I wonder what the acceptable collisions/delivery needs to be for it to match last mile truck safety level (ie UPS trucks are big and run into things with non-zero frequency)
Throwing up hands and saying: internalizing the externalized cost is "ridiculously expensive" is not proof it doesn't work.
The examples of the a la carte exercise brands referenced (SoulCycle, etc) are quite ineffective arguments -- those are successful businesses with loyal, high retention users because they provide specific, high value products to the users.
On this topic (parallel postulate), it took ~2000 years from Euclid and then 3 people all came to the same conclusion independently within ~10-20 years.
"Lobachevsky [mathematician contemporary of Gauss, who claimed parallel postulate was unnecessary] was relentlessly criticized, mocked, and rejected by the academic world. His new “imaginary” geometry represented the “shamelessness of false new inventions”"
Further, many claimed premature success in finding logical contridictions in geometry lacking parallel (Euclid's 5th) postulate; which meant they believed a 4-postulate geometry to be fundamentally false.
Non-Euclidean geometry (geometric axioms in which one postulate is rejected such that the 3 angles of a triangle are not exactly 180 degrees) was considered a meaningless word game and fundamental mistruth.
Later, non-Euclidean geometry was actually essential to modern physics.
It's intellectually sketchy to judge future value by the present.
The legit criticism with a legit recommended change is even better.
A time and technological gap always exists between theory and a plan for experimental confirmation. Some gaps are fairly short. String theory's gap is undoubtedly long, not for lack of resources.
This gap justifies tapering the allocation of attention and research resources (funding, students, etc), which got lopsided following the strong marketing campaign driven by Greene.
It's ~$1/person direct, but how much indirect cost? Things become inexpensive when automation and technological progress makes them so, and the code copyright and restriction is a major barrier to do that.
The bigger issue is software development, imo. eg the IT team at a hospital or the medical office can't build a tool to help coding efficiency or accuracy. Further, public resources/websites with code-related information get DMCA takedowns. It's a stranglehold on innovation/progress.
The American Medical Association owns copyright to all the codes and their descriptions. They have an extremely restrictive and expensive licensing options and they strictly forbid training models with the codes.
The AMA (a nonprofit!) clears ~$300M/year revenue from the codes, which is the direct cost passed through to consumers, but the indirect costs are the byzantine nightmare of OP.