Tomanr
From espians
|
This is a project proposal. The project proposals page is out-of-date, but the projects proposed may still be of some interest.
Tomanr is a reviews site initially concentrating on reviews of services, like plumbers. It allows users to rank their personal trust levels of friends and acquaintances, so that a personalised trust level can be displayed for each review. It even displays the trust levels of anonymous reviews!
Scenario
Say someone wants a plumber. They obviously want someone local but not a cowboy. They can look in the Yellow Pages, but that won't assure them that they're not a cowboy. The best way would be to find multiple independent reviews of the plumber.
The problem about finding reviews of plumbers and the like is that, apart from the fact that there isn't really a site for them to exist on, you can't really be sure that the people involved have no vested interest. Rather like some of the decidedly dodgy tech product reviews that you get across the web... the reviewers might be from the company, or given a free unit to try, or whatever.
If someone wants to post a review about a plumber, they've also got the problem that they might incur some wrath from said plumber if they're writing a bad review about them. so there's a problem there: a) the reviewer can only give a good review if they do it anonymously, but b) reviews are only useful if you can verify their credibility; i.e. if they're not anonymous!
Some site that allows people to post anonymously but in such a way that everybody was networked in providing a secret "I trust this person this much" ballot would be what I envisage for a least-working tomanr.com
How it Works
The cool thing is that the working effect of the site would be this: all reviews would be public and anyone could see them, but when you're logged into the site, even if it's an anonymous post you'd have a ratings thing up saying how much you should trust this given your trust network.
Possible ways to exploit this include the plumber rating his clients very high directly one by one to see how that affects the rating of some bad post about him. Not sure how to get around that exactly, unless there were some jitter in the site, or some bidirectional trust mapping were to take place. The bidirectionality would probably be sufficient.
FAQ
Q: How about the review being done by a peer, i.e. another plumber who can professionally review the work done?
A: But then it's not grounded in your personal levels of trust: you don't know that the plumber that rates his peer highly isn't a friend of that plumber. You don't know anything about it, really; and there are going to be more clients of plumbers really and willing to rate the plumber's performance than plumbers rating other plumbers!
You get the flip side too: by rating all the other plumbers down, you increase your own standing. We're just modelling how trust works in real life, but expanding it a little through a) the internet network effect, and b) anonymisation.
If I'm going to use any plumber, I'm going to use one that my cousin brought in and said did a good job, right? Much more likely to do that than to ring up a different plumber in the Yellow Pages and ask them what they think of the plumber that I'm thinking of getting! That makes no sense whatsoever.
Q: Surely one drawback of the model is that it needs a heck of a lot of anonymous networking before it can really scale up properly?
A: Actually that's something that's emergent. In other words, most reviews would be named, and they're still trivially useful even without the trust network. You have three layers: 1) raw reviews of services, 2) a trust network showing how much you can trust those reviews (iff you trust our trust metric!), and 3) anonymity showing a benefit of the trust network that you can't get without this technology; the novel factor.
3 depends on 2 which depends on 1; but 1 works without 2 which works without 3; and in terms of value: 1 < 2 < 3. So it can start out as just a well engineered reviews site. then you bolt trust on, and then you bolt anonymity on. it's pretty easy to make a trivial reviews site anyway; but making a well engineered one would be quite a task.
Summary
What I like about it is that it's a boring idea with a lot of actual potential and one that it made less boring by working on it with lots of intelligent friends. Ripe for 24 Weeks!
Further Reading
- IRC chat about tomanr.com, 11th June 2007
- Toman.