Cryonics and Pascal’s wager
The Cambridge UK meet-up on Saturday 12 February went really well. Many thanks to everyone who came and provided a wonderful and entertaining discussion.
One of the topics that came up was that of cryonics. This is the idea of having your body (or maybe just your brain) frozen after death, to be thawed and revived in the far future when medical technology has advanced to the point where it can heal you. Is this a rational thing to do?
The argument I heard from some of the other attendants effectively boils down to “what have you got to lose?” In other words, have yourself frozen just in case it works and you can be resurrected.
This struck me as awfully reminiscent of Pascal’s Wager, which is similarly a “what have you got to lose?” type argument. Cited in its original form, it is about belief in a god and goes something like this:
You can either believe in God or not. If you do, you will either be rewarded with eternity in heaven (if you’re right) or nothing happens (if you’re wrong). But if you don’t believe, you will either be punished by eternal torture (if you’re wrong) or nothing happens (if you’re right). It’s a no-brainer! You’re better off believing.
This argument falls down on many counts, but I’ll concentrate on a specific one. It makes a far-fetched assumption about the set of possible outcomes. It assumes that there are only the two possibilities quoted and no others. It ignores the possibility of a god that only rewards sceptical atheists.
Coming back to cryonics, the argument seems to proceed approximately like this:
You can either have yourself frozen or not. If you do, you will either wake up in a wonderful, happy-go-lucky utopian future with amazing technological advances (if cryonics works) or nothing happens (if it doesn’t). But if you don’t have yourself frozen, nothing happens either way. It’s a no-brainer! You’re better off in cryopreservation.
If I haven’t already made it abundantly clear, the assumption that the future you wake up in is at all desirable for you is a far-fetched one. It ignores the possibility of waking up as a slave with no opportunity for suicide.
What are everybody’s thoughts on this?
Welcome to Less Wrong, Timwi. It looks like you’re relatively new here, so you might not know that the topic of cryonics has received a lot of discussion on LW, enough to appear in the tag cloud on the front page. People here tend to have relatively favorable views of it: some (including Eliezer Yudkowsky) are convinced that it’s the right choice, have signed up, and advocate that everybody sign up, and even those who aren’t convinced tend to at least see cryonics as a reasonable possibility that’s worth considering.
Before getting too caught up in debating particular aspects of cryonics, you may want to take some to familiarize yourself with some of the ideas and arguments that have already been discussed on LW to get a better sense of the case for cryonics. The LW wiki entry and this post by Eliezer are good places to start, and they also give you various links to follow to learn more.
I think it’s OK for a newcomer to post topics in the discussion forum that have already been discussed at length. You learn a lot more if you can actively engage in conversation rather than just reading what other people have written.
Cryonics is a good bet if the expected utility from being restored in the future exceeds the expected utility you lose from funding it in the present, modulo any discounting you may choose to perform. As best I can tell, there are two free variables in that equation: first, the probability of being restored; and second, the expected utility of your future life given restoration.
I don’t think we can put firm bounds on either of these, but I’d be very surprised if the latter isn’t positive. Your slavery hypothesis strikes me as unlikely: if a society capable of resuscitating cryonics patients found itself in need of cheap labor, it would almost certainly have cheaper sources of it. Ideological reasons for resuscitating cryonics patients and making their new lives unpleasant seem somewhat more likely (perhaps the future finds some aspect of our behavior repugnant and demands retribution?), but I can’t think of many that don’t require several things to go wrong at once.
See Reversal test, Motivated skepticism, Rationalization.
(The assumption that you can go to a grocery store and buy some orange juice is a far-fetched one, since it ignores the possibility of a global conspiracy led by Sarah Palin bent on not letting you do that.)
These possibilities are not equally unlikely.
I agree with this. It is worth considering, though, that there are possibilities other than slavery and utopia, and also that “desirable for you” is a bit more complicated of an issue that material calculations of utility would suggest.
I think human life is better in the 21st century than it was in the 16th century, and that it was better in the 16th century than it was in the 11th century, etc. I am also pretty sure that if you could grab people from the 19th century and bring them into the present, they would adapt and ultimately end up with more utility in our time. I’m less certain this would be true for someone from the 11th century, however. The world might be so alien to their preferences, and their ability to acquire meaningful skills to contribute in any way other than as a museum curiosity be so limited, that they aren’t grateful, even though in material terms life is objectively better. So if cryonics is successful, but takes a thousand years instead of a hundred, the results might not be as “desirable” as one would hope.
Perhaps not quite as implausible as waking up (or staying asleep) as batteries, but certainly in the same vein!
There are not all that many scenarios in which future agents would bother rebuilding us to use as slaves. We just wouldn’t be worth the hassle.
The technology needed to revive you would almost certainly make slavery obsolete.
I disagree. Robin Hanson has extensively argued that the development of software emulations (such a development is a likely necessity for cryonics to work) would lead to a crash in wages (since supply of labor would be able to rise quickly enough to meet demand for the first time since the Industrial Revolution). In that scenario, almost all market power would fall to the owners of capital, and since wages would be near subsistence anyway, lots of ems might offer ownership of themselves in exchange for the right to survive. That situation has historical precedent (and lingers in parts of Africa to this day).
The technology needed to revive you would almost certainly make humans obsolete.
It very well may be that a future society might value the first-hand accounts of the early 21st century as told by a resuscitated mom or popsicle. (no disrespect intended; just anticipating future slang).
I think the chances of revival are much better than the sliver thin chances suggested by the analogy to Pascal’s Wager. While there are many non-technical ways it can fail, no-one is in a position to be very confident that we’ll see any of those failure modes—that there will be a general societal failure that stops the LN2 deliveries, or that the cryonics organisations themselves will collapse, or such. And while there are a great many unknowns on the technical side, the more I look into it the more I think that the chances of recovery being technically plausible are high—well into the “more likely than not” side of things.
For reference, Yudkowsky’s comments on the topic.
Pascal’s wager is for what seems like infinite utility times a finite although small probability of happening. Infinite trumps small, how could devoting your life to a non-vanishing probability of infinite utility ever be wrong within the kinds of mathematical approach favored around here? Within that mathematical model, the fact that there are infinite negatives (if the muslim god turns out to be right you get an infinite negative utility from believing in the Christian god). So Infinity—Infinity is undefined, we have no idea what the real payoff is for Pascal’s wager, no idea if the expectation value is positive or negative.
But with cryonics, the downside seems close to nothing, I guess the vanishingly small probability that you delay your entry into heaven for a few hundred to billion years while you remain frozen, until cryonics fails or the earth is destroyed, or the other even lower probability that you are revived by people advanced enough to revive you but economically retarded enough to value humans as slaves.
I’m still not signing up. I’d still rather have a summer house or a really kick-ass vacation in Africa or something. But Yudkowsky’s article I think does clarify the issue re: Pascal’s wager.
If we stay here ten billion years,
In the embers of the last shining sun,
We’ll have no less days to sing God’s praise,
Than if we cark it when it’s barely begun.
That regrettably fails to fit the usual tune.
Another place Pascal’s wager fails on is the cost one has to pay for believing (such as attending services, time/money for religious holidays, ect.). In Cryonics case, it is a literal price, the cost of keeping you frozen.
So, what this boils down to is calculating the utilities for the probability of being able to live in the future vs. the value of the money that you and your family could have used for something else.
Pascal’s wager does not fail there. It would hold up under torture, early death and poison that transforms your blood into fire ants too. Cryonics is not nearly so robust in that particular respect.
If you believe that nothing is worse than being dead, then cryonics does legitimately become a no-brainer (it’s not like it could make you more dead).
But accepting that nothing is worse than being dead, despite being common on LW, is not trivial. It seems at a minimum you’d have to accept Fun Theory (I don’t yet understand Fun Theory well enough to accept or reject it).
Not quite. Even then there is opportunity cost to consider. Those resources could be directed to other methods of life extension.
None of which happens to work now, despite propaganda I’ve heard since the 1970′s about imminent breakthroughs in superlongevity. The people who say we’ll have 150 years life expectancies or whatever by some randomly postulated year in this century don’t understand what “life expectancy” means. We determine life expectancy retrospectively from statistics gathered about populations of individuals who have already died, and we don’t know until a significant number of people have died past the age of 150 to see if they constitute a trend, or instead represent statistical noise. Clearly we won’t have the ability to gather those data in this century—I would have to wait until the year 2109 to see my 150th birthday—and I think all these “immortalist” obsessives like Ray Kurzweil just waste money and possibly damage their health by ingesting their “life extension” quackery.
By contrast, we can conduct experiments in brain cryopreservation which generate useful data in a timely fashion, like most other scientific experiments. If you want to see research into something which could show tangible returns for your survival, cryonics has some advantages over chasing after an anti-aging breakthrough which won’t arrive for many decades, if not centuries.
Up front, I’ll say that I am not signed up for cryonics, and have no particular plans to. (Yes, I know this means that I will almost certainly be permanently dead within 50 years.) Nobody else at the meet-up was signed up, although some expressed a vague intention to do it one day.
Anyway, an argument in favour is not only that you are likely to wake up (if you wake up at all) in a favorable situation, but you can look forward to a greatly extended lifespan, since making living bodies last longer seems like an easier problem than resuscitating corpsicles frozen with the crude technology of today. This multiplies the utility by a huge amount.
Despite that and other arguments, and a personal desire to live a healthy life for as long as possible, I remain unenthusiastic about devoting a substantial proportion of my resources to the project (not just money—you cannot buy cryonics like you can buy a picture to hang on your wall, you need a plan for the actual suspension). Small probabilities of stupendous outcomes fail to move me even if their product is greater than the cost. You can play the figures like a guitar and get any answer you want.
A downside that I haven’t seen much attention paid to, although it does get mentioned from time to time, is the problem of having the organisation that has the care of your corpsicle surviving long enough, and taking good enough care. You can’t freeze a social structure and put it in a vat for a century—it has to live through all the time that you don’t. What are the chances?
There’s a flip side to that, which is that a cryonicist who takes the idea of reanimation seriously must also take the idea of future stability more seriously. Unlike any other living person they have reason to anticipate the long-term results of today’s political stances, investments, and social developments in terms of actual sensory experiences. I’m not sure if this actually translates to increased rationality, but seems like should.
A surprisingly large number of firms have lasted for centuries, even in war-torn countries like Germany:
http://en.wikipedia.org/wiki/List_of_oldest_companies
2% for a century. Multiply that into the chances of cryonics working.
That surprises me, too—about 1% are at least 200 years old, though I don’t know where the 2 million in the database come from. That seems very low for all the companies in the world, or even in the developed world.
See http://lesswrong.com/lw/z0/the_pascals_wager_fallacy_fallacy/
FM-2030 expressed a more optimistic assessment about the values of Future World in his ebook Countdown to Immortality:
I would also point out the fact that we don’t get “the future” all at once. Even if our initial living conditions seem suboptimal at first, if we have radically extended lives, we’ll have the time to work towards situations more to our liking.
Cryonicist Thomas Donaldson (Ph.D. in mathematics from the University of Chicago) pointed out the bad assumptions with this sort of reasoning back in 1989:
http://www.alcor.org/Library/html/probability.html
In other words, cryonicists can get off their butts and start to do some constructive things to make the project more likely to succeed. I’ve donated some money towards cryonics-related research that few other people seem interested in, for example. I’d like to see a lot more of that instead of the tendency for cryonicists, who jumped onto Drexler’s “nanotechnology” distraction early on, to fantasize about “how cool it would be if we had nanotech factories which would give us genie-like superpowers.” Nano-nonsense: 25 years of charlatanry