(warning brain dump most of which probably not new to the thinking on LW) I think most people who take the Tegmark level 4 universe seriously (or any of the preexisting similar ideas) get there by something like the following argument: Suppose we had a complete mathematical description of the universe, then exactly what more could there be to make the thing real (Hawking’s fire into the equations).
Here is the line of thinking that got me to buy into it. If we ran a computer simulation, watched the results on a monitor, and saw a person behaving just like us, then it would be easy for me to interpret their world and their mind etc. as real (even if I could never experience it viscerally living outside the simulation). However, if we are willing to call one simulation real, then we get into the slippery slope problem which I have no idea how to avoid whereby any physical phenomena implementing any program from the perspective of any universal Turing machine must really exist. So it seems to me if we believe some simulation is real there is no obvious barrier to believing every (computable) universe exists. As for whether we stop at computable universes or include more of mathematics, I am not sure anything we would call conscious could tell the difference, so perhaps it makes no difference.
(Resulting beliefs + aside on decision theory) I believe in a Level 4 Tegmark with no reality fluid measure (as I have yet to see a convincing argument for one) a la http://lesswrong.com/r/discussion/lw/jn2/preferences_without_existence/ . Moreover, I don’t think there is any “correct” decision theory that captures what we should be doing. All we can do is pick the one that feels right with regard to our biological programming. Which future entities are us, how many copies of us will there be, and who should I care about etc. are all flaky concepts at best. Of course, my brain won’t buy into the idea I should jump off a bridge or touch a hot stove, but I think it is unplausable that this will follow from any objective optimization principle. Nature didn’t need a decision theory to decide if it is a good idea to walk into a teleporter machine if two of us walk out the other side. We have our built in shabby biological decision theory, we can innovate on it theoretically, but there is no objective sense in which some particular decision theory will be the right one for us.
However, if we are willing to call one simulation real, then we get into the slippery slope problem which I have no idea how to avoid whereby any physical phenomena implementing any program from the perspective of any universal Turing machine must really exist.
My approach is that everything is equally real, just not everything is equally useful. In a meta level, talking about what’s more real is not useful outside a specific setting. Unicorns are real in MLP, cars are real in the world we perceive, electrons are real in Quantum Electrodynamics, virtual particles are real in Feynman diagrams, agents are real in decision theories, etc.
(warning brain dump most of which probably not new to the thinking on LW) I think most people who take the Tegmark level 4 universe seriously (or any of the preexisting similar ideas) get there by something like the following argument: Suppose we had a complete mathematical description of the universe, then exactly what more could there be to make the thing real (Hawking’s fire into the equations).
Here is the line of thinking that got me to buy into it. If we ran a computer simulation, watched the results on a monitor, and saw a person behaving just like us, then it would be easy for me to interpret their world and their mind etc. as real (even if I could never experience it viscerally living outside the simulation). However, if we are willing to call one simulation real, then we get into the slippery slope problem which I have no idea how to avoid whereby any physical phenomena implementing any program from the perspective of any universal Turing machine must really exist. So it seems to me if we believe some simulation is real there is no obvious barrier to believing every (computable) universe exists. As for whether we stop at computable universes or include more of mathematics, I am not sure anything we would call conscious could tell the difference, so perhaps it makes no difference.
(Resulting beliefs + aside on decision theory) I believe in a Level 4 Tegmark with no reality fluid measure (as I have yet to see a convincing argument for one) a la http://lesswrong.com/r/discussion/lw/jn2/preferences_without_existence/ . Moreover, I don’t think there is any “correct” decision theory that captures what we should be doing. All we can do is pick the one that feels right with regard to our biological programming. Which future entities are us, how many copies of us will there be, and who should I care about etc. are all flaky concepts at best. Of course, my brain won’t buy into the idea I should jump off a bridge or touch a hot stove, but I think it is unplausable that this will follow from any objective optimization principle. Nature didn’t need a decision theory to decide if it is a good idea to walk into a teleporter machine if two of us walk out the other side. We have our built in shabby biological decision theory, we can innovate on it theoretically, but there is no objective sense in which some particular decision theory will be the right one for us.
Can you expand on this a bit?
My approach is that everything is equally real, just not everything is equally useful. In a meta level, talking about what’s more real is not useful outside a specific setting. Unicorns are real in MLP, cars are real in the world we perceive, electrons are real in Quantum Electrodynamics, virtual particles are real in Feynman diagrams, agents are real in decision theories, etc.