Nelz's Blog

Mah blogginess

Proposal: Social Credit Score

The Background

Yesterday, via Reddit Programming http://programming.reddit.com/ I was brought to a forum discussion on RSS technology.

The original post (by Bruce Eckel) was about RSS and inconsiderate clients, but Bruce ends up commenting on fundamental issues of the Web in general. The root question he seems to be asking is whether or not we always need to adhere to the Web’s implicit anonymity? Doesn’t social discourse become orders of magnitude more civil and helpful (and less spam-y) when comments can be attributed to a known source?

John Cowan then comments on the ineffectiveness of several other identity solutions. (He even included one example of a site trying to verify identity all the way back to a user’s government-issued ID. (But, as Rod Begbie helped me to see at BarCampBlock, even the way our government ID’s are associated to us is broken.)

The Concept

In a community, it is not your ‘identity’ (the correlation of your name to your physical reality) that matters, it is actually your ‘reputation’, or the body of your participations that matters.

Okay. I agree, that’s not a terribly new idea. You can see this idea in execution in many places. As an example, look at the Artima.com forum, you’ll see an accounting of the number of posts from any given contributor right underneath their name. The higher a number associated with a submitter, the more social weight the submitter carries.

My complaint with this system is that all the reputation developed by these people are siloed into disassociated stores. Much like some of the value offered by the OpenID system, wouldn’t users (and our communities) benefit from aggregating these contributions, this ‘social capital’?

The Proposal

Why don’t we, as a community, create an (open-source?) algorithm that will generate a "social credit score"? (Kind of like how Equifax (and others) generates generates a financial credit score.)

This algorithm would have to be very robust and it would have to take many different data inputs. It would have to be able to calculate not only the user’s overall participation quality, but also the quality of input from the involved sites as well. And don’t forget esteem or detractor votes by other ranked users.

Also, this algorithm must not be overly influenced by attackers, neither in a single instances nor at scale.

What other characteristics should this algorithm have?