Certainly parts of this have already been done, or attempted or written about, but I haven't seen it expressed as a whole like this.

Reputation management online works for some, but generally within defined boundaries - e.g. you have a reputation on the eBay website, but this is not transferable to other sites. This idea is to have a portable reputation which can be used across all websites to give others an idea of whether or not to trust you. The problem of what identity the reputation should be attached to is not easy. There's no 'strong' way of proving identity online and people's preference for multiple online identities works against the idea of a portable reputation. To make it easier, I shall assume the reputation is attached to your primary email address. Then, if this email address is used to initiate separate identities on various sites, that's OK, as your reputation will follow you.

How will your reputation be communicated? The idea is that if someone you trust trusts your counterparty, then you should trust them too. So, there will need to be a central 'reputation broker' where you can identify all the people and organisations you trust. Equally, organisations and people will state that they trust you (e.g. your employer might trust you - while you're working for them).

It would then be possible to gain a measure of the trustworthiness of someone on the internet - you might be told that they are, for example trusted by one person you trust, and 3 people who are trusted by an organisation you trust, and one organisation you don't trust.

Possibly more useful than this though is the potential for extending the model to judging the trustworthiness of websites. Most banks are well aware of the phishing websites which are trying to defraud their customers, they're just powerless to shut them down. But if your browser automatically looked up the trust links for a new website you're visiting, this would all be different: You would get a message like "This website (www.yourbank.ru) is not trusted by 2 organisations you trust (YourBank and UK Government) and is not trusted by 3 people you trust".
-- hippo, Oct 23 2009

Identity 2.0 http://identity20.com/media/OSCON2005/
Nice talk in the vein of the idea. [Jinbish, Oct 23 2009]

Trusted Thid Party http://en.wikipedia...Trusted_third_party
An idea often used in cryptography and security circles. I have seen reputation management proposed using such trusted third parties. (I've also seen peer-to-peer style distributed authentication -rather than 'trust ' - systems. [Jinbish, Oct 23 2009]

I think, Mr Hippo - that there are two things going on here - one, the nature of identity, and two - the real nature of trust - and really, how the two things are actually the same thing - and how top-down approaches must fail, natural ones (like this) must emerge, and really, we can only ever ultimately be recognised and vetted by clues stirred up by our activity and relationships, rather than being a fixed, known record that sits in a file in a government building somewhere.
-- zen_tom, Oct 23 2009


//I've always said that everybody should unanimously 'dirty up' their profile// - good idea. You go first.
-- hippo, Oct 23 2009


Didn't we have this discussion before? I distinctly remember writing something much like this where the relationships are public but the levels of trust between relationships are hidden. That way there's no pressure on one party to artificially inflate their evaluation of another person.

Any given evaluation of a person or relationship could be established by distance (number of intervening relationships), value (either the average of the levels of relationships or the single highest value) or by the number of paths between you and the other person - or by any combination thereof.

A single person's evaluations of other people could be weighted by how closely they match other evaluations of others.

I just know I wouldn't want to write it.
-- phoenix, Oct 23 2009



random, halfbakery