Immortality and its Critics: Why Rationalism Should not Treat its Darling as Unpopular or Simple

Hear this article read aloud here!

Author’s Note: this post based on an old Discord conversation

To open this, I feel like I should address the question of whether this is an insider’s or outsider’s critique. Am I a Rationalist? Nick has quipped that the classic praxis act of Rationalism is writing a Rationalist blog. Thinking Much Better is run by Nick, a self-identified Rationalist, and is named almost like a parody of site titles throughout the Rationalsphere, like LessWrong, Overcoming Bias, and Clearer Thinking 1. In my own articles I have written about aspects of discourse relevant to Rationalism, and have regularly admired and linked to the writing of self-identified Rationalists like Scott Alexander and Kelsey Piper.

At the same time, being part of an ideology can sometimes depend on fine points of category difference, for instance, if I am a sentientist, am I necessarily a humanist, or necessarily not a humanist? In my case, I consider myself an Effective Altruist. On one view, this automatically makes me a Rationalist of some sort. On another view, in order to be a Rationalist as well, I need to side with Rationalism where it differs from Effective Altruism in at least some places, which isn’t that common for me, otherwise I am merely an Effective Altruist, and therefore bound to be somewhat aligned with Rationalists.

I don’t pretend that mine is either the most serious or most general criticism of the Rationalist movement worth making. A more classic criticism for example is the fact that it has helped incubate neoreactionary thought, as an iffy outsider’s critique of Rationalism I previously interacted with mentioned. While I consider this criticism serious, it is limited in several key ways. Neoreactionary thought isn’t close to a majority view among Rationalists (according to one stat Scott Alexander has mentioned, it seems to be around 3%), and it has also been heavily influenced by continental philosophy, a set of ideologies Rationalism generally doesn’t care for and which generally doesn’t care for Rationalism. I also get the impression that Rationalist discourse is a slightly saltier, less welcoming environment than Effective Altruist discourse as a rule, however it is still better than most circles, and I can’t evidence this concern well. It may just be idiosyncratic to my readings (not to mention with the salt levels of my last couple of articles, it feels somewhat hypocritical, ah well).

There is, however, a specific area of Rationalist discourse that has always stood out to me as especially aggravating, and which I feel I have some relative competence to highlight as a trend both common among a very large portion of Rationalists who bring the topic up, and one where the discursive conduct of these Rationalists leaves a very great deal to be desired. Immortality, unlike other causes and ideas favored by Rationalists, seems to be very much treated as the movement’s baby. Something they have uniquely worked to popularize, mostly all agree on, and which fits very well with their ideological aesthetic. All of the Rationalish media Nick has brought to my attention that deals with immortalism has shared similar pathologies. Putting transparently hokey arguments in the mouths of their critics, entirely ignoring more practical issues of implementation, sometimes with a Stockholm Syndrome diagnosis, and of course, my very favorite, calling critics and detractors killers.

Although I am sympathetic to their goals, I have in the past briefly interacted with the trend of parts of leftist/socialist twitter treating its own baby, Medicare For All, with extremely similar limited-scope narratives and hostility against critics. While the worst Medicare For All activists seem to have done much more damage to their discourse than immortalists have thus far done to their own, I don’t want to see something similar happen in the future to this topic, and suspect most Rationalists would agree: this is not the trend in argument they want to be compared to. The main exception to this trend I’ve seen is Anders' Sandberg’s relevant interview with the 80,000 Hours podcast, which actually interacted with some of the more challenging objections these other media leave out or minimize. Sandberg however (admittedly like Yudkowsy and Bostrom) is also an Effective Altruist, so this is a weak defense against the toxicity of immortalism being a pathology of Rationalism.

To start, one of the things that annoys me about these pieces of media is that they often read to me as strawmen. The usual counterargument I see skewered is the idea that “death gives life its meaning”. This is something that I don’t believe in the least, and I am annoyed by immortalists who spend their energy kicking around the same impotent objection as though it is the most compelling thing skeptics could worry about. Still, I say that it only “reads to me as a strawman” because, unlike a strawman, I believe it is actually an extremely common objection. It is understandable, it has a tempting elegance to it, but the issue is that it doesn’t meet its burden of proof for describing something that convincingly accounts for life’s meaning. Unless you rig the argument by making claims about the subconscious or about denial, it seems clear to me that when I am doing the activities that have the greatest claim to being meaningful by most accounts, loving, thinking, creating, experiencing in general, I am not thinking about my death. Even when I’m planning I usually don’t factor in death, though its absence would certainly change some plans of mine. It seems like an account of life’s meaning that omits all the parts of my life I consider relatively untouched by my mortality would be deeply impoverished. More fatally, unless again you make some argument about subconscious knowledge, we need to be taught that we are going to die. There is an age before which we live, in ways it seems like we should consider meaningful, when we just don’t know that life is going to end. I could easily imagine someone spending their whole life believing they would never die. Either you must argue that their eventual death makes their life meaningful even though the way their life seemed from the inside was no different from a similar portion of a real immortal’s life, or you must deny that this person had a meaningful life at all, which seems implausible to me.

The more convincing, related argument I see in some media appeals to the idea that, if we have forever, we will get sick of life, just plain bored forever. The strongest counter argument to this, it seems to me, pays a significant philosophical cost. You could point out that people don’t have infinite minds and identities, they can’t approach a limit for experience, because even as they have infinite experience, their minds and identities will end up gradually changing and leaving things behind, to the point where a mind is unrecognizable as the same one a trillion years down the line.

The cost this pays is that immortalists become seriously vulnerable to “replaceability” arguments concerning the ethics of death. When is it that someone’s continued life is of no more value than its replacement with a new life? Normally, this is a difficult, controversial line to explicitly define, but unless you take the “further fact” view of personal identity in which someone’s identity consists in something deeper than psychological properties that come in degrees, which it seems to me you shouldn’t, these immortal beings are almost indistinguishable from the case where they are replaced anyway. The key difference is that their replacement is gradual, their death is a slow transition superimposed on another’s person’s slow birth. While I’m sympathetic to the view that people won’t fear death in this world in the way they would in the other, you don’t see immortalists advocating for engineering away people’s fear of death, so this seems unlikely to be the reason.

The bigger counter seems to be autonomy, the idea that it is fundamentally wrong for someone to die against their wishes, even if all this means is a sharper discontinuity. I have a fairly hot take on this topic however, if we assume that this autonomy consideration is at the bottom of the apparent difference. In my experience most people who have thought about this issue, including more skeptical people like me, agree with the immortalist Rationalist types pretty much exactly on actual policy, seriously eroding the “literal killing” appeal. That is, we pretty much all agree that a good ideal for eventual human (or transhuman) life is for people to live indefinitely, and only die when they choose to.

For example, the key source of 2010s philosophical pop-culture and occasional philosophical adversary of mine, “The Good Place”, ended with a meditation on the badness of immortality appealing largely to the boredom concern, as well as vaguely hinting at the meaning issue more deeply. Based on the ideas concentrated on in the last episodes, the most plausible way to read the ending is as a piece of anti-immortalist media, but its resolution is basically what I see Rationalists arguing for. Live as long as you want, only die when you choose to. There is a dispositional difference here, and a difference on some values and ways of thinking, but all of them the sorts of differences that are pretty petty to call people murderers over when the actual policy advocacy is the same (not the first time I’ve been irritated by a debate where the confrontational tone hasn’t matched the material disagreements).

One possible source of difference may be the point that indefinite life means you could live forever, because you could choose to live forever, or it seems to be in some essential sense similar to immortality, and that this immortality is itself desirable in a way we shouldn’t knock as media like “The Good Place” does. I can’t imagine the Rationalists who have thought about this for very long coming to this conclusion, but perhaps the size of “forever” really is just that difficult to capture in an ideology, and some people manage to take this idea seriously. It seems to me that there is some evidence for this in this Wait But Why post about Graham’s number, where imagining how big this really big but infinitely smaller than infinity number really is makes Tim Urban more skeptical of the idea of immortality.

There are other things I have seen that resort to cartoonish extremes to try to make the inaccessibility of infinity feel more present, this long quote from a memorable chapter from “a Portrait of the Artist as a Young Man” in which a preacher is describing Hell comes to mind as perhaps the classic literary example:

“You have often seen the sand on the seashore. How fine are its tiny grains! And how many of those tiny grains go to make up the small handful which a child grasps in its play. Now imagine a mountain of that sand, a million miles high, reaching from the earth to the farthest heavens, and a million miles broad, extending to remotest space, and a million miles in thickness: and imagine such an enormous mass of countless particles of sand multiplied as often as there are leaves in the forest, drops of water in the mighty ocean, feathers on birds, scales on fish, hairs on animals, atoms in the vast expanse of the air: and imagine that at the end of every million years a little bird came to that mountain and carried away in its beak a tiny grain of that sand. How many millions upon millions of centuries would pass before that bird had carried away even a square foot of that mountain, how many eons upon eons of ages before it had carried away all. Yet at the end of that immense stretch of time not even one instant of eternity could be said to have ended. At the end of all those billions and trillions of years eternity would have scarcely begun.”

In the end, the mathematics alone isn’t satisfying, but everyone who has looked at even just the mathematics rationally should already know that indefinite life means mortality of some sort, and mortality of a sort infinitely smaller than immortality, and, assuming you care about immortality in terms of what it would be like on the ground rather than just some more abstract property, infinitely different.

Another alternative view is that immortality is good, not just indefinite life, and if indefinite life won’t give you anything like immortality, then the immortality should be absolute, we should push for an average probability of death approaching or equal to zero in each moment. That involuntary death, and voluntary death, should both be made impossible. I think very few people believe this, I have seen Amanda Askell at least explicitly distinguish this unendable life from what she calls “living forever”, suggesting some sort of language or conceptual gap. I definitely don’t believe this, but someone might make the argument that you never have a right to cut off a life infinitely larger than what you have experienced so far, that there is some deep sense in which you are never qualified to make this choice to cut off an infinitely larger remaining period of your life. If you have death-centered moral values of some sort, this might make sense.

If you can be motivated to believe that the value of life can easily be overridden by imaginable-if-rare circumstances, for instance extreme levels of suffering, you should be profoundly risk averse about the possibility of infinite, inescapable net negative lives. An example that troubles me deeply when imagining immortality of this sort is to picture certain types of situations in which one is trapped. Some people end up sucked into tar pits, or buried in avalanches that become part of the landscape, or fall into deep, remote crevices they can’t get out of. Whatever specific situation you imagine, there are some situations you can get in that will trap you in a nightmare of claustrophobia for tremendous, indefinite periods of time. It is very conceivable to me that scenarios like this exist which will keep you trapped for long enough in expectation, that the period you can expect to be trapped in such situations would, in total, exceed the expected gaps in between these periods of being trapped 2. This is especially notable in some types of scenarios for being trapped, like being lost drifting in space. After a period long enough for those searching to forget about you, I can’t even imagine how long you would be expected to drift, quite possibly through the heat death of the universe.

While I think the savvier immortalists imagine ways of life that dodge these specific risks like simulations, the idea of risks for long-term entrapment of this sort is worth worrying about more generally, even if we can’t imagine the specific cases that go with a given way of life. Another way to think of this risk is that immortality forces you to pay attention to Pascal’s Mugging type extremes of expected value. While normally a given very small risk will just be sailed past, and seem unreal, infinite time spans make the actual results inevitably approach the expected value, including in the obscure cases we normally laugh off as defects of blind expected value. Abolishing a physical limit like death makes the existence of edge-cases like these, even unknown ones, far more concerning. It is undeniable that actual immortality presents a possible route for S-Risks, and, especially if you give some credence to a pleasure/pain asymmetry, this is worth taking very seriously. As I personally have no sympathy for forced immortality anyway, this is not something I devote great thought to. If you believe in forced immortality, it seems to present an especially good reason not to be smugly dismissive of counter-arguments. To take seriously that there are worse things than death out there, and that the worst of them are often limited by death under normal circumstances.

Not only do I think that, even for immortality critics who say things like “death gives life meaning”, this indefinite life ideal is the most common belief when pressed, I think that life-extension like this opens up unusually counter-intuitive ethical principles as well. Affirming the indefinite life line is, functionally, saying that in an ideal world everyone will die of suicide. This is rarely highlighted by immortalists making arguments like these, but there are several coherent positions here, and none of them give you everything you would expect of an idea that claimed to be totally commonsense. You could have involuntary mortality, voluntary mortality, involuntary immortality, voluntary immortality, or voluntary death. The best of these seems to me to be voluntary death, that when death happens, it is technically some form of suicide.

Voluntary mortality is fairly similar, it is when someone eventually or from the beginning chooses to be mortal, able to die, but doesn’t explicitly off themself. To some this will seem different, better to some degree, I agree that it should be an option, but it is hard to imagine a uniquely good ethical reason for it. A sense that it is the natural way to die seems like it might be common, but of course that just means someone has committed suicide by “natural” death. Alternatively someone might want to die soon, but not have the stomach to do it themself. This also seems like another form of suicide, when there is an alternative. Basically, however I frame voluntary mortality, it seems like suicide with some personally appealing feature, and a few added inconveniences for planning and autonomy.

Someone who subscribes to an act/omission view of morality might say that it is preferable to suicide, but the fact that it introduces more steps and more uncertainty doesn’t seem to be enough to satisfy the principle. Someone has to start out mortal, and opt to remain mortal for this to be enough of an omission to avoid becoming a form of disguised suicide. The idea that someone can avoid being guilty of suicide only by doing something that would shorten their life while not changing how they end up dying seems profoundly unappealing from any commonsense morality perspective (though I’m not the best person to ask for a sympathetic account of the act/omission principle anyway).

Involuntary mortality, by contrast, is the current situation. People will die whether they want to or not, without planning for when it happens. Let’s say that we have the means to allow people to choose to survive what would normally be their death, can we be justified in denying them this choice to survive? This seems unthinkable if keeping them alive hurts no one else. If someone could be allowed to survive some possible death, and wants to survive, you seem to have little excuse to let them die, certainly not just to prevent them from killing themself later. An example for thinking about this is as follows, if we cure ageing at some point, we won’t have cured accidents, people could still get stabbed to death or fall off a cliff. If this makes us feel relief because we can avoid death being voluntary, we should be able to bite the bullet that, if we found cures for knives and cliffs, we had better keep people in the dark.

Then we have voluntary immortality and involuntary immortality. These are the possibilities I have already discussed at length. Voluntary immortality seems like the nicest option out of the ones I provided, but for the reasons I mentioned, it seems hopelessly unrealistic. Involuntary immortality seems to range between someone’s ugly compromise scenario and someone else’s eldritch nightmare. I lean towards the latter.

Mind you, the idea that “voluntary death” presents the best case scenario out of these isn’t a complete answer. Just because every ideal death is suicidal doesn’t mean that every suicide is an ideal death, but immortalism opens up difficult and speculative bioethical questions concerning suicide that deserve their own article, and are likely to upend at least some current intuitions about the ethics of suicide/euthanasia. It should certainly be seen as more challenging to current norms than it seems to be that “The Good Place”, a fairly mainstream and inoffensive show, ended with an ideal world in which everyone commits suicide.

There is one possible escape route, which I haven’t seen developed, which says the following:

  1. There are circumstances under which suicide must be allowable in indefinite life, but this is not every circumstance, and it is possible to reduce the prevalence of these circumstances.

  2. There are circumstances under which any given “immortal” could still die, but we can approach having fewer and fewer such fatal circumstances.

  3. While voluntary immortality does not seem possible in theory, we can approach it by making it less and less likely that a situation causing either voluntary or involuntary death will arise.

This is a nice thought, perhaps not spelled out enough for implementation, but worth entertaining, especially for those who see the given options and find voluntary immortality to be worlds better than the others. While it still seems to me that indefinite life is the best possibility, the fact that it seems to bring up such counterintuitive implications on sensitive topics in ethics contradicts the implications of much of this media. Yudkowsky’s Rational!Harry describes fearing death as being like fearing a “great big monster with poisonous fangs”, Bostrom’s depiction of death (ok aging) is as a dragon. Both works feature a child making an emotional appeal to common sense against a pretentious authority figure. Whatever argumentative aesthetics Yudkowsky and Bostrom may hope to cultivate, discussing what is involved in ending death is not a matter of common sense, it is an exercise in highly conceptual speculations and bullet biting. To insist otherwise is not only insulting, it is wrong.

But we can put this aside for a second, to return to the assumption that, more or less, everyone agrees on the ideal policy. As I said, it seems to me like this is roughly true. Is there still something to be angry about or resist? Implementation matters. An immortalist might say that this research should be one of our biggest priorities (most of the linked pieces of media point out that some people will make it to indefinite life, while some won’t, and the priority given to this issue matters very materially for those people at the edge of this), a skeptic might say that the issue is not worth working on at all, that it will never be worth the tradeoffs that would be necessary, or would never be likely enough to work. Do these differences give a justification for the attitudes of Rationalists towards this issue?

The last view, that this will never be likely to work, seems to clearly not be worth anger. It is a factual mistake at worst. The practical concerns, those that might relate to how power structures will shift, or what other causes research in this area is competing for resources with, are largely ignored (with the exception of the Sandberg interview) by the cited Rationalist media. While it is hard to call the difference in lives lost involuntarily between an earlier and a later cure insignificant, treating this as justification in itself for overriding priority means claiming that the concerns people might have (such as the unpredictable worst-case consequences for power-structures) and the other causes they might be prioritizing over indefinite life (such as saving QUALYs with existing medicine, combating factory farming, and mitigating long-term risks that would by the way threaten the project of indefinite life as well) cannot be morally comparable.

The effects of this delay are no more direct than the effects of neglecting any of these other issues, all of these issues are mostly comparable on consequentialist terms, which is to say empirical. If there was a deep difference in what people thought was a theoretically desirable policy, as I have denied is usually the case, the anger might derive from the recognition of some deep, principled issue with discourse. As the area it seems most of the material disagreement can be located is an area where deep empirical uncertainties are relevant and the counter-considerations are arguably comparably strong, a careful and humble tone seems far more appropriate for the issue than what I see.

  1. Ed. Note: TMB Trivia: I originally bought to blog about education and pedagogy. I’d been researching many things about pedagogy from different sources (mostly popular books…), and I wanted to try and spend a year writing a blog about the key ideas, in my spare time. Then Ultralearning came out and summarized basically all the things I’d seen anyway, so I used the domain name for this blog. ↩︎

  2. Ed. Note: This probably happens all the time in 17776, unless I misinterpreted what the nanobots are doing. ↩︎