality and control of the trust computation is probably another point. Other ideas/comments?

  8. paolo Post author

    Check Ripple, it is really blowminding!
    The paper “Money as IOUs in Social Trust Networks & A Proposal for a Decentralized Currency Network Protocol” or just the site.

    And read “down and out in the magic kingdom” (cc-released so download it for free) or just the Whuffie page in wikipedia http://en.wikipedia.org/wiki/Whuffie: “Whuffie is the ephemeral, reputation-based currency of Cory Doctorow’s sci-fi novel, Down and Out in the Magic Kingdom.” BUT if you intend to read the book (suggested!!!), don’t read the wiki page!

    Well, one difference is that you cannot lose all your money in one second but you can lose the trust all the people (or many of them) pose on you in one second. This invites you to “don’t turn evil”. If you are a rich politician (read Mastella), you don’t care about normal people because you have a lot of money (and power) accumulated somewhere. This is with money. In a reputation/whuffie society, I have some ideas where Mastella would be.

    As a consequence, accumulation of capital is not possible in the reputation/whuffie/trust economy. This means that if you mother is rich, but you are a jerk, you don’t inherit automatically its trust.

    Does it make sense?
    Are there downsides in the whuffie society? Of course! If you read “down and out in the magic kingdom” you will start to see some of them. But there are many and of course I cannot be sure that that society will be “better” that the present society based on currency capital, whatever definition of better we give and whatever point of view we use for evaluating it …

  9. Marco

    Mmmh, money seems to work differently from credit worthiness.

    For example, as you say, you can’t lose all your money in one second, but you definitely can trash your credit score in one second (e.g., identity theft, bankruptcy). You can transfer money around easily, but not so much credit worthiness (which seems more similar to the transitive transfer of reputation).

  10. Pingback: MobBlog » Blog Archive » Can I rent your battering ram?

Leave a Reply

Your email address will not be published. Required fields are marked *