I've been interested in online reputation systems ever since reading Cory Doctorow's Down and Out in the Magic Kingdom which featured "Whuffie" the reputation-based currency of the book's post-scarcity world.
We already have silo'd reputation services (likes on Facebook, recommendations on LinkedIn, buyer/seller feedback on eBay) which many of us use to judge someone on a whole range of attributes, from their suitability as an employer/employee to the likelihood that the item they are selling on an auction site lives up to its description.
However, there's a big challenge in combining these disparate sources into one overall "score" that gives others a handle on our reliability outside of those silos. Karma (havekarma.com) tries to do that for a decent range of services, although they are being understandably coy about their weighting system. It'll be interesting to see how and if it grows.
These systems have lots of challenges ahead, including exchange rates (is a LinkedIn recommendation worth more or less than one Facebook like?) and applicability (is being a good seller on eBay really relevant to your suitability as an airbnb host?) but the biggest challenge for a successful reputation measurement system is likely to be balancing the desire for higher numbers of contributing services against quality. After all you wouldn't want the reputation of your reputation system coming under threat because you've granted contribution rights to a reputation-farming service (and I'd expect to see these spring up shortly after a successful system is opened up and an API provided).
If these systems ever catch on I wonder how long it'll be before your score becomes portable as well as aggregated, allowing you to carry it into a service you've never used before, where it becomes the baseline score by which others users judge you?