Who do you trust?

Is this an idea worth pursuing?


  • Total voters
    23

DuaneJackson

Free Member
Jul 14, 2005
8,642
1,100
Brighton / London
FlyingCards said:
Prober Industrial Ltd. [email protected]
This is my agent in China I can vouch for him 100%
http://honestasia.en.ec21.com/company_info.html
Paul

I just saw this post from Paul and emailed someone I know that might find it useful with the details saying someone I trust 90% (established forum member but never done business with) is recommending this guy saying he trusts him 100%

It got me thinking, if you put all your contacts in a databse along with trust levels, and I put mine in, along with a trust level for you, then I could seach for anyone in any industry and get an idea of how much I should trust them dependant on the trust level between me and you and you and the supplier.

A good idea for a site or a crap one?
 

cjd

Business Member
  • Nov 23, 2005
    15,987
    3,427
    www.voipfone.co.uk
    It's quite interesting and a little similar to the ebay rating scheme. Obviously open to abuse with 'ringing'.
     
    Upvote 0
    C

    compuclean

    But wouldn't your level of trust differ to mine? You see, you may not have had bad experiences with someone, but I may have, and may not trust him. So who gets to decide what the 'trust level' is, and can that be amended?
     
    Upvote 0

    DuaneJackson

    Free Member
    Jul 14, 2005
    8,642
    1,100
    Brighton / London
    The trust level is on a one-to-one basis.

    If I trust you 10% and you trust John 80% then I trust John 8%

    However if Colin trusts you 90% then he would trust John 72%

    Or something like that

    If there are multiple paths to John through my contacts then an evergae rating could be calculated, along with a breakdown of how it was calculated.
     
    Upvote 0
    A

    amo crafts

    A good idea IMO but with caveats.

    I may be cynical but I do wonder how many people are willing to give out their suppliers to their competitors. Though this site does seem to be more forthcoming in that regard than many.

    You would need to only allow pre-trusted members to enter details. Otherwise you'll have every scam artist and his dog posting details of their 'company'. Although you may have a 0% trust of that member, the database may soon fill up with spam.

    You'll need something like the amazon review summary, using your trust level as the weight before taking the average.

    AVG( SUM( their_trust * your_trust )) ? or something along those lines. Okay, okay, syntactically wrong but you get the idea. :)
     
    Upvote 0
    F

    Free Line Rental

    Sounds a great idea.

    How many of us have paid for credit reports....... only to do the deal and get stung.

    If your idea could be managed on a subscription basis it would stop spammers and provide real time info.

    Already working on the site for this lol
     
    Upvote 0
    It's much like a recomendation site really or personal recomendation with having to get a recomendation if you know what i mean.

    As for trust, well generally I trust everyone %50 when I first meet them then it builds or decreases as the relationship, other things influence it like if someone I trust puts in touch with them I almost instantly trust them 75% or more from the start and so on. Never judge a book by it's cover either way.
     
    Upvote 0
    An interesting idea for a website, I think the question is who trusts the trusters, if you see what I mean.



    Case study


    This lot http://www.118trades.com/

    Keep pestering us, to become members. They got hit a while back by BBC's rougue traders for not checking references, which was one of their supposed USP's.

    After that. This appeared, check at the top left of the screen. That's right that meant to make you think it's all above board.

    Because http://www.verificationassociation.org/ check the checkers right?

    Interesting that the same bloke owns both companies.
     
    Upvote 0

    Mac Yeti

    Free Member
    Aug 22, 2005
    1,161
    13
    Arctic
    Much like the man Gary says, you could set up a network of people that you trust, and therefore trust their recommendations.

    The more those people are trusted by other people, the more their influence on the percentage of trust to give others. Does that make sense? Probably not :)
     
    Upvote 0

    DuaneJackson

    Free Member
    Jul 14, 2005
    8,642
    1,100
    Brighton / London
    GillOakWeb said:
    Trust with what though Duane?? Your life, your kids or your money??

    Its a very subjective thing so why then not just ask people to give a recommendation and how and when they have used the service/person and as on ebay and various other places, rate them according to various levels of service etc

    Trust them to do a good, professional job. I wouldn't use a website to find a childminder or a heart surgeon : )

    Ebay-type systems have their place, but it's very black and white. You just have a number based on what a group of other people that mean nothing to you have said.

    with this system there would be value in the rating because it's based on your own opinion of others judgment
     
    Upvote 0

    DuaneJackson

    Free Member
    Jul 14, 2005
    8,642
    1,100
    Brighton / London
    Ooooh, look at Jilly doing her diagrams on how it could all work! go girl!

    Get the felt tips out, get it all down on paper and I'll give you a percentage : )

    You're right though. A number of influences could be thrown into an algorithm to get the end result. Someone trusted 95% by 40 people i trust are of more interest to me than someone trusted 100% by 1 person I trust 100%and this would have to be reflected in search results.
     
    Upvote 0

    cjd

    Business Member
  • Nov 23, 2005
    15,987
    3,427
    www.voipfone.co.uk
    You need a penalty system too - otherwise there's no reason not to lie.

    If someone recommended turns out to be a wrong 'un all the recommender's other recommendations need to be downgraded and his status reduced.

    you don't need to worry about monetizing it - if it turns out to be a good idea, you'll get the traffic
     
    Upvote 0

    Mac Yeti

    Free Member
    Aug 22, 2005
    1,161
    13
    Arctic
    DuaneJackson said:
    Ooooh, look at Jilly doing her diagrams on how it could all work! go girl!

    Get the felt tips out, get it all down on paper and I'll give you a percentage : )

    You're right though. A number of influences could be thrown into an algorithm to get the end result. Someone trusted 95% by 40 people i trust are of more interest to me than someone trusted 100% by 1 person I trust 100%and this would have to be reflected in search results.

    Ah see now you mentioned the word algorithms and I just saw my brain running out of the room shouting "you won't take me alive!"
     
    Upvote 0

    DuaneJackson

    Free Member
    Jul 14, 2005
    8,642
    1,100
    Brighton / London
    cjd said:
    You need a penalty system too - otherwise there's no reason not to lie.

    If someone recommended turns out to be a wrong 'un all the recommender's other recommendations need to be downgraded and his status reduced.

    surely that's inbuilt? Scenario:

    i trust you 90%, therefore my downstream are likely to use your services.

    I use someone you say was also 90% trustworthy and they stitch me up.

    I edit my trust rating for you down to 20%. Resulting in less business for you from my downstream?
     
    Upvote 0

    DuaneJackson

    Free Member
    Jul 14, 2005
    8,642
    1,100
    Brighton / London
    cjd said:
    If someone recommended turns out to be a wrong 'un all the recommender's other recommendations need to be downgraded and his status reduced.

    further to my reply above...

    It's all based on opinion. I would downgrade you, which in effect downgrades all your recommendations for me and my downstream. Andy never had a problem with you so he doesn't adjust his rating for you, so his downstream still see you in the same light.

    In fact, Andy used the same guy I did via your recommendation and was well chuffed. He even added the guy to his own list of recommendations which then made him (the person you recommended) even more highly recommended to Andys downstream whilst my downstream remains totally unaffected.

    I could look at someone and see a 10% rating, whereas someone else could look at the same guy and see 90%. this is becuase, as gill says, it's all subjective.
     
    Upvote 0

    cjd

    Business Member
  • Nov 23, 2005
    15,987
    3,427
    www.voipfone.co.uk
    Don't you dare downgrade me; 'cos if you do I'm going to downgrade you and tell all my friends to do the same.

    <discuss>
     
    Upvote 0
    What about the following scenario?

    You offer a recommendation service. To join, someone must pay $10/month, say (forgive me using a dollar sign; I don't have pound on this keyboard). Now, as a member, you can recommend people and you can benefit from the recommendations of others.

    Let's say that someone makes a purchase based on your recommendation. At the end of the job, they get to rate the supplier. If the rating is positive, you get $1 for a good recommendation; if the rating is negative, you lose $1 for a bad recommendation.

    Everyone wins:

    1) You win because of the monthly membership fee.
    2) Recommenders win because they can earn money.
    3) Purchasers win because they can make decisions based on recommendations.
    4) Good suppliers win because they get business.

    The only ones who lose are bad suppliers and people who recommend them.

    There would need to be more to it than this, but to obtain genuine recommendations, the recommenders must have some skin in the game.
     
    Upvote 0

    DuaneJackson

    Free Member
    Jul 14, 2005
    8,642
    1,100
    Brighton / London
    it sounds like a good idea. I'm dubious of offering money as an incentive because it then enourages people to list others with a high rating in order to improve their chanced of earning a dollar. money clouds judgement.

    But then, if they do give a high rating to someone that's crap then they'll lose money.

    Certainly a structure worth pondering.

    Anyone else inclined to run with this idea? It is in the public domain now so I'm not bothered at all.
     
    Upvote 0
    If such a system were to be implemented I wouldn't use it. Although I know these aren't your intentions, but the system would appear money grabbing. Recommendations would also have less influence for me as people would be more liable to recommend someone based on profit (if they are optimistic) and less willing based on loss (pessimistic) so the results would be skewed.

    I'm from Scotland too so I'm too tight for $/£10 a month! ;)
     
    Upvote 0

    ImproveSearchListings

    Free Member
    Dec 5, 2006
    704
    35
    There is already a site that does this, but rather than give ratings you actually give reviews.

    It means that anyone that's adding to the site actually has to put a bit of effort in rather than just clicking 'yes i like it' or 'no i don't' or even a percentage.

    In your example it would also makes it very hard for any automated system to play the game for money benefits. Money is definately a bad idea as it encourages underhand tactics in my opinion (for something like this).
     
    Upvote 0
    M

    Mortime Business Software

    What's happening with this idea then? Any developments?

    I think it is an interesting idea which could probably be developed in a number of ways. I can understand the points about subjectivity, collaboration, cheating, etc., but I can't think of many mathematical models that do not have to cope with some contaminants.

    Most businesses play the roles of both supplier and demander. We could form a network of "trust-lines" between them and use Duane's "diminishing trust" idea to try and solve the problem below, and even to arrive at a "trust level" for each member in the system. Here is a diagram (there are notes about the diagram at the bottom of this message):

    http://img15.imgspot.com/?u=/u/06/348/08/TRUST1166190337.GIF

    Suppose we wanted to find out how much A can trust C. There are a number of routes through the diagram from A to C. There are routes with 3 hops and routes with 2 hops. There are two routes with 2 hops, A->B->C which evaluates to 0.85 x 0.67 = 0.57, and A->E->C which evaluates to 0.77 x 0.95 = 0.73. So which trust level do we choose, 0.57 or 0.73? Do we take the average? And why shouldn't the 3-hop routes be used in the calculation of the trust level? For example, the 3-hop route A -> B -> E -> C evaluates to 0.85 x 0.94 x 0.95 = 0.75, which is higher than both of the 2-hop trust levels.

    I think I might pinch your idea if you don't mind Duane, and have a little play around to see if I can get a trust level for each member.

    Dave
    -----------------------------
    Some notes about the diagram.

    Draw an arrow from A to B with arrowhead touching B, and label the arrowhead 0.85. This denotes that A has bought from B, and has placed 0.85 of "trust" in B.

    (All trust values are between 0 and 1 inclusive, i.e. if t is the trust value, then 0 <= t <= 1).

    Do the same for all the other members that A has bought from.

    Notice that some trust lines have an arrowhead at each end. For example, the trust line between A and D is a double headed arrow. This means that A has bought from D and places 0.93 trust in D, and D has bought from A and placed 0.59 trust in A.

    You don't need to use seperate trust-lines to show mutual trust between two members. It will be clearer if you use double arrowheaded lines (as in the diagram) to show mutual trust between two members.
     
    Upvote 0

    Latest Articles