An additional part of the story is that if a superintelligence want to compute the distribution of outcomes of an intelligent civilization doing a singularity, it will want to sample histories leading to the singularity. But these histories have a branching tree structure.
Suppose you’ve already simulated the evolution of life and humanity up to the singularity, and you’re interested in other nearby ways it could have played out. You could simulate the whole thing again, but that’s probably unnecessary. Computationally cheaper, would be to cache the simulation, go back to a specific branching point and restart the simulation (with some trivial noise) from that point. eg you only need to compute the evolution of life and humanity up to WWI, once, in order to simulate all of the possible historical trajectories that could have followed from WWI.
The more time passes, the more branching points there are, and the more branches. (Depending on the parameters), it’s an exponential explosion. It seems like almost all of the total simulation time is spent simulating the period in the decades leading up to the singularity.
Yeah, this talk for instance.
An additional part of the story is that if a superintelligence want to compute the distribution of outcomes of an intelligent civilization doing a singularity, it will want to sample histories leading to the singularity. But these histories have a branching tree structure.
Suppose you’ve already simulated the evolution of life and humanity up to the singularity, and you’re interested in other nearby ways it could have played out. You could simulate the whole thing again, but that’s probably unnecessary. Computationally cheaper, would be to cache the simulation, go back to a specific branching point and restart the simulation (with some trivial noise) from that point. eg you only need to compute the evolution of life and humanity up to WWI, once, in order to simulate all of the possible historical trajectories that could have followed from WWI.
The more time passes, the more branching points there are, and the more branches. (Depending on the parameters), it’s an exponential explosion. It seems like almost all of the total simulation time is spent simulating the period in the decades leading up to the singularity.