[Wikipedia-l] Re: Webs of trust are a better idea (was: trust metrics)

[Michael Snow]

Jimmy Wales wrote:

> Daniel Mayer wrote:
>
>>That would prevent any incentive to create sock puppets since my
>>selections only affect what *I* see and what the people who trust my
>>judgment see (if they set their preferences accordingly).
>>    
>>
>
>I think that's really fascinating.  The incentive only arises if the
>web of trust is "summed up" across different people to arrive at an
>overall "score".
>
>As long as we don't do that, there's no incentive for sock puppetry.
>
>So, hmm, why did I want to do it that way in the first place?  Well, a
>"summed up" score could be really handy for certain types of decision
>making.  It could provide people with feedback on their overall
>behavior.
>
>But the real point is just to find a way for us to scale better as the
>number of editors grows, to ensure that newcomers are assisted, that
>vandalism is properly watched for, etc.
>
>I like your idea a lot.
>
>--Jimbo
>
At the risk of me-too-ism, I think mav's web of trust concept at least 
avoids most of the dangers I see in a feedback-based reputation system. 
That alone is wonderful progress. I'm a little more skeptical about how 
widely the concept would be adopted, but I'm also not that much of a 
Recent Changes junkie, so maybe I'm missing the appeal. Anyway, the 
web-of-trust system wouldn't have to be that widespread to be useful.

However, I would still avoid the pitfalls of generating a reputation 
score for individual users, even using this framework. Better just to 
let user X know that user Y trusts or distrusts X.

--Michael Snow
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.wikimedia.org/pipermail/wikipedia-l/attachments/20040216/b4326116/attachment.htm