4 ms·

Conveniently, you don't need to simulate the entire universe, just a single human's experience of it. That cuts computing costs tremendously. If a tree falls i

Conveniently, you don't need to simulate the entire universe, just a single human's experience of it. That cuts computing costs tremendously. If a tree falls in the forest and nobody is around to observe it, do you need to simulate it down to the quantum level? Or do you just do coarse-grained simulation, with fine-grained steps reserved to where it causes observable differences? If later somebody comes around to observe the fallen tree in too much detail you can always just simulate that part in more detail, and hide the difference behind what we call "inherent randomness at the quantum level".

rendall 3y ago
Ok, but then why is a computer needed at all? Put a brain in a vat and stimulate it with chemicals and electricity to hallucinate that it is in a consistent, coherent Universe.
- wongarsu 3y ago
  Sure, you could do that. But that sounds mostly useful for psychology or brain research. The number of institutions that would even consider doing that is dwarfed by the number of simcity players. In a theoretical future where simulating a world in sufficient detail is viable on a smallish hardware budget, you are much more likely to be in a simulation than in a research lab.
- vidarh 3y ago
  So, a computer.
breuleux 3y ago
I think this is a much harder problem than you make it seem. A tree no one is looking at, sure, you can take shortcuts, but if I'm not looking at my computer, it still has to run my programs and you cannot approximate its answers. If I, a being in your simulation, was to hypothetically build a simulator of the kind that you use to simulate me, it is obvious that it could not work: if it could approximate the running of itself to a sufficient level of detail, it already would be doing that. So the question is: what else wouldn't work? The closer the technology I build gets to physical optimality, the more it will act as a bottleneck in the simulation. If I build a physically optimal circuit to perform any useful calculation that I can verify, for example, the simulation speed will have to drop below one second per second. What this means is that modern times are probably too expensive to simulate, so you would likely stick to simulating medieval times (we may also have to eschew simulating the brains themselves, if it turns out the biological ones are too close to optimality for what they do). At least that's what we would have to do. You can imagine that the universe that simulates us works differently, but hypothesizing a world that has different laws of physics doesn't really help us to understand ours. It's fun to think about, but nothing more than that.
- rendall 3y ago
  > You can imagine that the universe that simulates us works differently Moreover, you have to make that assumption, or the equivalent assumption that computers can become so advanced that we are the equivalent to their middle ages. My objection to the simulation hypothesis is that, if one must make those assumptions to make it work, why precisely and only those assumptions that make our Universe a "simulation": that we are here because of the intentions of creatures, that "simulation" applies to whatever it is they're doing, that the creatures are human enough to find simulating us useful.
  - breuleux 3y ago
    The probabilistic simulation argument ("we are probably in a simulation") actually requires you to reject that assumption (edit: not reject outright, but you need the assumption that observers of the real and simulated universes cannot tell the difference, so no 6D stuff). People who raise the point that the universe doing the simulation could be very different from ours don't understand what is required for the argument to work. The simulation argument is that if you observe a universe X, and you work out that a real universe X will simulate two universes just like it, then someone who observes that they are in a universe that looks like X is probably in a simulation. That is because for every real X you have two simulated X: even if you assume that there exists a truly real and physical X, it still follows that most observers of X are simulated. I don't think our universe will simulate more observers than there are real observers, but at least this is something we can meaningfully argue about. If, in contrast, you say that there is a universe Y, and Y simulates X, there are no interesting conclusions you can draw from that. You are comparing "X is real" to "Y is real, and Y simulates X", but there's no reason to think Y is more likely to be real than X. It's possible, but a lot of things are possible. It's not interesting.
    - rendall 3y ago
      If I understand the argument, it seems that case is even more unlikely. Not only must one assume that the composition of our Universe just happens to be a super version of one of our own latest technological innovations, one must also take on faith that a computer could one day spin off multiple instantiations of entire Universes that are indistinguishable from the Universe the computer itself is within. Personally, I don't find the argument compelling because of its unexamined, embedded assumptions. What is meant by computer? What is meant by a simulation? What is meant by Universe? What is an observer? How is it falsifiable? By the time these terms are pinned down, the argument disappears in ontological haze.

rendall 3y ago

Ok, but then why is a computer needed at all? Put a brain in a vat and stimulate it with chemicals and electricity to hallucinate that it is in a consistent, coherent Universe.

wongarsu 3y ago
Sure, you could do that. But that sounds mostly useful for psychology or brain research. The number of institutions that would even consider doing that is dwarfed by the number of simcity players. In a theoretical future where simulating a world in sufficient detail is viable on a smallish hardware budget, you are much more likely to be in a simulation than in a research lab.
vidarh 3y ago
So, a computer.

breuleux 3y ago

I think this is a much harder problem than you make it seem. A tree no one is looking at, sure, you can take shortcuts, but if I'm not looking at my computer, it still has to run my programs and you cannot approximate its answers. If I, a being in your simulation, was to hypothetically build a simulator of the kind that you use to simulate me, it is obvious that it could not work: if it could approximate the running of itself to a sufficient level of detail, it already would be doing that. So the question is: what else wouldn't work? The closer the technology I build gets to physical optimality, the more it will act as a bottleneck in the simulation. If I build a physically optimal circuit to perform any useful calculation that I can verify, for example, the simulation speed will have to drop below one second per second. What this means is that modern times are probably too expensive to simulate, so you would likely stick to simulating medieval times (we may also have to eschew simulating the brains themselves, if it turns out the biological ones are too close to optimality for what they do). At least that's what we would have to do. You can imagine that the universe that simulates us works differently, but hypothesizing a world that has different laws of physics doesn't really help us to understand ours. It's fun to think about, but nothing more than that.

rendall 3y ago
> You can imagine that the universe that simulates us works differently Moreover, you have to make that assumption, or the equivalent assumption that computers can become so advanced that we are the equivalent to their middle ages. My objection to the simulation hypothesis is that, if one must make those assumptions to make it work, why precisely and only those assumptions that make our Universe a "simulation": that we are here because of the intentions of creatures, that "simulation" applies to whatever it is they're doing, that the creatures are human enough to find simulating us useful.
- breuleux 3y ago
  The probabilistic simulation argument ("we are probably in a simulation") actually requires you to reject that assumption (edit: not reject outright, but you need the assumption that observers of the real and simulated universes cannot tell the difference, so no 6D stuff). People who raise the point that the universe doing the simulation could be very different from ours don't understand what is required for the argument to work. The simulation argument is that if you observe a universe X, and you work out that a real universe X will simulate two universes just like it, then someone who observes that they are in a universe that looks like X is probably in a simulation. That is because for every real X you have two simulated X: even if you assume that there exists a truly real and physical X, it still follows that most observers of X are simulated. I don't think our universe will simulate more observers than there are real observers, but at least this is something we can meaningfully argue about. If, in contrast, you say that there is a universe Y, and Y simulates X, there are no interesting conclusions you can draw from that. You are comparing "X is real" to "Y is real, and Y simulates X", but there's no reason to think Y is more likely to be real than X. It's possible, but a lot of things are possible. It's not interesting.
  - rendall 3y ago
    If I understand the argument, it seems that case is even more unlikely. Not only must one assume that the composition of our Universe just happens to be a super version of one of our own latest technological innovations, one must also take on faith that a computer could one day spin off multiple instantiations of entire Universes that are indistinguishable from the Universe the computer itself is within. Personally, I don't find the argument compelling because of its unexamined, embedded assumptions. What is meant by computer? What is meant by a simulation? What is meant by Universe? What is an observer? How is it falsifiable? By the time these terms are pinned down, the argument disappears in ontological haze.