Actually, I updated slightly towards Alcor being the best choice for cryonics because of this information, since all 3 are likely more intelligent, rational, and informed than I am.
If Yudkowsky is signed up with C. I., then that is slight evidence that C. I. is superior to Alcor. This is so not because he is famous, but because he is more intelligent, rational, and informed than I am. Of course, if I actually studied cryonics in-depth, then my new knowledge would screen-off most of the evidential weight of the opinions of these “famous” people.
As far as I can tell, Eliezer picked C.I. to minimize the cost of signaling his views about cryonics, not because he thought it was better than Alcor. See this comment.
See the comment below: My primary reason for signing up for cryonics was because I got sick of the awkwardness, in important conversations, of trying to explain why cryonics was a good idea but I wasn’t signed up for cryonics.
Consider that it might actually be evidence for a different conclusion: Eliezer signed up for cryonics some years ago, when he had little income, bravely foregoing well-paid employment in favor of pursuing his core goals. (I can relate to that!) I would very much like to talk to E.Y. about whether it’s time to reconsider his past decision based on current information and current finances. I’m just an email or a phone call away, Eli...
Consider that it might actually be evidence for a different conclusion
I’d express it this way: by conservation of evidence, Eliezer signing up for CI is evidence for CI and against Alcor. Within the set of reasons/scenarios which lead to him signing up for CI, the observation about when Eliezer signed up is evidence for the ‘economizing’ explanation in which his signing up is not evidence for CI over Alcor.
(This may sound contradictory, but the important thing is that A as a set can be shrinking in total probability even as individual members of A become more likely.
An example of this would be the hope function: if you’re searching drawers one at a time for a letter, each time you search a drawer, you expect more strongly that the next drawer will hold the letter, even as you also expect more strongly that the letter is not in your desk at all.)
Because it was an excuse to bring in the hope function by way of correcting Max’s statistical reasoning, something I find really cool given how simple & obscure it is.
Someone did ask EY whether the fact that he signed up with CI whereas Hanson signed up with Alcor meant he disagreed with him about something important about the two institutes, and IIRC he answered it was just that Hanson was richer and older than him so of course he’d chosen the higher-end option. (I cannot find that comment, since even a search for Eliezer CI Hanson Alcor in the internal LW search engine turns up umpteen pages.)
Indeed, I’m young and not yet rich. If I was rich, though, given my prior state of knowledge I would’ve gone with SA + CI on the belief that CI seemed more long-run stable—CI seems more risk-averse and more financially prudent. I’ve updated somewhat on the financial prudence of Alcor as a result of reading these threads, and if the decision suddenly mattered for some reason, I would now require more investigation to figure out whether SA + CI or Alcor was the better long-run bet.
Depends what you mean by risk-averse. Alcor has an unquestionable history of fighting for its members’ wishes, rather than giving up on them at the mere hint of a legal battle. The only way in which CI could be said to be more financially prudent (but in a way with its own costs) is in its remarkable ability to hold down operating costs. I’m working hard on reducing our costs without undesirable penalties in terms of capabilities. I think we are also now at a point where further membership growth will yield significant economies of scale.
But take a look at both organizations’ financial statements. You will see that CI expects to maintain patients indefinitely—and revive them—on a small amount of per-patient funding. That takes some heroic and highly risky assumptions to accept. Alcor has carefully structured institutions and policies to manage sustainably for the long-term, including strict limits on what can be charged to the patient care trust fund, a 2% draw on the Endowment Fund, and an investment policy that has been giving us gains (while CI has been losing on its investments). If you continue to delve into the gory details, I think you may continue to update your views further.
The first question you need to ask Yudkowsky (and yourself) is a damned difficult one to answer “simply,” and that is what are the currently well known, well understood, and well documented BIOLOGICAL differences in outcome that are likely to pertain using the two different approaches in the reasonably optimum case. Reasonably optimum means that the member is experiencing medico-legal death under controlled conditions with competent cryonics organization personnel in attendance, My bet is that only a few people on the planet can answer that question, and that Yudkowsky isn’t one of them.
Of course, if you do not believe the degree of molecular, histological or gross damage to the patient matters, within broad limits, then such differences are immaterial. For instance, if you think that several hours of warm ischemic injury, followed by 12 to 24 hours of cold ischemic injury, followed by reperfusion injury, followed by grossly inadequate cryoprotective perfusion/equilibration in the brain resulting in large areas of massively ice injured brain tissue will be easily repairable with Nanotechnology, then you will be largely insensitive to the differences between Ci and Alcor, or a well done cryoprotective perfusion and a poorly done one.
My question for such people is, “Why bother with perfusion at all? The ischemic delays are very damaging. Why not just have yourself packed in dry ice as soon as you are pronounced and get shipped off to CI? It would be about $10K to $15 cheaper and you’d only be faced with Nano-repair of cryoinjury?” No need for Nano, Nano, one Nano will do.
I’m in the final stages of preparing Part 3 of THE EFFECTS OF CRYOPRESERVATION ON THE CAT for publication on Chronosophere. Part 3 is the transmission electron microscopy of the tissues under different conditions of cryopreservation (Part 2 was the histology: http://chronopause.com/index.php/2012/02/14/the-effects-of-cryopreservation-on-the-cat-part-2/). You can look at those pictures of cell and tissue structure and decide for yourself which condition you’d rather be in.
I realize I’m probably going to lose some points with you by stating this. But assuming the limit of perfect technology and the absolute correctness of the pattern theory of identity—if you can’t accept these hypotheses, please just say so, instead of answering based on a different hypothesis—is there any definitive rejection of my admittedly naive notion that if you can literally read out every single atomic position, then “Chop off the head with a guillotine and drop it into a bucket of liquid nitrogen” should, yes, just work? I admit that my actual belief and assumption is that current cryonics efforts are massive overkill by people who don’t realize that liquid nitrogen is not a secure encryption method for brains.
Can you refine what you mean by “the limit of perfect technology”? If you expect atomic tweezers, you’re probably right. If you expect superpowered but still annoying analogues of current methods for manipulating individual atoms, you’re probably wrong. Nanotech is surprisingly hard—it looks less like surgery with a knife you made with a rock and more like using the rock to pound on the knife’s handle during surgery. (But I’m an amateur.)
You lose whatever information is no longer in the atoms, which might be a lot because the skull is not designed to assist cooling, and the brain is a considerable thermal mass. It’s going to cool slowly, be shredded to mush by crystal formation, and be warped and cracked by thermal stress, while undergoing runaway chemical reactions and cell death. Your “limit of perfect technology” is then faced with an awe inspiring task of running the reaction products backwards, modelling and reversing the thermal damage, un-killing the cells, and splicing the cracks, in 3D on tissue that does not come with alignment hints, and then inferring a mind. There’s going to be some level of physically unavoidable data loss even in the perfect case, the data is entailed in thermal noise and random photons and the damage is no longer reversible without reversing the universe. Presumably the perfect technology will paper over these cracks by copying in mind structures from Mr Perfectly Average. But the end result would be that you’re less you.
I am a cryoskeptic because I don’t believe the pattern theory of identity, but in any case, it seems that this is a rather important issue for people who do, and who are seeking cryonic suspenstion. This thread (and Mike Darwin’s blog) are full of detailed histories and analysis about numerous aspects of cryonics. But I don’t see an analysis anywhere of how the organizations rate, when evaluated specifically from the perspective that atomic-scale mapping and reconstruction/simulation of the suspended brain will become possible, and that this is enough for personal survival. If we assume this to be true, and if we put aside considerations about the relative ability of cryonics organizations to keep their patients frozen—just focusing on the specific suspension procedures that they apply—how do they rate? Are any of them “not good enough”, even by these assumptions? Or do they all get a pass?
is there any definitive rejection of my admittedly naive notion that if you can literally read out every single atomic position, then “Chop off the head with a guillotine and drop it into a bucket of liquid nitrogen” should, yes, just work?
Logically, it’s possible that there is dynamic information not conveyed by “every single atomic position” that is lost when making a static copy. One could imagine that a recording of the positions over some amount of time would do better.
Admittedly, our current understanding of physics might already rule out this possibility without my knowing.
I’ve wondered, before, whether there’s any way to get yearly MRI, eeg, fMRI, etc. recordings on durable media for future preservation with your corpsicle. I couldn’t afford it, but it seems like it could be useful.
Actually, I updated slightly towards Alcor being the best choice for cryonics because of this information, since all 3 are likely more intelligent, rational, and informed than I am.
Surely there are three famous people signed up with CI. Eliezer’s one, I suppose. This evidence isn’t evidence.
If Yudkowsky is signed up with C. I., then that is slight evidence that C. I. is superior to Alcor. This is so not because he is famous, but because he is more intelligent, rational, and informed than I am. Of course, if I actually studied cryonics in-depth, then my new knowledge would screen-off most of the evidential weight of the opinions of these “famous” people.
As far as I can tell, Eliezer picked C.I. to minimize the cost of signaling his views about cryonics, not because he thought it was better than Alcor. See this comment.
Consider that it might actually be evidence for a different conclusion: Eliezer signed up for cryonics some years ago, when he had little income, bravely foregoing well-paid employment in favor of pursuing his core goals. (I can relate to that!) I would very much like to talk to E.Y. about whether it’s time to reconsider his past decision based on current information and current finances. I’m just an email or a phone call away, Eli...
I’d express it this way: by conservation of evidence, Eliezer signing up for CI is evidence for CI and against Alcor. Within the set of reasons/scenarios which lead to him signing up for CI, the observation about when Eliezer signed up is evidence for the ‘economizing’ explanation in which his signing up is not evidence for CI over Alcor.
(This may sound contradictory, but the important thing is that A as a set can be shrinking in total probability even as individual members of A become more likely.
An example of this would be the hope function: if you’re searching drawers one at a time for a letter, each time you search a drawer, you expect more strongly that the next drawer will hold the letter, even as you also expect more strongly that the letter is not in your desk at all.)
Umm, here’s a suggestion: WHY DON”T YOU JUST ASK ELIZER HOW AND WHY HE MADE THE DECISION? Why speculate?
Because it was an excuse to bring in the hope function by way of correcting Max’s statistical reasoning, something I find really cool given how simple & obscure it is.
This is precisely why I both love and hate Less Wrong.
Someone did ask EY whether the fact that he signed up with CI whereas Hanson signed up with Alcor meant he disagreed with him about something important about the two institutes, and IIRC he answered it was just that Hanson was richer and older than him so of course he’d chosen the higher-end option. (I cannot find that comment, since even a search for
Eliezer CI Hanson Alcorin the internal LW search engine turns up umpteen pages.)Indeed, I’m young and not yet rich. If I was rich, though, given my prior state of knowledge I would’ve gone with SA + CI on the belief that CI seemed more long-run stable—CI seems more risk-averse and more financially prudent. I’ve updated somewhat on the financial prudence of Alcor as a result of reading these threads, and if the decision suddenly mattered for some reason, I would now require more investigation to figure out whether SA + CI or Alcor was the better long-run bet.
Depends what you mean by risk-averse. Alcor has an unquestionable history of fighting for its members’ wishes, rather than giving up on them at the mere hint of a legal battle. The only way in which CI could be said to be more financially prudent (but in a way with its own costs) is in its remarkable ability to hold down operating costs. I’m working hard on reducing our costs without undesirable penalties in terms of capabilities. I think we are also now at a point where further membership growth will yield significant economies of scale.
But take a look at both organizations’ financial statements. You will see that CI expects to maintain patients indefinitely—and revive them—on a small amount of per-patient funding. That takes some heroic and highly risky assumptions to accept. Alcor has carefully structured institutions and policies to manage sustainably for the long-term, including strict limits on what can be charged to the patient care trust fund, a 2% draw on the Endowment Fund, and an investment policy that has been giving us gains (while CI has been losing on its investments). If you continue to delve into the gory details, I think you may continue to update your views further.
Yes, many (almost all) events are evidence for more than one hypothesis.
The first question you need to ask Yudkowsky (and yourself) is a damned difficult one to answer “simply,” and that is what are the currently well known, well understood, and well documented BIOLOGICAL differences in outcome that are likely to pertain using the two different approaches in the reasonably optimum case. Reasonably optimum means that the member is experiencing medico-legal death under controlled conditions with competent cryonics organization personnel in attendance, My bet is that only a few people on the planet can answer that question, and that Yudkowsky isn’t one of them.
Of course, if you do not believe the degree of molecular, histological or gross damage to the patient matters, within broad limits, then such differences are immaterial. For instance, if you think that several hours of warm ischemic injury, followed by 12 to 24 hours of cold ischemic injury, followed by reperfusion injury, followed by grossly inadequate cryoprotective perfusion/equilibration in the brain resulting in large areas of massively ice injured brain tissue will be easily repairable with Nanotechnology, then you will be largely insensitive to the differences between Ci and Alcor, or a well done cryoprotective perfusion and a poorly done one.
My question for such people is, “Why bother with perfusion at all? The ischemic delays are very damaging. Why not just have yourself packed in dry ice as soon as you are pronounced and get shipped off to CI? It would be about $10K to $15 cheaper and you’d only be faced with Nano-repair of cryoinjury?” No need for Nano, Nano, one Nano will do.
I’m in the final stages of preparing Part 3 of THE EFFECTS OF CRYOPRESERVATION ON THE CAT for publication on Chronosophere. Part 3 is the transmission electron microscopy of the tissues under different conditions of cryopreservation (Part 2 was the histology: http://chronopause.com/index.php/2012/02/14/the-effects-of-cryopreservation-on-the-cat-part-2/). You can look at those pictures of cell and tissue structure and decide for yourself which condition you’d rather be in.
I realize I’m probably going to lose some points with you by stating this. But assuming the limit of perfect technology and the absolute correctness of the pattern theory of identity—if you can’t accept these hypotheses, please just say so, instead of answering based on a different hypothesis—is there any definitive rejection of my admittedly naive notion that if you can literally read out every single atomic position, then “Chop off the head with a guillotine and drop it into a bucket of liquid nitrogen” should, yes, just work? I admit that my actual belief and assumption is that current cryonics efforts are massive overkill by people who don’t realize that liquid nitrogen is not a secure encryption method for brains.
It doesn’t have to be a secure encryption method to be a lossy compression method.
Can you refine what you mean by “the limit of perfect technology”? If you expect atomic tweezers, you’re probably right. If you expect superpowered but still annoying analogues of current methods for manipulating individual atoms, you’re probably wrong. Nanotech is surprisingly hard—it looks less like surgery with a knife you made with a rock and more like using the rock to pound on the knife’s handle during surgery. (But I’m an amateur.)
You lose whatever information is no longer in the atoms, which might be a lot because the skull is not designed to assist cooling, and the brain is a considerable thermal mass. It’s going to cool slowly, be shredded to mush by crystal formation, and be warped and cracked by thermal stress, while undergoing runaway chemical reactions and cell death. Your “limit of perfect technology” is then faced with an awe inspiring task of running the reaction products backwards, modelling and reversing the thermal damage, un-killing the cells, and splicing the cracks, in 3D on tissue that does not come with alignment hints, and then inferring a mind. There’s going to be some level of physically unavoidable data loss even in the perfect case, the data is entailed in thermal noise and random photons and the damage is no longer reversible without reversing the universe. Presumably the perfect technology will paper over these cracks by copying in mind structures from Mr Perfectly Average. But the end result would be that you’re less you.
I am a cryoskeptic because I don’t believe the pattern theory of identity, but in any case, it seems that this is a rather important issue for people who do, and who are seeking cryonic suspenstion. This thread (and Mike Darwin’s blog) are full of detailed histories and analysis about numerous aspects of cryonics. But I don’t see an analysis anywhere of how the organizations rate, when evaluated specifically from the perspective that atomic-scale mapping and reconstruction/simulation of the suspended brain will become possible, and that this is enough for personal survival. If we assume this to be true, and if we put aside considerations about the relative ability of cryonics organizations to keep their patients frozen—just focusing on the specific suspension procedures that they apply—how do they rate? Are any of them “not good enough”, even by these assumptions? Or do they all get a pass?
Logically, it’s possible that there is dynamic information not conveyed by “every single atomic position” that is lost when making a static copy. One could imagine that a recording of the positions over some amount of time would do better.
Admittedly, our current understanding of physics might already rule out this possibility without my knowing.
I’ve wondered, before, whether there’s any way to get yearly MRI, eeg, fMRI, etc. recordings on durable media for future preservation with your corpsicle. I couldn’t afford it, but it seems like it could be useful.
There are also more esoteric uses for regular baselines of that sort of thing. They come in handy while recovering from brain damage, for example.