I'd love to see ProPublic or other investigative journalists do some research here. Especially on the driver side, because they seem to have even less leverage than average riders.
(1) N drivers authorized on the Rideshare platform are engaged.
(2) these N drivers drive simultaneously and in a similar geographic preference- as if they live in the same neighborhood, but also are optimizing as drivers do. And yet drawn home.
(3) pricing, earnings, etc. are recorded per driver and compared across drivers on the Rideshare platform(s).
Now we engage the experimental component. With N drivers we can have N/2 drivers engage in behavior A or behavior B. Then, when the Rideshare platform denies they do this or that depending on that or this, there is good evidence that is not the case. Statistical evidence. Tallyho, bandits!
- positional pricing, certain geocodes command what is called pricing elasticity, aka. people going to the hospital dont care about the price. Time based pricing elasticity is also exploited, for example, people leaving the concert at 12am likely dont have other options
- driver bonuses and rider coupons optimizing for marginal rides. The riders and drivers at the margins of pricing are typically switchers, which mean that if you are loyal to any particular app, you will likely never get see these coupons or bonuses