This sounds very much like the phenomenon described in From Personal to Prison Gangs: Enforcing Prosocial Behavior, where the main reason for regulation/getting the government to step in has become more and more common is basically the fact that at scales larger than 150-300 people, we lose the ability to iterate games, which in the absence of acausal/logical/algorithmic decision theories like FDT and UDT, basically mean that the optimal outcome is to defect, so you can no longer assume cooperation/small sacrifices from people in general, and coordination in the modern world is a very taut constraint, so any solution has very high value.
(This also has a tie-in to decision theory: At the large scale, CDT predominates, but at the very small scale, something like FDT is incentivized through kin selection, though this is only relevant for 4-50 people scales at most, and the big reasons why algorithmic decision theories aren’t used by people very often is because of the original decision theories that were algorithmic like UDT basically required logical omniscience, which people obviously don’t have, and even the more practical algorithmic decision theories require both access to someone’s source code, and also the ability to simulate another agent either perfectly or at least very, very good simulations, which we again don’t have.)
This link is very helpful to illustrate the general phenomenon:
This sounds very much like the phenomenon described in From Personal to Prison Gangs: Enforcing Prosocial Behavior, where the main reason for regulation/getting the government to step in has become more and more common is basically the fact that at scales larger than 150-300 people, we lose the ability to iterate games, which in the absence of acausal/logical/algorithmic decision theories like FDT and UDT, basically mean that the optimal outcome is to defect, so you can no longer assume cooperation/small sacrifices from people in general, and coordination in the modern world is a very taut constraint, so any solution has very high value.
(This also has a tie-in to decision theory: At the large scale, CDT predominates, but at the very small scale, something like FDT is incentivized through kin selection, though this is only relevant for 4-50 people scales at most, and the big reasons why algorithmic decision theories aren’t used by people very often is because of the original decision theories that were algorithmic like UDT basically required logical omniscience, which people obviously don’t have, and even the more practical algorithmic decision theories require both access to someone’s source code, and also the ability to simulate another agent either perfectly or at least very, very good simulations, which we again don’t have.)
This link is very helpful to illustrate the general phenomenon:
https://www.lesswrong.com/posts/sYt3ZCrBq2QAf3rak/from-personal-to-prison-gangs-enforcing-prosocial-behavior