I was not surprised by this, because I know many Catholics honestly try to be rational… of course only within the limits given by the Church.
They would have absolutely no problem with Bayesian updating; the only problem would be the Solomonoff prior. If you replace it by “the Catholic Church is always right” prior, you are free to update rationally on everything else and remain a good Catholic.
This is why Catholics don’t have a problem to accept e.g. evolution, as long as someone can provide an explanation how evolution can be compatible with “the Catholic Church is always right”. (A possible explanation could be e.g. that God created the first life forms; that evolution is a consequence of physical laws created by God, therefore any result of evolution is still indirectly created by God; and that humans are somehow an exception to this process, because even if their bodies are a result of evolution, they also have an immaterial soul created directly by God.)
I don’t believe theists would have any problem with Solomonoff prior. Some ten-state two-symbol machine with a blank tape can be a God for all we could ever know, and then it could create us within it’s machine and do what ever it wants (and the souls could be just the indices it keeps on us).
You know who actually has problem with Solomonoff prior? People who understand it.
We’d want to know what proportion of ten-state two-symbol machines with blank tapes turned out to be gods.
Do we? I would think we would want to know what proportion of universes are created by ten-state two-symbol machines that are gods as opposed to ten-state two-symbol machines that are not gods.
Isn’t this the wrong question? We’d want to know what proportion of ten-state two-symbol machines with blank tapes turned out to be gods.
One short god will suffice if laws of physics require substantially larger program. And for all we know they do.
edit: Also, there’s only what, 20^10 = about 10 trillion possible ten state two symbol machines? Maybe 9 old British billions after you eliminate non-universal machines. That’s less than the data in physical constants we haven’t derived.
I was not surprised by this, because I know many Catholics honestly try to be rational… of course only within the limits given by the Church.
They would have absolutely no problem with Bayesian updating; the only problem would be the Solomonoff prior. If you replace it by “the Catholic Church is always right” prior, you are free to update rationally on everything else and remain a good Catholic.
This is why Catholics don’t have a problem to accept e.g. evolution, as long as someone can provide an explanation how evolution can be compatible with “the Catholic Church is always right”. (A possible explanation could be e.g. that God created the first life forms; that evolution is a consequence of physical laws created by God, therefore any result of evolution is still indirectly created by God; and that humans are somehow an exception to this process, because even if their bodies are a result of evolution, they also have an immaterial soul created directly by God.)
I don’t believe theists would have any problem with Solomonoff prior. Some ten-state two-symbol machine with a blank tape can be a God for all we could ever know, and then it could create us within it’s machine and do what ever it wants (and the souls could be just the indices it keeps on us).
You know who actually has problem with Solomonoff prior? People who understand it.
Isn’t this the wrong question? We’d want to know what proportion of ten-state two-symbol machines with blank tapes turned out to be gods.
In so far as that’s what we want, Catholicism still falls to being a huge conjunction of propositions.
Do we? I would think we would want to know what proportion of universes are created by ten-state two-symbol machines that are gods as opposed to ten-state two-symbol machines that are not gods.
That was implied by “proportion”.
One short god will suffice if laws of physics require substantially larger program. And for all we know they do.
edit: Also, there’s only what, 20^10 = about 10 trillion possible ten state two symbol machines? Maybe 9 old British billions after you eliminate non-universal machines. That’s less than the data in physical constants we haven’t derived.