Having just reread your objections to uploading without reverse engineering I think it merits a response more detailed than the one I am about to give. It may be correct or at least have some room for middle ground where a lot of the short timescale/easy stuff is directly simulated and then corrections are spaghetti coded added on to prevent particular failures with data from real experiments.
That said, my (limited) experience with trying to reverse engineer what is going on in a mouse’s brain during social interaction makes me feel utterly hopeless and everyday I dream about how much easier it would be if we could do a barebones direct simulation like the fruit fly simulation to see if we are even on the right track. Because of this (again, quite limited) experience trying to do something like the reverse engineering you suggest, I expect it to take ~forever whereas disentangling the mess that is all the higher order corrections past simple electrical models of cells connected with one way chemical synapses would merely take a really, really long time.
Also, I think doing the kind of reverse engineering on humans is challenging for purely ethical reasons whereas sufficiently detailed models for neuron/other components from mice would just carry over to human WBE much better than a fully reverse engineered mouse. I may be misunderstanding what depth you feel reverse engineering is necessary and what experiments it would require.
oh oops sorry if I already shared that with you, I forgot, didn’t mean to spam.
My actual expectation is that WBE just ain’t gonna happen at all (at least not before ASI), for better or worse. I think the without-reverse-engineering path is impossible, and the with-reverse-engineering path would be possible given infinite time, but would incidentally involve figuring out how to make ASI way before the project is done, and that recipe would leak (or they would try it themselves). Or even more realistically, someone else on Earth would invent ASI first, via an unrelated effort. So I spend very little time thinking about WBE.
You left a comment one a previous post of mine about this but that was almost a year ago I think, so hardly spam.
My perfect very wishful thinking world involves ASI miraculously not happening and normal human neuroscience efforts shifting toward uploading and away from what it is now, which is a lot of wheel spinning and performative science. I do not assign a high probability to either of these. I also feel I am not well informed enough on either to make such sweeping claims.
then corrections are spaghetti coded added on to prevent particular failures with data from real experiments
My guess would be that the failures would be quite systematic, and would reflect the absence of substantial algorithms. That would suggest that you either have to come up with more algorithms, and/or you have to learn them from data. But to learn them from data without coming up with the algorithms or with algorithmic search spaces that sufficiently promote the relevant pieces, you need a lot of data; and brain algorithms that work on a time scale of an hour or a day have correspondingly 104 or 105 less data feasibly available compared to ~second-long events.
I’m skeptical of WBE happening for reasons discussed here but still enthusiastic about the connectomics stuff for other reasons discussed here.
RE Dale’s Law, I recall reading that almost every neuron in the brain detects and produces one or more neuropeptides.
Having just reread your objections to uploading without reverse engineering I think it merits a response more detailed than the one I am about to give. It may be correct or at least have some room for middle ground where a lot of the short timescale/easy stuff is directly simulated and then corrections are
spaghetti codedadded on to prevent particular failures with data from real experiments.That said, my (limited) experience with trying to reverse engineer what is going on in a mouse’s brain during social interaction makes me feel utterly hopeless and everyday I dream about how much easier it would be if we could do a barebones direct simulation like the fruit fly simulation to see if we are even on the right track. Because of this (again, quite limited) experience trying to do something like the reverse engineering you suggest, I expect it to take ~forever whereas disentangling the mess that is all the higher order corrections past simple electrical models of cells connected with one way chemical synapses would merely take a really, really long time.
Also, I think doing the kind of reverse engineering on humans is challenging for purely ethical reasons whereas sufficiently detailed models for neuron/other components from mice would just carry over to human WBE much better than a fully reverse engineered mouse. I may be misunderstanding what depth you feel reverse engineering is necessary and what experiments it would require.
oh oops sorry if I already shared that with you, I forgot, didn’t mean to spam.
My actual expectation is that WBE just ain’t gonna happen at all (at least not before ASI), for better or worse. I think the without-reverse-engineering path is impossible, and the with-reverse-engineering path would be possible given infinite time, but would incidentally involve figuring out how to make ASI way before the project is done, and that recipe would leak (or they would try it themselves). Or even more realistically, someone else on Earth would invent ASI first, via an unrelated effort. So I spend very little time thinking about WBE.
You left a comment one a previous post of mine about this but that was almost a year ago I think, so hardly spam.
My perfect very wishful thinking world involves ASI miraculously not happening and normal human neuroscience efforts shifting toward uploading and away from what it is now, which is a lot of wheel spinning and performative science. I do not assign a high probability to either of these. I also feel I am not well informed enough on either to make such sweeping claims.
My guess would be that the failures would be quite systematic, and would reflect the absence of substantial algorithms. That would suggest that you either have to come up with more algorithms, and/or you have to learn them from data. But to learn them from data without coming up with the algorithms or with algorithmic search spaces that sufficiently promote the relevant pieces, you need a lot of data; and brain algorithms that work on a time scale of an hour or a day have correspondingly 104 or 105 less data feasibly available compared to ~second-long events.