How hard would this be? I tried to come up with a lower bound for how many distinct human brains there are. This was an act of great hubris. Confidence is 0.6 that I’ve given a true lower bound. 8.6*10^10 neurons in the brain. Each neuron is connected to between 1 and tens-of-thousands of neurons. So I represent the brain is represented as a 100-regular undirected graph on 300 neurons (lower bound!) . If there are N 100-regular graphs on n, then how many such on n+1? I (n-choose-100)*N (I think?). This gives ~10^109829 possibilities. Google tells me there are roughly 10^80 atoms. if the period of simulation lasts 1 second, and we can simulate 10^80 brains simultaneously, it would take ~10^109742 years to simulate them all. Google says heat death is roughly 10^1000 years away, so we would have completed almost 1⁄100 of them by then.
The 100-regular graph is a poor model for this purpose because it includes all such graphs, and some neurons just shouldn’t be connected in any sensible brain. But given that I’ve under-sized the brain model by 8 orders of magnitude, and given us the ability to simulate a brain using an atom, and assumed the simulation hardware functions even into heat death, I think this is a convincing lower bound.
An event like this occurs in Charles Stross’s Accelerando. Great book!
it would take ~10^109742 years to simulate them all. Google says heat death is roughly 10^1000 years away, so we would have completed almost 1⁄100 of them by then
How hard would this be? I tried to come up with a lower bound for how many distinct human brains there are. This was an act of great hubris. Confidence is 0.6 that I’ve given a true lower bound. 8.6*10^10 neurons in the brain. Each neuron is connected to between 1 and tens-of-thousands of neurons. So I represent the brain is represented as a 100-regular undirected graph on 300 neurons (lower bound!) . If there are N 100-regular graphs on n, then how many such on n+1? I (n-choose-100)*N (I think?). This gives ~10^109829 possibilities. Google tells me there are roughly 10^80 atoms. if the period of simulation lasts 1 second, and we can simulate 10^80 brains simultaneously, it would take ~10^109742 years to simulate them all. Google says heat death is roughly 10^1000 years away, so we would have completed almost 1⁄100 of them by then.
The 100-regular graph is a poor model for this purpose because it includes all such graphs, and some neurons just shouldn’t be connected in any sensible brain. But given that I’ve under-sized the brain model by 8 orders of magnitude, and given us the ability to simulate a brain using an atom, and assumed the simulation hardware functions even into heat death, I think this is a convincing lower bound.
An event like this occurs in Charles Stross’s Accelerando. Great book!
10^1000 x 100 = 10^1002