Uber charges what it thinks you can pay

The exhibit is a Reddit screenshot pair: two phones requesting the same short Manhattan ride across Central Park, same hour, one quoting 76.82 dollars and the other 23 ▸ 26:11, roughly 300,000 versus 88,000 pesos for the same two kilometers. His first instinct, an A/B test, gets corrected on camera: Uber is too mature for that, this is “surveillance pricing” ▸ 27:26.

The mechanism is inferred willingness to pay. The platform reads your habits, your device, even your battery level, and if it decides you can absorb more, it charges more ▸ 28:01; a phone nearly dead reads as a rider who needs the car and can’t shop around ▸ 28:41. The tell that they’ve felt it themselves: at airports they set the pin to the building across the street and watch the fare drop 50,000 pesos ▸ 29:12.

el precio no mide la distancia; mide cuánto cree el algoritmo que aguantas →

Then the move that turns a gripe into a teardown: the asymmetry is two-sided. The driver never sees the rider’s fare, so on the 76 dollar ride he may still be paid 20; the platform charges the rider up and pays the driver down, and “todo esto de acá a la mitad se lo queda Uber” ▸ 33:43. And the reason that margin compounds forever is the nature of the good. The heavy cost was the first two years, the algorithm and the market fit; after that it’s a feature here, a feature there, while revenue climbs, because “el software es replicable, no cuesta nada replicarlo” ▸ 35:26. A chef who makes great burgers can’t be copied; the software that prices your ride can, so they earn millions on something built years ago ▸ 35:31. The sting in the tail, aimed at every builder watching: write that same software as an employee and none of it is yours, “es la empresa” ▸ 36:27…