Different customers like different things. Netflix doesn’t get the average of customer ratings and show me that average. Netflix attempts to predict my rating for a movie and shows me that prediction. Some customers of Uber/Lyft prefer more talkative drivers, drivers who prefer the AC on different settings, drivers who like talking about different things, like different cars, etc. I believe that predictions could be done with some fairly simple analysis of the data you have. With predicted ratings, you can try to match riders and drivers better. This could be even more helpful in the case of products like uber pool.
What is the use of so much micro optimization. I just want to reach my destination ASAP. Time and cost are the only things I would want to optimize
This is not a micro optimization...
Want to wait an extra 5 mins to find a chatty driver? Didn’t think so
I'd definitely be into it if I had a nice UI to adjust preference sliders and be able to estimate trip wait time in advance based on preferences
Wouldn’t be surprised if they’re working on that at the moment