Google, Reputation, and more Reputation
Victor sent me a link about Google Payments Seller Reputation a while ago, but with all the craziness at the time I wasn’t able to check it out until now.
Looks like they’re going for the standard five stars approach, along with some qualitative comments about the process. Again, the comments aren’t standardized, and I don’t really understand the point behind star rating systems in general. With the reputation management system Didier and I worked on for the Applied Dream, we went with a thumbs-up or thumbs-down to evaluate a transaction, because in the end you’d either engage in another transaction with that person or you wouldn’t.
I think a yes or no is much easier to decipher than a star rating, particularly because the stars don’t actually equate to anything meaningful. I understand what five stars means and one star, but what differentiates a two from a three from a four? It becomes a matter of subjective opinion, and at that point the common basis for understanding the rating degrades. Simply put, because of my particular value system that I attach to those stars, my two-star rating may mean something completely different from your two-star rating or even your three-star rating.
What seems better is to use a binary rating scheme (yes or no) and track trends over time. Several online games use a rating scheme whereby your freshness of results hold greater weight over older results. In other words, your rating automatically degrades if you haven’t logged in or played for a while. Attaining and maintaining a ranking therefore requires that you play frequently in addition to being good.
While this mechanism can be applied in a gaming situation to keep things interesting, it also contains a component of accuracy of results. Simply put, a recent rating is more likely to reflect your behavior now than a rating from a year ago. This behavior has some interesting implications. It means that if you were once bad and are now good, your rating isn’t penalized for that behavior. That history will still be there, but it won’t skew your rating within the value system of recent results holding greater weight. People might still have concerns, just as they would about someone with a criminal record. The difference here is that good, recent behavior outweighs bad behavior of the past. In essence, every moment is an opportunity to raise your rating by doing good. It’s a very positive proposition.
If you haven’t been rated for a while, then your rating degrades back to a baseline. This isn’t bad or good; it just means that you haven’t had any recent rated activities and that there is an element of uncertainty surrounding any future transactions you engage in. This uncertainty might sound bad, but if transaction histories are known, then you can use past actions and trends of actions as a basis for judgement. So maybe your rating is at baseline, but your history shows general positive ratings.
Just as it’s used in games, this “decay” could be used to stimulate activity within the system. If you know that you will have a better rating if you have recent activity, then perhaps this will spur you to engage in activities which helps your rating, thereby producing a positive feedback loop.
The downside to ratings and reputations is that there is always uncertainty. There’s always the chance that someone with a great rating will walk off with your Ferrari, for example. I can just say that this is part of human nature, or, coming from a litigious society, you can sue. Or something to that effect. I’m sure there’s room for things like insurance in these transactions. For one thing, that risk represented by the presence of insurance was covered in the Applied Dream project through a reduction in cost. So riskier individuals (those with a low reputation) pay more, while those with better reputations pay less.
However, in the end I think there aren’t any real mechanisms to put in place which can prevent or insure against certain types of behavior. The real power of reputation management systems is as a support to human relationships. They make the invisible visible, and they help you make better decisions. Like any tool, however, it just reports the facts. It can’t predict human nature, and that’s why you have to rely on things like intuition and social savvy.
Going back to comments as part of the ratings, the qualitative information comments supply can cover the specifics of any reservations you might have regarding the transaction. Here the stock-market analogy comes into play, because I think it’s ultimately the responsibility of the parties involved to do their due diligence and to make the final call. If you invest solely on today’s stock price, you’re probably in for some rude surprises. You have to use the tools available to you to make informed decisions, and I think that’s something which people are familiar with: if you don’t do your homework, you are taking your chances. That’s true of anything in life.
Ok, that’s about it for now. Need to get back to the last touches on the thesis report draft.