This post seems to focus too much on Singularity related issues as alternative arguments. Thus, one might think that if one assigns the Singularity a low probability one should definitely take cryonics. I’m going to therefore suggest a few arguments against cryonics that may be relevant:
First, there are other serious existential threats to humans. Many don’t even arise from our technology. Large asteroids would be an obvious example. Gamma ray bursts and nearby stars going supernova are other risks. (Betelgeuse is a likely candidate for a nearby supernova making our lives unpleasant. If current estimates are correct there will be substantial radiation from Betelgeuse in that situation but not so much as to wipe out humanity. But we could be wrong.)
Second, one may see a high negative utility if one gets cryonics and one’s friends and relatives do not. The abnormal after death result could substantially interfere with their grieving processes. Similarly, there’s a direct opportunity cost to paying and preparing for cryonics.
The above argument about lost utility is normally responded to by claiming that the expected utility for cryonics is infinite. If this were actually the case, this would be a valid response.
This leads neatly to my third argument: The claim that my expected utility from cryonics is infinite fails. Even in the future, there will be some probability that I die at any given point. If that probability is never reduced below a certain fixed amount, then my expected life-span is still finite even if I assume cryonics succeeds. (Fun little exercise, suppose that my probability of dying is x on any given day. What is my expected number of days of life? Note that no matter how small x is, as long as x>0, you still get a finite number). Thus, even if one agrees that an infinite lifespan can give infinite utility, it doesn’t follow that cryonics gives an expected value that is infinite. (Edit: What happens in a MWI situation is more complicated but similar arguments can be made as the fraction of universes where you exist declines at a geometric rate so the total sum of utility over all universes is still finite)
Fourth, it isn’t even clear that one can meaningfully talk about infinite utility. For example, consider the situation where you are given two choices (probably given to you by Omega because that’s the standard genie equivalent on LW). In one of them, you are guaranteed immortality with no costs. In the other you are guaranteed immortality but are first tortured for a thousand years. The expected utility for both is infinite, but I’m pretty sure that no one is indifferent to the two choices. This is closely connected to the fact that economists when using utility make an effort to show that their claims remain true under monotonic transformations of total utility. This cannot hold when one has infinite utility being bandied about (it isn’t even clear that such transformations are meaningful in such contexts). So much of what we take for granted about utility breaks down.
This post seems to focus too much on Singularity related issues as alternative arguments. Thus, one might think that if one assigns the Singularity a low probability one should definitely take cryonics. I’m going to therefore suggest a few arguments against cryonics that may be relevant:
First, there are other serious existential threats to humans. Many don’t even arise from our technology. Large asteroids would be an obvious example. Gamma ray bursts and nearby stars going supernova are other risks. (Betelgeuse is a likely candidate for a nearby supernova making our lives unpleasant. If current estimates are correct there will be substantial radiation from Betelgeuse in that situation but not so much as to wipe out humanity. But we could be wrong.)
Second, one may see a high negative utility if one gets cryonics and one’s friends and relatives do not. The abnormal after death result could substantially interfere with their grieving processes. Similarly, there’s a direct opportunity cost to paying and preparing for cryonics.
The above argument about lost utility is normally responded to by claiming that the expected utility for cryonics is infinite. If this were actually the case, this would be a valid response.
This leads neatly to my third argument: The claim that my expected utility from cryonics is infinite fails. Even in the future, there will be some probability that I die at any given point. If that probability is never reduced below a certain fixed amount, then my expected life-span is still finite even if I assume cryonics succeeds. (Fun little exercise, suppose that my probability of dying is x on any given day. What is my expected number of days of life? Note that no matter how small x is, as long as x>0, you still get a finite number). Thus, even if one agrees that an infinite lifespan can give infinite utility, it doesn’t follow that cryonics gives an expected value that is infinite. (Edit: What happens in a MWI situation is more complicated but similar arguments can be made as the fraction of universes where you exist declines at a geometric rate so the total sum of utility over all universes is still finite)
Fourth, it isn’t even clear that one can meaningfully talk about infinite utility. For example, consider the situation where you are given two choices (probably given to you by Omega because that’s the standard genie equivalent on LW). In one of them, you are guaranteed immortality with no costs. In the other you are guaranteed immortality but are first tortured for a thousand years. The expected utility for both is infinite, but I’m pretty sure that no one is indifferent to the two choices. This is closely connected to the fact that economists when using utility make an effort to show that their claims remain true under monotonic transformations of total utility. This cannot hold when one has infinite utility being bandied about (it isn’t even clear that such transformations are meaningful in such contexts). So much of what we take for granted about utility breaks down.
And if the expected utility of cryonics is simply a very large yet finite positive quantity?
In that case, arguments that cryonics is intrinsically the better choice become much more dependent on specific estimates of utility and probability.
And so they should.