What is the probability of cryonics working?

And how do you decide if it's worth it?

Do you like being alive? Do you like your loved ones being alive? Would you like that period of living to be longer, so long as you are in good health? Suppose you answer: “Yes” or, like me, “Hell yes!” What if you also have doubts that life extension advances will come soon enough to keep you alive? Then you have good reason to wonder whether cryonics could work for you. How likely it is to work? And is the cost and effort worth it?

You will not have any difficulty finding people who claim that cryonics is impossible. They may be unwilling to think seriously about the possibility and so dismiss it as “like jumping from here to the moon” — an actual assertion! They may assert that memories don’t survive the process or that revival technology will never be developed. They may assert the cryonics violates laws of physics without being able to show how. There’s not much you can say to someone who has made up their mind without truly considering the matter.

On the other side, you will find it extremely challenging to find a cryonics supporter who claims that it will certainly work. “Just sign up with a cryonics organization and you’re all set!” This is not something I have ever heard. You may hear one person say they estimate the probability at 1%, or 5%, or 30%, or occasionally 90%, but never 100%. The success of cryonics – whether it is the odds of it working for you or for someone – has a probability. Reasonable people will ask three questions:

• What is that probability?

• How do we go about trying to estimate it?

• How accurate is our estimate likely to be?

To make it personal, let’s interpret the question to mean: What is the probability that cryonics will work for me?

If this were a question like “what are the odds that I will win ten million dollars in the lottery?", we would need to specify a time frame. Winning the lottery after you are dead isn’t of much use. You might want to know the odds of winning it while you are still vigorous enough to enjoy it. Cryonics, by its nature, doesn’t have any practical time limit. Part of cryonics “working” involves successful revival from the cryopreserved state. It doesn’t fundamentally matter how long it takes to develop the technology. Of course, the longer it takes the greater the odds of something going wrong. But no specific period of time is relevant to the nature of success in cryonics.

I am not going to limit the question to the odds of cryonics working from purely a technical perspective. The odds that we could make it work if we chose to do so is higher than the actual probability for any individual in the actual world. The probability that we will be able to land a person on Mars is higher than the probability that you will ever walk on Mars. So, we are not merely considering the chances that cryonics would work in principle, but also that you would get cryopreserved successfully, stay cryopreserved long enough, and then be successfully revived. This means considering technical, organizational, and social factors.

What would success look like?

We haven’t clarified the question sufficiently yet. We are asking about the probability of you successfully going through the cryonics process, but what does “you” mean? To be successful, would you require that 100% of your body and brain and memories be fully preserved? You may have reasons to preserve your entire body but would probably still count the effort as a success if your brain survived intact and could be housed in a replacement body. So, success doesn’t require more than reviving the brain.

Even if we limit the criterion for success to the brain – or the mind and personality that it embodies – surely 100% fidelity is not necessary. The contents of our minds changes continually. You are not 100% the same as you were yesterday, and less like the person you were twenty years ago. Just how much continuity is necessary and of what type is a deep philosophical question that cryonicists have often pondered.

Most of us would prefer to retain 100% of who we were at the point of being cryopreserved. Speaking for myself, I would require that most of my core personality be revived – my values, dispositions, and basic ways of looking at things. If I lost all my specific declarative memories but retained most of my personality, I would consider that good enough to count as success. You might feel differently and require at least some memories to be retained. Obviously, I would prefer that but, with personality intact, I might be able to fill in some memory gaps from external records. This isn’t very different from a stroke survivor who does not recover 100% but is able to piece together his personality over time and does have a continuous awareness of himself.

If you don’t regard the process as successful unless you retain 99.9% or 100% of what makes you you, clearly the odds of it working are much lower. The lower the fidelity you allow, the better your odds. This will not matter for some factors – such as the probability that your cryonics organizations survives – but it might matter for other factors – such as the probability of getting cryopreserved under sufficiently good conditions.

How should we estimate the probability?

The common approach to answering the question “What are the odds that cryonics will work for me?” is to intuitively settle on a number that feels right. “80% is about right. Things are going in the right direction. James Bedford is still cryopreserved after 56 years.” “5 or 10%. So much could go wrong.” If you are not actually calculating the odds, you should refrain from giving a specific number. It is no more meaningful or helpful than someone who says AI will kill us all with a probability of 20% or 90%. This amounts to no more than throwing out a number that simply reflects your more or less optimistic or pessimistic view.

This approach is similar to someone claiming that cryonics causes “damage” without getting specific about this damage and why future technologies would not be able to infer the healthy state from a damaged state.

Apart from going for an intuitive overall probability estimate – or let’s be honest and call it a guess – you could go to the trouble of putting together a Delphi panel. That’s an awful lot of work. Another option is to look at betting markets. In many areas, these outperform experts by structuring the wisdom of crowds effectively. They are more reliable when used by many bettors. Cryonics has been tackled in several forms by the forecasters in the Metaculus markets.

The first attempt at phrasing was this: “If you die today and get cryonically frozen, will you "wake up"?”” Success is defined as the revived individual has full mental faculties, an essentially complete set of memories of their former life, and a personality that is at least difficult for them or others to discern from the original. 436 forecasters collectively came up with 5.8%. It had been as high as around 30% in 2021 but went down considerably over the next six months for some reason.

Another bet is phrased differently: “Will any person that has been cryopreserved for more than 1 year be resuscitated or emulated before 2200?” Only 56 forecasters contributed. Collectively, they came up with a 58% probability.

Another question asks: “What will the earliest preservation date of any resuscitated cryonics patient be?” 38 forecasters and 111 predictions come up with a median date of 2060. The most pessimistic 25% come up with 2097. In all cases, the number of forecasters is small, so the results are not reliable. If the number of forecasters reaches the thousands, the outcome would be worth taking more seriously. We would have a major additional reason to take it seriously if real money were at stake.

The conjunctive probability approach

Let’s assume that you are not going to form a Delphi panel, nor set up a prediction market. Another way to figure out the odds is to break the problem down into sub-problems or sub-questions. Those of a mathematical mind may want to do this in a particular way: conjunctive probability. The idea is simple: Think about what must happen for cryonics to work for you. Break it down into specific events, problems, or conditions. Estimate the probability of each factor required for success. Multiple those probabilities together.

For instance, suppose you can break some event down into three things that have to happen for the event to occur. To make it simple, let’s assume each step has a probability of 50%. The probability of the event is 0.5 x 0.5 x 0.5 = 0.125 or 12.5%.

This approach has been taken by Steve Harris (1989), Gregory Benford (2011) in his novel Chiller, and by Brook Norton at cryonicscalculator.com. In theory, there are some strong advantages of this method. The cognitive psychology literature has extensively studied how people estimate the probability of the outcome of events that can be decomposed into necessary steps. These studies find that people typically come up with a probability that is too high. We tend to overestimate the probability of an event that is really the product of multiple necessary prior events or conditions. Even experts in the subject matter are prone to this problem. Breaking the matter down into a conjunctive analysis can give us a more accurate estimation of the probability.

The most familiar example of conjunctive analysis is the Drake equation for estimating the probability that intelligent life exists beyond Earth. More specifically, the Drake equation attempts to calculate the number of civilizations in the Milky Way galaxy with which communication might be possible. Some of the eight factors in the equation could at least potentially be determined scientifically: the average rate of star formation in our galaxy; the fraction of those stars that have planets; and perhaps the average number of planets that can potentially support life per star that has planets. Others require a lot of subjective guesswork, such as the fraction of planets with life that actually go on to develop intelligent life.

Conjunctive probability analysis has also been used to estimate the risk that AI will destroy all humans. I recently looked at two of these analyses, by Jon Evans and Rohit Krishnan. In estimating the probability of AI doom, you are really estimating the probability that multiple things will happen. These might include being able to create an artificial general intelligence (AGI); the probability of not developing alignment or control; the probability of AI being able to act dangerously in the physical world; the probability of AI being self-improving, and so on.

Steve Harris’s probability equation has 13 components. These items consider the probability:

• Pa: that personal identity is a purely physically defined quantity.

• Pb: that personal identity resides in the mechanical structure of the brain

• Pc: of suffering clinical death in such a fashion as to have the physical control of one’s brain be passed to cryonicists before one’s mechanical identity patterns have been degraded too much.

• Pd: that the cryonic suspension process does not destroy too much mechanical information in the brain.

• Pe: that your brain will make it to future revival time without a mechanical accident thawing you.

• Pf: that your cryonics organization will make it to revival time.

• Pg: that your society will make it to revival time intact without major social upheavals that destroy cryonics organizations.

• Ph: that cryogenic storage of bodies or brains will stay continuously legal until revival time.

• Pi: that full scale development of nanotechnology is possible within the context of physical law.

• Pj: that, if nanotechnology can be done, mankind will do it.

• Pk: that mankind, your society, and your cryonics organization will survive the development of nanotechnology.

• Pl: that the cryonic revival process will ever be inexpensive enough to be paid for by your cryonics organization or somebody else.

• Pm: that society will permit the revival of cryonauts, once possessed of the ability to do so.

As you can imagine, with 13 factors multiplying together, the overall odds are going to be low. If you assigned a 90% probability to each factor, the end result is a probability of 25%. At 70% each, the result is a probability of 0.1%.

How can conjunctive probability analysis go wrong?

One challenge in carrying out this kind of analysis is in specifying the factors. These can be individuated or grouped in different ways. When we are not talking about clearly quantifiable factors, you can always break a factor down into subfactors. This will reduce the overall probability. We should be suspicious because you could take any real-world event that we know is likely to occur, break it down into numerous discrete events that we know are not 100% certain to occur, and end up showing that it is unlikely.

Consider the factors listed by Harris.

Pa: It’s not obvious that an assumption of materialism is needed. Even a dualist could accept cryonics because there is no obvious reason why a soul could not (a) remain attached to a cryopreserved person (as perhaps happens with frozen embryos), or (b) return to the body upon revival.

Pe and Pf: It seems reasonable to combine these under “sufficient competence of your cryonics organization”.

Pg and Ph: The organization could move patients to another country and, eventually, into space, or members could move or relocate when near clinical death. Rather than an objective probability, this depends in part on the actions of yourself and your cryonics organization.

Pj: Nanotechnology may not be needed. For patients cryopreserved later in excellent conditions, nanotechnology might not be needed. To the extent that “nanotechnology” refers to a specific technology such as described by Drexler and Freitas, perhaps some other technology will be used instead.

Pk: This is not completely independent of Pg.

Pl and Pm should probably be combined into a single factor. We don’t need “society” to be able to afford to revive us; we need our cryonics organizations to be able to do it.

We might also question the 50% survival definition, or at least expand how it is determined. Information stored outside your brain may be able to fill in gaps

Are the factors objective and independent?

For conjunctive probability analysis to give an answer worth anything, the probabilities of the component factor must be objective. This may be possible in a mechanistic context where you’re considering the odds of a machine operating reliably by looking at the chance that each component might independently fail, causing the whole system to fail. If you have a large number of such machines, you can observe the frequency of failure of the parts. That gives you an objective basis for the probabilities.

Of the 13 items in Harris’s equation only one arguably might or could be objective: Pc: the probability of undergoing clinical death in such a fashion as to have the physical control of one’s brain be passed to cryonicists before one’s mechanical identity patterns have been degraded too much. If we have a large enough number of cases, we might be able to come close to a probability. Even here, learning on the part of cryonics organizations and changes in culture and law could change that probability.

All 12 other factors are clearly not objective. May of them are future one-time events for which we have no statistical guidelines. The probability of these events depends on not only the organization’s choices but also the individual’s choices regarding (a) your health – and what kind of health (cancer vs. heart disease/stroke); (b) where you choose to live; (c) who you live with; (d) whether you keep your financing and paperwork up to date — including planning for inflation; and so on. We might become better at meeting these challenges. Or new, unforeseen challenges may arise.

The dependence of these events on individual and organizational actions also means that they are not independent factors, making conjunctive analysis inappropriate. Underlying factors also make the events dependent to some degree. As Mike Perry put it:

The major difficulty, as I see it, is that a good many of the social variables that could kill cryonics, and which are treated as independent, are not really independent at all.

Seven of the thirteen conditions are not related to the technical feasibility of cryonics, or the requirement that cryonic suspension be performed under reasonable conditions, but instead are what I would lump under “the social problem,” which I call condition n. I don’t believe these seven conditions can be treated as independent.

This is why we must make a clear distinction between the probability that cryonics practiced under good conditions will work and the probability that it will work for you. For example, CPR performed under good and timely conditions has a good probability of working but that does not mean that there is a high probability that CPR will be successful for you in a random out-of-hospital cardiac arrest.

The probabilities of these factors will change over time. Assuming we don’t discover new problems, the odds will go up – technology will advance, organizations will get stronger, the ability to anticipate and plan your clinical death will improve, and less time will remain for the organization to fail or for legal problems.

Aschwin de Wolf summarized these problems in a 2009 essay.

Although such calculations give the semblance of objectivity, they are equally vulnerable to the fundamental objection that assigning one single number to the probability that cryonics will work is just a lot of hand waving.  How many independent events are there and how do we know that they are independent? What is the basis for assigning  specific probabilities to these conditions?

Thomas Donaldson criticized the supposed objectivity of probabilities in stirring terms:

It is wrong for us either individually, or as a class of people (cryonicists) to take occurrence of world-wide or even Solar System-wide events as necessarily our own personal fate. It is wrong and far worse than wrong. Habits of mind which identify ourselves with the general fate of humanity assume an abdication of that exact responsibility we must take over our own fate. Putting our fate into a model of passive probability assumes our own passivity.

We very much should not identify our own fate with that of “society” or “mankind” or even the Earth (cryonics societies should found offices off the Earth as soon as that becomes possible). Didn’t we become cryonicists because we proposed to escape that which all the philosophers said was the common fate of all mankind? Floods, earthquakes, meteor impact, major war, mobs searching for us in every cranny, how could we be fazed by such trivialities having once adopted our major goal?

In the end, you can’t take the numbers very seriously. The most useful thing to take away from the conjunctive probability approach is to use it as a way of clarifying the crucial factors. Then you and your cryonics organization can work at improving the probability that everything will work out.

Is cryonics a worthwhile investment?

In deciding whether to make cryonics arrangements, you could estimate the changes of it working for them and then take a number for the value of a life. By multiplying the payoff by the probability of it being achieved, they can then compare it to the cost.

In deciding whether to make cryonics arrangements, you could estimate the changes of it working for them and then take a number for the value of a life. By multiplying the payoff by the probability of it being achieved, they can then compare it to the cost.

This calculation requires that we assume a lifespan following revival. We have no way of knowing what this will be. It seems to reasonable to expect us to continue to improve safety measures and to use advanced technologies to make our bodies more robust. When we have conquered aging, we will look closely and sternly at the remaining causes of death and work on reducing them further. Life expectancy could rise to tens of thousands of years or more. If you accept the possibility in backing up your personal identity critical information, redundant storage could statistically extend your lifespan enormously.  

Let’s take a highly conservative approach by using current death rates from accidents. If you never age but accidents kill at the same rate as today, how long can you expect to live? Estimates vary widely but two sources put it at 2000 years and 8,938 years. Some of the current causes of death – such as automobile accidents in human-driven vehicles – are likely to plummet. Even taking the second source and assuming, ridiculously, no reduction in hazards, our lifetime post-revival would be around 120 times our current lifespan.

Finally, we should acknowledge that making cryonics arrangements may not be completely based on estimates of individual success. We may do it because we support the cause – that’s why I made arrangements in 1986 even though I was only 22 and unlikely to need it. And we may want to help the organization succeed, to set a social example, and so on. Many people support causes that they don’t expect to benefit from themselves.

Just do it!

By thinking about the situation carefully we can clarify the factors that support and threaten our successful cryopreservation and revival. In the end, though, probabilities in this area are, at best, estimates. The decision to take the journey is a highly personal one. To me, it’s a grand adventure, a leap into the unknown. Many people are understandably deterred by the unknown and wouldn’t take the trip even if they had high confidence that it would work. To me, the noble choice is to be venturesome, to take the chance, to assert what control I can over entropy.

Summary

• In the end, it is not possible to objectively calculate the odds of cryonics working.

• We can form a better idea by considering the things that have to go right, or the things that could go wrong.

• It is important to distinguish between the probability that cryonics will work as a technical proposition and the probability that it will work for an individual person.

• Once we know the critical factors, we can shift the odds in our favor by taking steps — where we live, which organization we join, building support for our wishes among the people around us, especially family, setting up an asset preservation trust, creating an appropriate living will.