Because when the rated party also rates the rater, ratings are essentially useless. The question you ask yourself is: is this person so absolutely awful as a passenger/driver that I'm willing to risk being kicked off the platform to get them kicked off?
You can’t see how the other person rated you before you rate them in Uber. Also this would explain why you’d always rate high, not choose two stars over one which is still a bad rating.
Airbnb has this problem though. A loud apartment on the highway will have no mention of it being on the loud highway. Just ratings like “5 stars, great for early risers!”
Airbnb has no motive to fix it because their service looks great with rating inflation.
If there’s a problem with Uber’s rating system, it’s that it’s a five-star system.
I thought I was the perfect Uber passenger (always on time, sat quietly) until I saw my rating. It was 3.5/5. Too arbitrary to be actionable. Is it because I’m a gringo living in Mexico? Should I start small talk more often? Tip them in cash? Is it because I only use it to go to the airport which is a long and traffic-ridden drive which must not be fun for the driver? Or is 3.5 actually a good rating because it’s above 3.0, etc.
> Airbnb has no motive to fix it because their service looks great with rating inflation.
They have no motive to fix it if they have an incredibly short term view of things. It's fairly obvious that this damages the entire platform over time.
Framing a stupid decision as rational because it's rational if you squint enough doesn't ring true. It's rational to think ahead a bit and take into account secondary and tertiary effects.
> If there’s a problem with Uber’s rating system, it’s that it’s a five-star system.
"Single axis" ratings systems are unintuitive, unreliable, prone to "gaming", and thus often self-defeating.
I think person-to-person services should have at least 2 rating axes:
• Was the service rendered [as advertised]? Yes / No
This would be for whether the driver dropped you off at the destination, or the seller delivered the advertised item, or the room and facilities matched the description.
A simple confirmation that the contract was completed; emotions should not enter into this rating.
• How do you feel? Disappointed / Neutral / Pleased
This must be presented in such a way that users DO NOT ASSUME that "Neutral" == "Bad." It should be OK to rate Neutral, because that's how we feel about most of the things in life.
I believe this was the original intent of the 5 star systems to begin with: 3 would be neutral/average/okay, 4 would be above average, and 5 would be exceptional/rare.
Or, since they are STARS, even 1 would be good.
But in practice, our collective psychology thinks anything <5 is bad.
Maybe introduce tomatoes or poo in addition to stars. Stars would be how good it was, and poo would be how many things annoyed you.
Why? In a situation where repeat encounters are rare, and each party cannot see the other party's rating until they've submitted theirs (or not at all), what's the risk of giving an accurately-bad rating? Also, how is rating 2 stars less likely to trigger retaliation than rating 1 star?
