Has centralised quality control for taxi drivers worked somewhere?

Our "grand old" taxi company in my town who advertises for being the only reliable option with professional drivers failed on me five times on a row. On successive rides I got a standard neo-nazi lecture about immigrants, my Visa credit card was refused apparently for transaction costs, two of my drivers got lost and one tried to drive to my destination using mostly sidewalks for driving on.

I sent feedback each time to only receive a generic "we are sorry, we have failed our quality controls and this will never happen again" copy-pasted message. Maybe it's more straightforward to advertise than getting rid of drivers who can't behave.

With Uber I know my bad ranking (I have always rated my drivers 5 stars, so far) has at least some effect on the misbehaving driver.

It's not like Black cabs in London are guaranteed to be safe: https://en.wikipedia.org/wiki/John_Worboys

I think the driver ratings could be largely derived from app data. How close to pickup do they get on average? How close to drop off do they get on average? How erratic is their driving? Do they speed a lot? Other traffic violations? A lot can be gleaned from the GPS and other data. The only remaining factor would be the human side of things: are they nice, is their car nice to sit in?

You really want to rely on a single third party instead of a curation by democratic process?

> I don’t want to rate my driver. I want to be able to rely on a third party...

You sure can’t rely on the Uber, Lyft, Juno ratings. It’s 5 stars or bust. The social pressure on 5 stars is enormous.

Netflix moved to thumbs up, thumbs down. YouTube did the same, after showing a graph of the 5s and 1s:

https://techcrunch.com/2009/09/22/youtube-comes-to-a-5-star-...

I relentlessly give an average delivery or ride 3 stars, but feel bad every time. When the ride is quite good, 4 stars, and exceptional, 5 stars. Exceptional is the exception.

Three stars doesn’t make you a bad rider or a bad driver, just average. If it’s not the bulk of the ratings you give, you’re an unreliable rater and not helping the ratings anyway.