# Considerations on Cryonics

Is cryonics worth it, and if yes, should one cryocrastinate (i.e. postpone signing up for cryonics to a later date)? Betteridge’s law of headlines only applies partially here: Yes, it is probably worth it (under plausible assumptions $2.5m for a 20 year old, and more for older people), and no, cryocrastination is usually irrational. A cost-benefit analysis written in Lua. I also perform a Monte-Carlo simulation using Guesstimate, and find that signing up for cryonics at age 20 is worth in the mean $35m , median -$100k (90% confidence interval: -$1.59m, $63.2m). It therefore seems recommendable to sign up for cryonics immediately.

# Considerations on Cryonics

If I died, would I be rid of my senses?

Or will it retain, trapped within my corpse, in stasis?

If I died, would I be a woman in heaven?

Or would I fall asleep, not knowing what it’s like to feel alive?If I died, would I begin with a new life?

Or would I be gone as quickly as the breath I give last?

If I died, would I be a woman in heaven?

Or would I fall asleep, not knowing what it’s like to feel alive?

*– Patricia Taxxon, “Deconstruct” from “Foley Artist”, 2019*

One day they woke me up

So I could live forever

It’s such a shame the same

Will never happen to you

*– Jonathan Coulton, “Want You Gone” from “Portal 2: Songs to Test By (Volume 3)”, 2011*

Many would-be cryonicists cryocrastinate, i.e they put off signing up for cryonics until a later point in their life. This has often been explained by the fact that signing up for cryonics seems to require high conscientiousness and can be easily be delayed until another point in life: “I’ll get around to doing it eventually” – person who was cremated. However, it hasn’t yet been explored whether this procrastination might be rational: Many cryonics organisations have high membership fees, which might be avoided by waiting.

To find this out, I present a point-estimate model of whether (and if yes, when) to sign up for cryonics. The model is written in Lua.

## Note

This write-up is not intended as an introduction to the concept of cryonics. For a popular introduction to the topic that clarifies many common misconceptions about the practice, see “Why Cryonics Makes Sense” by Tim Urban.

For more basic information about the topic, the Cryonics FAQ by Ben Best, a former director of the Cryonics Institute, answers many questions, as well as Alcor’s Cryonics FAQ.

These texts should answer most questions people usually have about cryonics.

## Cost-Benefit Calculation for Cryonics

If you make 50k$/yr now, and value life-years at twice your income, and discount future years at 2% from the moment you are revived for a long life, but only discount that future life based on the chance it will happen, times a factor of

^{1}⁄_{2}because you only half identify with this future creature, then the present value of a 5% chance of revival is $125,000, which is about the most expensive cryonics price now.

*– Robin Hanson, “Break Cryonics Down”, 2009*

To find out whether to sign up for cryonics at all, one needs to make a cost-benefit calculation. This has been attempted before, but that analysis has been rather short (disregarding several important factors) and it might be productive to approach the topic independently.

The costs of cryonics are comparatively easy to calculate and contain little uncertainty: The price of cryopreservation and life-insurance are widely known, and can be easily added together. The benefits of cryopreservation, however, contain a lot more uncertainty: It is not at all clear that the technology for reuscitation will be developed, cryonics organizations (or humanity) survive to develop such technology, or that the future will be interested in reuscitating people from cryopreservation.

The model presented makes the assumption that a person has a given age and has the option of waiting for signing up for cryonics every year up to their expected year of death. So, for example, a person that is 20 years old now is able to plan signing up when they are 20 years old, 21 years, 22 years and so on up to 78 years. The value of cryonics is calculated, and the value of a regular death is tacitly assumed to be $0.

```
curage=20
actval={78.36, 78.64, 78.66, 78.67, 78.68, 78.69, 78.69, 78.70, 78.71, 78.71, 78.72, 78.72, 78.73, 78.73, 78.74, 78.75, 78.75, 78.77, 78.79, 78.81, 78.83, 78.86, 78.88, 78.91, 78.93, 78.96, 78.98, 79.01, 79.03, 79.06, 79.09, 79.12, 79.15, 79.18, 79.21, 79.25, 79.29, 79.32, 79.37, 79.41, 79.45, 79.50, 79.55, 79.61, 79.66, 79.73, 79.80, 79.87, 79.95, 80.03, 80.13, 80.23, 80.34, 80.46, 80.59, 80.73, 80.88, 81.05, 81.22, 81.42, 81.62, 81.83, 82.05, 82.29, 82.54, 82.80, 83.07, 83.35, 83.64, 83.94, 84.25, 84.57, 84.89, 85.23, 85.58, 85.93, 86.30, 86.68, 87.08, 87.49, 87.92, 88.38, 88.86, 89.38, 89.91, 90.47, 91.07, 91.69, 92.34, 93.01, 93.70, 94.42, 95.16, 95.94, 96.72, 97.55, 98.40, 99.27, 100.14, 101.02, 101.91}
for age=curage,math.floor(actval[curage]) do
print(value(age) .. “: ” .. age)
end
```

`curage`

contains the current age of the user of the
program. `actval`

is an actuarial table that contains at the
nth position the median life expectancy of a person that is
n years old at the moment for a western nation (in this case
Germany).

This model usually tries to err on the side of conservative estimates, think of the lower range of a 50% confidence interval.

## The Disvalue of Waiting

Two important factors play into the value (or disvalue) of waiting to sign up for cryonics: Motivation drift and the possibility of dying before signing up.

```
function value(age)
return prob_signup(age)*prob_liveto(age)*(benefit(age)-cost(age))
end
```

### Motivation Drift

`prob_signup`

is a function that calculates the probability of
signing up for cryonics after having waited up to having a certain
age. It seems clear that people loose motivation to finish plans over
time, especially if they are unpleasant or complex. A good example
for this is people being motivated at the start of the year to do
regular exercise: How many of those actually keep their promises to
themselves? They might start off exercising, but after the first few
weeks the first people drop out, and and a couple of months there is
nearly nobody left still going to the gym except the ones who already
did it before. It seems like there is a strong regression to the
mean in
regards to action: Most regular actions are replaced by inaction, most
strong values are replaced by apathy over time. A similar phenomenon
seems likely for signing up for cryonics: At first, people are very
enthusiastic about signing up, but then loose interest as time progresses.

It doesn’t seem obvious how strong motivation drift is and how it develops over time (some people might regain motivation after some time), but intuitively it seems like a geometric distribution. The reasoning is as follows: Imagine that a thousand people have the motivation to perform a given action n years into the future. Every year, a certain percentage p of the people still motivated loses interest in performing that action and drop out. After n years, the number of people who perform the action is (the percentage of people still motivated is ).

When trying to find out what the value of p is for oneself, one can imagine a thousand independent identical copies of oneself planning a complex plan one year ahead. How many of those would actually follow through on that plan? Intuitively, I’d say that it can’t be much higher than 95%, possibly much lower, especially for something as complex and time-consuming as signing up for cryonics.

```
decay=0.95
function prob_signup(age)
return decay^(age-curage)
end
```

Interestingly, this does not mean that the decision of whether to be cryonically preserved or not is then set in stone as soon as possible: Cryonics memberships are very easy to cancel, in nearly all cases a simple email and a cessation of paying membership fees suffices. Signing up for cryonics earlier protects against regression to the mean, which means apathy or lack of motivation towards cryonics, but does not protect against changing ones mind about cryonics: If one becomes convinced it’s bullshit later, one can easily get out (much more easily than getting in). On the other hand, there might be a feeling of considerable sunk cost due to already paid membership fees and the acquired life insurance.

It will be assumed that once one is signed up for cryonics, one stays signed up for it.

### Dying Before Signing Up

If you die before signing up, all possible value (or disvalue) of cryonics
gets lost. So we want to calculate the probability of dying before having
a certain age given being currently `curage`

years old.

Mortality rates are often calculated using a so-called Gompertz distribution. I determined the b and eta values by eyeballing Wolfram Alpha and using a calculator in Tomasik 2016

```
b=0.108
eta=0.0001
function gompertz(age)
return math.exp(-eta*(math.exp(b*age)-1))
end
```

`gompertz`

returns the probability of reaching `age`

starting from birth, but I need the probability of reaching
`age`

given one is already `curage`

years old. With Bayes
theorem one can calculate
that

is equal to because
being older than `age`

is (in this calculation) a subset of being older
`curage`

, and . Some precautions
have to apply in the case that the probabilities of reaching `age`

is
not independent of the probability of reaching `curage`

, but those are
difficult to estimate and will not be implemented here.

This way, one can implement the probability of living until `age`

given
`curage`

the following way:

```
function prob_liveto(age)
return gompertz(age)/gompertz(curage)
end
```

### Longevity Escape Velocity

Longevity Escape Velocity (short LEV) is the name for the possible year when anti-aging technology becomes so good that people can be rejuvenated faster than they age. Although the concept is considered idle speculation in many circles, many futurists justify not signing up for cryonics because they expect that LEV will arrive during their lifetime, and see no reason to sign up for a cryonics membership they are probably not going to need anyway. In this text, I will consider LEV by assuming there will be a certain year after which the probability of death is practically zero.

I somewhat arbitrarily set this year to 2080, though many futurists seem more optimistic:

```
levyear=2080
```

## Calculating the Cost

Calculating the cost is comparatively straightforward, but there are some hidden variables (like opportunity costs and social costs) that have to be considered (not all of these are considered in this text).

The raw cost for cryonics depends heavily on the organisation choosen for preservation, the basic price range is from ~$20000 to ~$250000. In this case, I chose the costs for neurocryopreservation at Alcor, though this analysis should be extended to other organisations.

Raw cryonics cost can be split into three different parts: membership fees, comprehensive member standby costs and the cost for cryopreservation.

```
function cost(age)
return membership_fees(age)+pres_cost(age)+cms_fees(age)
end
```

### Membership Fees

Membership fees for Alcor are calculated using the age of the member and the length of their membership.

#### Direct Fees

Current Membership Dues, net of applicable discounts, are:

First family member: $525 annually or $267 semi-annually or $134 quarterly.

Each additional family member aged 18 and over, and full-time students aged 25 and under: $310 annually or $156 semi-annually or $78 quarterly.

Each minor family member under age 18 for the first two children (no membership dues are required for any additional minor children): $80 annually or $40 semi-annually or $20 quarterly

Full-time student aged 26 to 30: $460 annually or $230 semi-annually or $115 quarterly.

Long-term member (total membership of 20 − 24 years): $430 annually or $216 semi-annually or $108 quarterly.

Long-term member (total membership of 25 − 29 years): $368 annually or $186 semi-annually or $93 quarterly.

Long-term member (total membership of 30 years or longer): $305 annually or $154 semi-annually or $77 quarterly.

Long-term member (total membership of 40 years or longer): $60.00 annually or $30.00 semi-annually or $15.00 quarterly

*– Alcor Life Extension Foundation, “Alcor Cryopreservation Agreement—Schedule A”, 2016*

The following assumptions will be made in the implementation:

The person considering signing up for cryonics is over 18 years old.

If the person is under 25 years old, they are a student. Considering the fact that cryonics members seem to be more likely to be rich and educated, this seems likely, though maybe a bit classist. The code can be changed if personal need arises.

If the person is over 25 years old, they are not a student.

The person stays a member until their death (otherwise the cryonics arrangement doesn’t work).

The membership fees will not be changed drastically over time. In fact, inflation adjusted prices for cryonics have mostly stayed constant, so this is a reasonable assumption.

The cryonicist will know when LEV has occurred, and will cancel their membership starting from that year.

The implementation is quite straightforward:

```
function alcor_fees(age)
local left=math.min(math.floor(actval[age])-age, levyear-curyear)
local cost=0
if age<25 then
newage=25
cost=(newage-age)*310
end
if left>=30 then
cost=cost+(left-30)*305
left=30
end
if left>=25 then
cost=cost+(left-25)*368
left=24
end
if left>=20 then
cost=cost+(left-20)*430
left=20
end
if age<=25 then
cost=cost+(left-(25-age))*525
else
cost=cost+left*525
end
return 300+cost
end
```

#### Comprehensive Member Standby

For Members residing in the continental U.S. and Canada: Alcor will provide Comprehensive Member Standby (CMS) to all Members (standby in Canada may be subject to delays due to customs and immigration requirements), which includes all rescue activities up through the time the legally pronounced Member is delivered to the Alcor operating room for cryoprotection.

This charge is waived for full-time students under age 25 and minors (under age 18).

*– Alcor Life Extension Foundation, “Alcor Cryopreservation Agreement—Schedule A”, 2016*

Emphasis mine.

Current CMS charges are:

$180 annually, $90 semi-annually, or $45 quarterly

*– Alcor Life Extension Foundation, “Alcor Cryopreservation Agreement—Schedule A”, 2016*

I will assume that the cryonics member starts paying a CMS fee starting 10 years before their actuarial age of death.

```
cms=180
function cms_age(age)
return actval[age]-10
end
function cms_fees(age)
return cms*(actval[age]-cms_age(age))
end
```

### Preservation Cost

There are several different methods of funding cryonics, the most popular of which seems to be life insurance. I haven’t spent much time investigating the exact inner workings of life insurances, so I will make the assumption that the insurance companies price their products adequately, so one doesn’t have much of a financial advantage by choosing life insurance as opposed to simply saving money & paying the cryonics membership in cash. I also assume that life insurance companies can accurately price in the arrival date of LEV.

Minimum Cryopreservation Funding:

• $200,000.00 Whole Body Cryopreservation […].

• $80,000.00 Neurocryopreservation […].

[…]

Surcharges:

• $10,000 Surcharge for cases outside the U.S. and Canada other than China.

• $50,000 Surcharge for cases in China.

[…]

*– Alcor Life Extension Foundation, “Alcor Cryopreservation Agreement—Schedule A”, 2016*

I assume that the person considering signing up lives outside of the U.S (but not in China), since a lot more people live outside the U.S than inside of it. I also assume that the person wants to sign up for neurocryopreservation. With these assumptions, the function that returns preservation costs becomes quite simple:

```
function pres_cost(age)
return 90000
end
```

### Other Possible Costs

There is a number of different additional costs that have not been considered here because of their (perceived) small scale or difficult tractability.

These include opportunity costs for the time spent informing oneself about cryonics (tens of hours spent), opportunity costs for the time spent signing up (tens of hours spent), social costs by seeming weird (though cryonics is easy to hide, and most cryonicists seem to be rather vocal about it anyways), and alienating family members who necessarily come into contact with cryonics (considering the “Hostile Wife Phenomenon”).

## Calculating the Benefit

Calculating the benefit of cryonics carries a great uncertainty, but basically it can be divided into six distinct components: The probability of being preserved, the probability of revival, the amount of years gained by cryonics, the value of one lifeyear, the probability of living to the year when one will sign up, and the probability of then dying before LEV.

```
function benefit(age)
return prob_pres*prob_succ*years_gain*val_year*prob_liveto(age)*prob_diebeforelev(age)
end
```

Here, I will only take point estimates of these values.

### Value of a Lifeyear in the Future

Much ink and pixels have been spilled on the question of the quality of the future, very little of it trying to make accurate or even resolvable predictions. One way to look at the question could be to create clear criteria that encapsulate the most important human values and ask a prediction market to start betting. This could include the power of humanity to make most important decisions regarding its development and resource management, diversity among human beings, average happiness and lifespans and other variables such as inequality regarding resources.

But a much simpler way of approaching the topic could be the following: One takes arguments from both sides (proclaiming positive futures and negative futures) and prematurely concludes that the future is on average going to be neutral, with a high variance in the result. But some problems present themselves: In different value systems, “neutral” means very different things. Strictly speaking, a utilitarian would see human extinction as neutral, but not net neutral (the utility of a world without any sentient beings is exactly 0, which is presumably lower than the current value of the world), anti-natalists consider an empty world to be a positive thing, and most people working on preventing human extinction would consider such a world to be a gigantic loss of opportunity, and therefore net negative.

There seems to be no simple way to resolve these conflicts, otherwise it would have been written down up to now. But it seems like most people would take the current state of affairs as neutral, with improvements in happiness, meaning and wealth to be positive, and decreases in those to be negative. Also, they don’t see dying tomorrow as a neutral event.

#### Caveats

Here I will assume that

Future life years can be averaged in their quality, and that average has monetary value

Future lifeyears are not temporally discounted, i.e a lifeyear 1000 years in the future is as valuable as the next lifeyear

There are no diminishing marginal returns to lifeyears, i.e 1000 life years are 1000 times as valuable as one lifeyear

These are presumably controversial assumptions, but they simplify the analysis. I will continue to read about philosophers’ and economists’ analyses of the relation between additional lifeyears and utility, and update this section.

#### QALYs and VSL

There are two different methods of putting a value on human life: the VSL and the QALY. The Wikipedia page on VSL lists $182000 for the value of a year of life in Australia, and $50000 as the “de facto international standard most private and government-run health insurance plans worldwide use to determine whether to cover a new medical procedure”. This number seems like a good conservative estimate.

Interestingly, this approximately equals a year of waking hours
worth the minimum wage ($$10*16*7*52=$58240$).

Intuitively, the probability distribution over the value of a year of life in the future should then look like this:

```
.l(“nplot”)
.l(“nstat”)
grid([-20000 120000 20000];[0 0.00004 0.000004])
xtitle(“Dollar value of a future life year”)
ytitle(“Probability”)
plot({n.pdf(x;50000;500000000)})
draw()
```

Note that this graph is not based on real data and only for illustrative purposes.

But one can take another factor into account: Most negative future
scenarios don’t lead to reuscitation (civilisational collapse, stable
totalitarianism, existential catastrophes like AI failure, nuclear
war, biotechnological disaster and natural catastrophe all reduce
human capabilities or keep them constant, preventing the development
of reuscitation technology). In most of the negative futures, there
are either no more humans around or people don’t have time, energy or
resources to bring people back from cryonic preservation (if indeed they
still *are* in preservation by that point), and for malicious actors, in
most scenarios it is easier to create new people than to bring preserved
people back.

This effect might be dampened by the consideration that most possible futures have net-negative value, but on the other hand, nearly all of those futures don’t lead to reuscitation.

This would mean that the probability distribution over the value of a lifeyear in the future conditional on being reuscitated could look like this:

```
.l(“nplot”)
.l(“nstat”)
grid([-20000 120000 20000];[0 0.00004 0.000004])
xtitle(“Dollar value of a future life year”)
ytitle(“Probability”)
plot({:[x>50000;n.pdf(x;50000;500000000);0.4472*n.pdf(x;50000;100000000)]})
draw()
```

Note that this graph is not based on real data and only for illustrative purposes.

#### Negative Scenarios

However, I can think of 3 very specific (and thereby highly unlikely) scenarios where people could be reuscitated into a (for them) net-negative world.

##### Ascended Economy

The ascended economy is a scenario where the development of capitalism diverges significantly from the desires of humans, leading to most (if not all) of humanity becoming extinct. It seems highly unlikely, but possible that cryopreserved humans are placed into the hands of an algorithm that invests the money in the relevant funds to reuscitate the cryopreserved humans at a certain point. This algorithm could receive little (or no) information on what to do with the reuscitated humans afterwards, leading either to these humans quickly dying again because of an economy where they are worthless, or being kept alive solely for fulfilling the contract that is embedded in the algorithm. This might lead to insanity-inducing boredom as the humans are kept alive as long as algorithm manages to, possibly hundreds or thousands of years. This would have net-negative value for the people reuscitated.

##### Malevolent Future Actors

A superintelligence becomes a singleton and starts behaving malevolently because of a near miss in its implementation or or because it has been set up by a malevolent human. This would lead to cryopreserved people being reuscitated, having their brains scanned and executed as a brain emulation, copied and put into very painful conditions.

##### Information from the Past is Valuable

In a future where agents that don’t care about humans find the cryopreserved remains of humans, they might be interested in extracting information from those brains. If it is not possible to extract this information without reviving the cryopreserved people, they might reuscitate them and then interrogate these revived people for a very long time, with little regard for their well-being.

#### Steps for Reducing the Risk from such Scenarios

b) When, in Alcor’s best good faith judgement, it is determined that attempting revival is in the best interests of the Member in cryopreservation, Alcor shall attempt to revive and rehabilitate the Member. It is understood by the Member that a careful assessment of the risks versus the benefits of a revival attempt will be material to determining when to attempt revival. […]

d) Where it is possible to do so, Alcor represents that it will be guided in revival of the cryopreserved Member by the Member’s own wishes and desires as they may have been expressed in a written, audio, or videoStatement of Revival Preferences and Desires, which the Member may at his/her discretion attach to this Agreement.

*– Alcor Life Extension Foundation, “Cryopreservation Agreement” p. ^{15}⁄_{16}, 2012*

Although not a failsafe measure, steps can be taken to reduce the risks from hellish scenarios above by making arrangements with cryonics organisations. This may include not wanting cryopreservation to continue in an ascended economy, objecting to revival as an emulation or revival after more than a certain number of years (to prevent being reuscitated in an incomprehensibly strange and alien world).

#### Other Thoughts

Many people argue that the value of a year of life in the future might be much lower than in the present, because friends and familiy are not around, and it is very likely that the future will be extremely alien and unfamiliar.

These are valid considerations, but can be dampened a bit: Humans have shown to adapt to very different and varied circumstances, and humans today feel that modern life in big cities with regular calendars and highly structured lives without any worries about survival is normal, while for most humans who ever lived, it would be anything but. One can speculate that very similar facts will also hold for the future (becoming increasingly unlikely the further reuscitation lies in the future). There would certainly be a big culture shock in the future, but it seems not qualitatively different from the shock people have when they visit different countries today. It is possible that future societies might try to help people with this kind of future shock, but that is of course far from certain.

It is true that most cryonicists will not be able to convince their friends and family to sign up for it too, so they will be alone in the future at first. People today sometimes leave their friends and even families to move to other places, but those people seem to be the exception rather than the norm. However, people nearly always move on with their life, even as they get divorced, get estranged from their friends or see their children less regularly – they don’t seem to prefer death to continuing their lives without specific people. This consideration doesn’t seem to be a True Rejection.

After these considerations, I conservatively set the value of a lifeyear in the future to $50000.

```
val_year=50000
```

### Probability of Revival

Specific equations and values have been proposed, usually yielding probability of success 0 < x < 10%. For example, Steven Harris in 1989 estimated 0.2-15%, R. Mike Perry in the same article runs a different analysis to arrive at 13-77%, Ralph Merkle suggests >85% (conditional on things like good preservation, no dystopia, and nanotech); Robin Hanson calculated in 2009 a ~6% chance, Roko gave 23%; Mike Darwin in 2011 (personal communication) put the odds at <10%; an informal survey of >6 people (LW discussion) averaged ~17% success rate; Jeff Kaufman in 2011 provides a calculator with suggested values yielding 0.2%; The 2012 LessWrong survey yields a mean estimate of cryonics working of 18% (n=1100) and among ‘veterans’ the estimate is a lower 12% (n=59) - but interestingly, they seem to be more likely to be signed up for cryonics.

*– Gwern Branwen, “Plastination versus Cryonics”, 2014*

Besides these estimates, there exist also two related questions on the prediction website metaculus. “Before 1 January 2050, will any human cryonically preserved for at least 1 year be successfully revived?” has a median probability of 16% (n=117), “If you die today and get cryonically frozen, will you “wake up”?” receives 2% (n=407). I am not sure where the difference comes from, perhaps either from worries about the quality of current preservation or from a low trust in the longevity of cryonics organisations. This google sheet contains 7 estimates of success: 0.04%, 0.223%, 29%, 6.71%, 14.86%, 0.23% and 22.8%, with various different models underlying these estimates.

Calculating the mean of these results in a chance of ~13%:

It would certainly be interesting to set up a prediction market for this question, or get a team of superforecasters to estimate it, but basically, it seems like for a young or middle-aged person, the estimated probability is around 10%. However, the people surveyed are often sympathetic to cryonics or even signed up, and people in general are overconfident, so being conservative by halving the estimate seems like a good idea.

```
prob_succ=0.05
```

### Years Gained

Conditional on being revived, what is the average life expectancy?

If revival is achieved, it seems highly likely that aging and most degenerative diseases have been eradicated (it makes little sense to revive a person that is going to die again in 10 years). Also, most revival scenarios hinge upon either the feasibility of very advanced nanotechnology, which seems to be highly conducive to fixing aging, or on whole brain emulation scenarios, which would likely make aging unnecessary (why on purpose degrade a digital brain?).

If revival happens, there are still risks from accidents and homicide or suicide that can kill the reuscitated cryonicist, as well as existential risks that face all of humanity.

The website Polstats illustrates the risks purely from accidents and homicides using data from the Information Insurance Institute. They arrive at “a much more impressive 8,938 years” average life expectancy. An answer on Mathematics StackExchange to the question “What’s the average life expectancy if only dying from accidents?” arrives at 2850 years.

Taking existential risks into account is a bit harder. It is unclear whether most of the probability mass for existential risks should be placed before reuscitation of cryonics patients becomes feasible, or after it. It is also unclear how high the existential risk for humanity is overall. Assuming that the existential risk for humanity over the next 10000 years is ~40% (this number is pretty much a guess), and half of that risk is placed before reuscitation, then the life expectancy of cryonics is .

That number should be qualified further in an “Age of Em” scenario: that scenario will contain less natural risks (emulation can be backed up, they live in a simulated world where homicide risks and care accidents make no sense), but an em also suffers from the risk of not having enough money to continue being run, and from the fact that the em era might not last several subjective millennia. This scenario deserves further consideration (see also Hanson 1994).

To conclude, it seems like reuscitated cryonicists will on average live around 4500 years, although there is room for debate on this number.

```
years_gain=4500
```

### Probability of Being Preserved

It seems like not all people who sign up for cryonics remain cryonicists until their death, and not all cryonicists who die as members actually get preserved.

There seems to be very little data about this question, but as an extremely conservative estimate I would put the ratio of members of cryonics organizations who actually get preserved at 60% (it seems likely that the actual number is higher). Fortunately, a cryonics member can increase this number by being diligent about their cryonics arrangement, living near the preservation facility before death, informing family members about their arrangement, trying to lead a safe life and keeping contact to their cryonics organisation.

```
prob_pres=0.6
```

### Surviving Until LEV

The benefit of cryonics is only realized in one case: One lives to the planned year of signing up, but then dies before LEV. Both dying before signing up or living until LEV make the value of cryonics $0. One can calculate the probability of this scenario by multiplying the probabilities of living until signup with the probability of then dying before LEV.

To calculate the probability of living to a given age, we can use the gompertz distribution again:

```
function prob_liveto(age)
return gompertz(age)/gompertz(curage)
end
```

The probability of dying before LEV is 0 if LEV has already occurred:

```
if curyear+(age-curage)>levyear then
return 0
```

Othewise, we assume that one has signed up for cryonics at `age`

and now wants to know the probability of dying until LEV. That is the
same as , or the probability of living until
`curage+(levyear-curyear)`

given one has already lived until `age`

.

```
else
return 1-(gompertz(curage+(levyear-curyear))/gompertz(age))
end
```

## Conclusion

The complete code for the model can be found here.

### Standard Parameters

With the parameters presented above, it turns out that it is optimal to sign up for cryonics right away, mainly because the motivation drift punishes the procrastination quite heavily.

#### Currently 20 years old

At the age of 20 years, the value of signing up for cryonics the same year is $2797894 () according to this model, prolonging the decision until one is 30 reduces this number to $1666580 (), and waiting until 40, 50 and 60 years yields a value of $982100 (), $559610 () and $287758 (), respectively.

```
.l(“nplot”)
data::.r()
grid([0],(#data),[10];[0],(|/data),[1000000])
xtitle(“Years from now”)
ytitle(“Dollar value of signing up for cryonics”)
barplot(data)
draw()
```

#### Currently 40 years old

The values of signing up for cryonics look very similar to the values for a 20 year old. Performing the signup immediately at age 40 is worth $6590556 ($~$6.6*10^6$) at age 40 and is the best time to do it.

```
.l(“nplot”)
data::.r()
grid([0],(#data),[10];[0],(|/data),[1000000])
xtitle(“Years from now”)
ytitle(“Dollar value of signing up for cryonics”)
barplot(data)
draw()
```

### Without Motivation Drift

Since many people question the idea of motivation drift and trust
themselves in the future a lot, one can simulate this trust by setting
the `decay`

parameter to 1.

In this model, a very different picture emerges:

```
.l(“nplot”)
data::.r()
grid([0],(#data),[10];[0],(|/data),[1000000])
xtitle(“Years from now”)
ytitle(“Dollar value of signing up for cryonics”)
barplot(data)
draw()
```

It is still optimal to sign up without hesitation, but now the difference is much lower.

```
$ lua cryoyear.lua 20 50000 0.05 0.6 4500 1 | sort -n | tail −10
2785676.2860511: 29
2787605.1801168: 28
2789611.0731771: 27
2791107.7280825: 26
2792420.5648782: 25
2793783.1701729: 24
2794997.5035013: 23
2796078.6567918: 22
2797040.1939684: 21
2797894.3040717: 20
```

This means that cryocrastination is not that much of a sin even with a lot of self trust.

### The Critic’s Scenario

Somebody who is very critical might object and argue that the probability of success is much lower, and even if cryonics succeeds, it will not lead to extremely long lifespans. Let’s say they also don’t believe in value drift. Such a person might propose the following assignment of variables:

```
curage=20
val_year=50000
prob_succ=0.01
years_gain=50
prob_pres=0.6
decay=1
```

In this case, signing up for cryonics has negative value that converges to 0 the older one gets:

```
$ lua cryoyear.lua 20 50000 0.01 0.6 50 1 | sort -n | tail −10
−80320.313507659: 69
−78526.595695932: 70
−77042.774290053: 71
−75002.570281634: 72
−72832.328023689: 73
−70916.116822976: 74
−68452.980090227: 75
−65840.832675399: 76
−63425.293847013: 77
−60490.80006618: 78
```

Please note that the following graph should have negative values on the y-axis. This should get fixed sometime in the future.

```
.l(“nplot”)
data::-.r()
grid([0],(#data),[10];0,(|/data),[10000])
xtitle(“Years from now”)
ytitle(“Dollar value of signing up for cryonics”)
fillrgb(0.4;0.4;1)
barplot(data)
draw()
```

### Other Modifications

It is possible to think of many other modifications to the parameters in the script, including the probability of cryonics success, the value of a lifeyear, the amount of years gained, or even bigger modifications such as adding models for the probability of the development of life extension technology in the near future.

The reader is encouraged to enter their own value and execute the script to determine whether it is advantageous for them to sign up for cryonics, and if yes, whether cryocrastination would be a good idea.

## Appendix A: A Guesstimate Model

The website Guesstimate describes itself as “A spreadsheet for things that aren’t certain”. It provides Monte-Carlo simulations in a spreadsheet-like interface.

I used Guesstimate to calculate the uncertainty in the value provided by signing up for cryonics as a 20 year old. The model is available here.

### Variables

Most of the parameters were simply taken from this text, but some deserve more explanation.

#### Year for Longevity Escape Velocity

When I give any kind of timeframes, the only real care I have to take is to emphasize the variance. In this case I think we have got a 50-50 chance of getting to that tipping point in mice within five years from now, certainly it could be 10 or 15 years if we get unlucky. Similarly, for humans, a 50-50 chance would be twenty years at this point, and there’s a 10 percent chance that we won’t get there for a hundred years.

*– Aubrey de Grey, “Aubrey de Grey on Progress and Timescales in Rejuvenation Research”, 2018*

The 90% confidence interval for this variable lies in : Aubrey de Grey gives a mean of 2038, I believe that number to be quite optimistic, but not completely so. He doesn’t give a lower bound, but judging from the reasonable assumption that longevity escape velocity is likely not 2 years away, this seems like a log-normal distribution-ish, which is also what I used in the spreadsheet, with a 90% confidence interval in .

#### Age at Death

Unfortunately, Guesstimate doesn’t support Gompertz distributions, so I had to approximate the age of death by assuming that it was a log-normal distribution with the 90% confidence interval in , but mirrored along the y-axis. The data by Wolfram Alpha looks similar to the end result, and both have a mean age of death of ~83 years.

#### Years Lived After Revival

This was another log-normal distribution, with a 90% confidence interval of years. Why the huge range? On the one hand, revival without sufficient rejuvenation technology seems unlikely, but possible; another possibility is being revived and then dying in an accident or war. The high upper range accounts for a very stable future with rejuvenation technology. Although the distribution is log-normal, the mean is still 32000 years, and the 50th percentile is around 1300 years.

#### Value of Lifeyears After Revival

Here, I assumed that both negative and positive development of the future is equally possible, resulting in a normal distribution with a 90% confidence interval in . I personally believe that being revived in a future with negative value is quite unlikely, as outlined in this section, but it’s always the thing that people bring up and want to argue about endlessly (perhaps trying to convince me of their values or test whether mine are acceptable), so I included the possibility of substantial negative development.

#### Provider Cost per Year

Implementing the whole `membership_fees`

in Guesstimate seems possible,
but incredibly burdensome. I approximated it using a normal distribution
with a 90% confidence interval of .

### Value

The result is certainly interesting: in this model, signing up for
cryonics has a mean value of $35m and a median of ≈-$100k
(perhaps because of longevity escape velocity arriving and making
the value simply the cost for signing up), but with very long tails,
especially on the positive side: a fifth percentile of -$1.59m,
and a 95th percentile of *squints* $63.2m – quite a range!

The minimum and maximum of the simulation are even more extreme: -$1b for the minimum and $11.3b for the maximum.

Because of these huge numbers, perhaps it makes sense to try to visualize them logarithmically. I exported the numbers for the variable ‘Value’ from Guesstimate and converted them into a Klong array.

```
.l(“math”)
.l(“nplot”)
.l(“https://niplav.github.io/values.kg″)
logvalues::_{:[x<0;-ln(-x):|x=0;x;ln(x)]}‘values
logvalues::logvalues@<logvalues
incidence::{(logvalues@*x),#x}‘=logvalues
grid((*logvalues),(*|logvalues),[5];[0],(|/#’=logvalues),[100])
scplot2(incidence)
draw()
```

Note that the scale is logarithmic to the natural logarithm (symmetrically for both negative and positive values), not the logarithm to base 10, because this makes the data more granular and therefore easier to understand.

As one can see, the distribution has turned out sort-of
bimodal:
Most cases of signing up for cryonics have a value of -$100k (presumably
because longevity escape velocity arrives first), the
rest is either very negative of very positive. To be exact,
`(+/{*|x}‘flr({*x<0};incidence))%#logvalues`

of cases have negative value, and
`(+/{*|x}’flr({*x>0};incidence))%#logvalues`

of cases
have positive value. Of the ones with negative value, most are simply
flukes where longevity escape velocity arrives first: `2286%#logvalues`

.

### Conclusion

In this model, signing up for cryonics is still a good idea from a strict expected-value perspective. But decision processes with a precautionary principle might be much more wary of doing anything rash before futures with negative value can be ruled out.

Does this analysis take into account the fact that young people are most likely to die in ways that are unlikely to result in successful cryopreservation? If not, I’m wondering what the numbers look like if you re-run the simulation after taking this into account. As a young person myself, if I die in the next decade I think it is most likely to be from injury or suicide (neither of which seems likely to lead to successful cryopreservation), and this is one of the main reasons I have been cryocrastinating. See also this discussion.

I have been putting this off because my medical knowledge is severely lacking, and I would have to estimate how the leading factors of death influence the possibility to get crypreserved mainly by subjectively evaluating them. That said, I’ll look up some numbers, update the post and notify you about it (other people have been requesting this as well).

Why would injury prevent cryopreservation, unless it’s head injury?

I spent time working in fatal car crash investigation (reading crash reports and doing engineering analysis, nothing as gory as you’re probably picturing), and car crashes often involved massive head trauma or would, at a minimum, require *hours* of lag time before the cryonics team could make it there. I’d say at a complete guess that only about 10% involved people dying in hospital later on (i.e. under circumstances that a cryo team could get to them in time to prepare the body).

My impression of the technology is that it’s too much in its infancy to be able to say with any sort of confidence that a body that had been left with minimal treatment for a good 8-10 hours would be in a good state for preservation. And my understanding is that after only a few minutes/hours the brain starts to really degrade.

This is a major reason I’m not considering yet. I also live in a country without a good cryo organisation, and the exchange rates make the fees for Alcor quite a lot when I am not convinced I’d get the value. I also think the 5% figure is way too high.

I mean admittedly, pascal’s wager comes into play a bit here, but I’m not convinced that my current jurisdiction is a good place to die and be cryopreserved, and I have no plans to move.