Even leaving out the whole “experimentally verified” thing, parts of a system can have longer description lengths than the whole system by breaking symmetries in what you include. Which means that if the system has a description length distributed according to the simplicity prior, such a part doesn’t. So it’s not that you just shouldn’t use the simplicity prior on everything, it’s that you can’t.
If you’re saying our observations can be complex even though the universe is simple, you’re essentially taking the first horn of the dilemma in my post. This means you cannot use Solomonoff induction to figure out the correct quantum physics, because Solomonoff induction works on the input stream of bits that it sees, not on the “universe as a whole”.
This means you cannot use Solomonoff induction to figure out the correct quantum physics, because Solomonoff induction works on the input stream of bits that it sees, not on the “universe as a whole”.
This is a general problem: most decision theories ignore the fact that their implementation runs on physics, or make some nonsensical assumptions like “everything is program”. But observations are not just parameters of a known idea (for agent’s decisions, and in particular epistemic decisions, are not known), they by themselves mean at least the physical facts that constitute them.
I’m not actually very familiar with Solomonoff induction, so if it can’t handle the observable universe being one possibility of many, pretend I’m advocating something that can :)
Even leaving out the whole “experimentally verified” thing, parts of a system can have longer description lengths than the whole system by breaking symmetries in what you include. Which means that if the system has a description length distributed according to the simplicity prior, such a part doesn’t. So it’s not that you just shouldn’t use the simplicity prior on everything, it’s that you can’t.
If you’re saying our observations can be complex even though the universe is simple, you’re essentially taking the first horn of the dilemma in my post. This means you cannot use Solomonoff induction to figure out the correct quantum physics, because Solomonoff induction works on the input stream of bits that it sees, not on the “universe as a whole”.
This is a general problem: most decision theories ignore the fact that their implementation runs on physics, or make some nonsensical assumptions like “everything is program”. But observations are not just parameters of a known idea (for agent’s decisions, and in particular epistemic decisions, are not known), they by themselves mean at least the physical facts that constitute them.
I’m not actually very familiar with Solomonoff induction, so if it can’t handle the observable universe being one possibility of many, pretend I’m advocating something that can :)