I’ve always been fascinated by the way Uber has implemented their market dynamics using surge pricing. Uber isn’t a market for rides in the way that, say, a stock exchange is a market for stocks. There aren’t drivers setting their asking price and riders making bids, nor is there an order book where matches are made at a clearing price, and understandably so - I can’t imagine how to build a comprehensible UX around a “true” bid-ask marketplace.
Instead, Uber presumably has historic estimates of the supply and demand curves at different locations, different times of the week, different passenger / rider populations (business travelers or tourists?) and then uses the measured “true” supply and demand to find a clearing price, and therefore decide whether or not a market is going to surge.
The UX of surge is important too - the raison d’être of surge pricing is to bring more drivers to an undersupplied market. That means that when you detect a supply or demand shock that would lead to surge pricing, you want to increase the surge as quickly as possible to send out the “we need more drivers” signal, because there’s a latency in getting more supply (drivers have to relocate). Conversely, Uber doesn’t want to drop the surge price too quickly - they want downward movement to be sticky - because you don’t want to tell drivers “there’s more money to be made over here” just to renege on that promise before the supply can even get there.
So if surge is sticky on the way down, these drivers may have found a way to exploit the pricing algorithm - simulate a price shock then reap the rewards. If surge were not sticky on the way down, this strategy might be much less effective - a few drivers would get better fares, but the market would return to equilibrium faster.
None of this is to say that you can’t have cartel behavior in a “bid-ask”-style market too, but I suspect this is a “hack” of Uber’s pricing UX as much as anything else.