Dave Wilcox, all round smart guy, has just started his blog¹. One of his posts is about Reputation Technology and it’s prompted me to write something here about how rating things works in society.

I’m going to talk mainly about Uber to begin with because it’s a well known example. When you finish a journey with Uber you need to rate your driver out of 5 before you can book another one. Drivers need to keep an aggregate rating of over 4.6 to stay on the system otherwise they get deactivated (which means fired without notice). If you know this then it limits your rating choices. This system compresses your rating options into 4 levels of punishment, and everything else.

I like that my rating is personal to me², I can rate whatever I want, but if that fifth star encompasses everything from did an OK job to this was the greatest cab ride of my entire life then it feels too restrictive.

I few years ago I read building web reputation systems and there were two things that have stuck in my mind. One was that people really only rate at the extremes (e.g. 1 and 5 stars, hardly any 2-4 star ratings). So telling people that 5* was the baseline, but that they could rate up to 10 would just result in a drift up to 10* being the standard.

It’s free to rate 5 stars, maybe you could make it cost something to rate above that?

What if there was a log or exponential scale of tip that it would cost to rate above 5? E.g. if a 6 rating cost 50c and a 10* rating cost $100 with a rising rate in between³.

That money could then go directly to the driver, overcoming some of the objections to the forced-obsequiousness that has been mentioned elsewhere.