Even though LW is far more open to the idea of cryonics than other places, the general opinion on this site still seems to be that cryonics is unlikely to succeed (e.g. has a 10% chance of success).
How do LW’ers reconcile this with the belief that mind uploading is possible?
I can’t speak for anyone else, but I don’t see a contradiction. Believing that a living brain’s data can be successfully uploaded eventually doesn’t imply that the same data can necessarily be uploaded from a brain preserved with current-day tech. The usual line I see quoted is that cryonics tech isn’t guaranteed to preserve key data, but it has a higher chance than rot-in-a-box tech or burn-to-ash tech.
The usual line I see quoted is that cryonics tech isn’t guaranteed to preserve key data, but it has a higher chance than rot-in-a-box tech or burn-to-ash tech.
So are you saying that this key data includes delicate fine molecular information, which is why it cannot be preserved with current tech?
Nope, I’m not saying that. There are many systems that both don’t depend on fine molecular information, and also are easier to restore from being vitrified than to restore from being burned to ash.
Certainly shminux’s reply isn’t what I had in mind initially, if that’s what you mean.
As for whether I agree with it on its own terms… I’m not sure. Certainly I lack sufficient neurochemical domain knowledge to make meaningful estimates here, but I’m not as sure as they sound that everyone does.
No one yet knows what the data substrate includes or how much of it has to be preserved for meaningful revival. For all we know, a piece of neocortex dropped into liquid nitrogen might do the trick in a pinch. Or maybe not even the best current cryo techniques would be enough. But it is not really possible to give a meaningful estimate, as cryonics does not appear to be in any reference class for which well-calibrated predictions exist.
Here is a parable illustrating relevant difficulty of both problems:
*Imagine you are presented with a modern manuscript in latin and asked to retype it on a computer and translate everything into English.
This is how uploading more or less looks like for me, data is there but it still needs to be understood, and copied. Ah, you also need a computer. Now consider the same has to be done with ancient manuscript, that has been preserved in a wooden box stored in ice cave and guarded by a couple of hopeful monks:
Imagine the manuscript has been preserved using correct means and all letters are still there.
Uploading is easy. There is no data loss, so it is equivalent to uploading modern manuscript. This means that monks were smart enough to choose optimal storage procedure (or got there by accident) - very unlikely.
Imagine the manuscript has been preserved using decent means and some letters are still there.
Now, we have to do a bit of guesswork… is the manuscript we translate the same thing original author had in mind? EY called it doing intelligent cryptography on a partially preserved brain, as far as I am aware. Monks knew just enough not to screw up the process, but their knowledge of manuscript-preservation-techniques was not perfect.
Imagine the manuscript has been preserved using decent means all traces have vanished without trace.
Now we are royally screwed, or we can wait a couple of thousands of millions years so that oracle computer can deduce state of manuscript by reversing entropy. This means monks know very little about manuscript-preservation.
Imagine there is no manuscript. There is a nice wooden box preserved with astonishing details, but manuscript have crumbled when monks put it inside.
Well, the monks who wanted to preserve manuscript didn’t know that preserving the box does not help to preserve the manuscript, but they tried, right? This means monks don’t understand connection between manuscript and box preservation techniques.
Imagine there is no manuscript. The box has been damaged as well.
This is what happens when manuscript-preservation business is run by people with little knowledge about what should be done to store belongings for thousands of years without significant damage.
In other words, uploading is something that can be figured out correctly in far, far future while the problem of what is proper cryo-storage has to be solved correctly right now as incorrect procedure may lead to irreversible loss of information for people who want to be preserved now. I don’t assign high prior probability to the fact that we know enough about the brain to preserve minds correctly, and therefore cryonics in the current shape or form is unlikely to succeed.
I don’t assign high prior probability to the fact that we know enough about the brain to preserve minds correctly, and therefore cryonics in the current shape or form is unlikely to succeed.
Are you saying that accurate preservation depends on highly delicate molecular states of the brain, and this is the reason they cannot be preserved with current techniques?
I don’t know what is conditional to accurate preservation of the mind, but I am sure that if someone came up with definite answer, it would be a great leap forward for the whole community.
Some people seem to put their faith in structure for an answer, but how to test this claim in a meaningful way?
I don’t know what is conditional to accurate preservation of the mind,
It seems like you’re saying you don’t know whether cryonics can succeed or not. Whereas in your first reply you said “therefore cryonics in the current shape or form is unlikely to succeed.”
I don’t know if it is going to succeed or not (my precognition skills are rusty today), but I am using my current beliefs and evidence (sometimes lack of thereof) to speculate that it seems unlikely to work, in the same way cryonics proponents speculate that it is likely (well, likely enough to justify the cost) that their minds are going to survive till they are revived in the future.
Even though LW is far more open to the idea of cryonics than other places, the general opinion on this site still seems to be that cryonics is unlikely to succeed (e.g. has a 10% chance of success).
How do LW’ers reconcile this with the belief that mind uploading is possible?
I can’t speak for anyone else, but I don’t see a contradiction. Believing that a living brain’s data can be successfully uploaded eventually doesn’t imply that the same data can necessarily be uploaded from a brain preserved with current-day tech. The usual line I see quoted is that cryonics tech isn’t guaranteed to preserve key data, but it has a higher chance than rot-in-a-box tech or burn-to-ash tech.
So are you saying that this key data includes delicate fine molecular information, which is why it cannot be preserved with current tech?
Nope, I’m not saying that. There are many systems that both don’t depend on fine molecular information, and also are easier to restore from being vitrified than to restore from being burned to ash.
Would you agree with shminux’s reply then?
Certainly shminux’s reply isn’t what I had in mind initially, if that’s what you mean.
As for whether I agree with it on its own terms… I’m not sure. Certainly I lack sufficient neurochemical domain knowledge to make meaningful estimates here, but I’m not as sure as they sound that everyone does.
No one yet knows what the data substrate includes or how much of it has to be preserved for meaningful revival. For all we know, a piece of neocortex dropped into liquid nitrogen might do the trick in a pinch. Or maybe not even the best current cryo techniques would be enough. But it is not really possible to give a meaningful estimate, as cryonics does not appear to be in any reference class for which well-calibrated predictions exist.
Here is a parable illustrating relevant difficulty of both problems:
*Imagine you are presented with a modern manuscript in latin and asked to retype it on a computer and translate everything into English.
This is how uploading more or less looks like for me, data is there but it still needs to be understood, and copied. Ah, you also need a computer. Now consider the same has to be done with ancient manuscript, that has been preserved in a wooden box stored in ice cave and guarded by a couple of hopeful monks:
Imagine the manuscript has been preserved using correct means and all letters are still there.
Uploading is easy. There is no data loss, so it is equivalent to uploading modern manuscript. This means that monks were smart enough to choose optimal storage procedure (or got there by accident) - very unlikely.
Imagine the manuscript has been preserved using decent means and some letters are still there.
Now, we have to do a bit of guesswork… is the manuscript we translate the same thing original author had in mind? EY called it doing intelligent cryptography on a partially preserved brain, as far as I am aware. Monks knew just enough not to screw up the process, but their knowledge of manuscript-preservation-techniques was not perfect.
Imagine the manuscript has been preserved using decent means all traces have vanished without trace.
Now we are royally screwed, or we can wait a couple of thousands of millions years so that oracle computer can deduce state of manuscript by reversing entropy. This means monks know very little about manuscript-preservation.
Imagine there is no manuscript. There is a nice wooden box preserved with astonishing details, but manuscript have crumbled when monks put it inside.
Well, the monks who wanted to preserve manuscript didn’t know that preserving the box does not help to preserve the manuscript, but they tried, right? This means monks don’t understand connection between manuscript and box preservation techniques.
Imagine there is no manuscript. The box has been damaged as well.
This is what happens when manuscript-preservation business is run by people with little knowledge about what should be done to store belongings for thousands of years without significant damage.
In other words, uploading is something that can be figured out correctly in far, far future while the problem of what is proper cryo-storage has to be solved correctly right now as incorrect procedure may lead to irreversible loss of information for people who want to be preserved now. I don’t assign high prior probability to the fact that we know enough about the brain to preserve minds correctly, and therefore cryonics in the current shape or form is unlikely to succeed.
Are you saying that accurate preservation depends on highly delicate molecular states of the brain, and this is the reason they cannot be preserved with current techniques?
I don’t know what is conditional to accurate preservation of the mind, but I am sure that if someone came up with definite answer, it would be a great leap forward for the whole community.
Some people seem to put their faith in structure for an answer, but how to test this claim in a meaningful way?
It seems like you’re saying you don’t know whether cryonics can succeed or not. Whereas in your first reply you said “therefore cryonics in the current shape or form is unlikely to succeed.”
Yes.
I don’t know if it is going to succeed or not (my precognition skills are rusty today), but I am using my current beliefs and evidence (sometimes lack of thereof) to speculate that it seems unlikely to work, in the same way cryonics proponents speculate that it is likely (well, likely enough to justify the cost) that their minds are going to survive till they are revived in the future.