Point taken. I guess the main difference for me is that with age reversal, we’ve got a framework that makes sense and de-mystifies it, and it implies that if we just do this and this and this and this then the problem should be solved. We can actually see a path between where we are right now and where we want to be. Do we have something like that with AGI, or is it more a matter of “these language models are starting to look pretty smart”? I’m not saying I would bet against AGI, the way things are going… but I wouldn’t want to rely on it without a rational model of it, either.
PhilJackson
Karma: 131
The Fountain of Health: a First Principles Guide to Rejuvenation
There are papers showing senescent cells in humans with signs of shortened telomeres and replicative stress (e.g. [1,2]), so I wouldn’t say the Hayflick limit is nonsense or irrelevant in humans. But there is perhaps a broader point that the Hayflick limit / telomere erosion has been over-hyped as the fundamental driver of ageing. Case in point, I once talked to a guy who told me “Immortality is impossible, because: telomeres”.
The problem is that telomere erosion tells a simple and compelling story about what ageing is, like every cell has a built-in clock that prevents us living too long. From skimming the article, the point Ray Peat seems to be making, in a rather snarky and conspiratorial way, is that it’s much more complicated than that, which is true. The point I’m making in this post is that there is no primary mechanism or process which is or causes ageing. Ageing is driven by a multitude of unavoidable cellular and molecular changes, but the list is finite, and fixing it is a tractable engineering problem.
The one that always come to mind is: how will we spare space for the youths?
I’m unusually chill when it comes to population, both over and under. I’m not enormously concerned by how many people there are, I’m more concerned by how healthy the existing people are, and how healthy the environment is, both of which are mostly orthogonal to population and depend primarily on technological development. It’s important to note that fertility rates are generally declining worldwide, seemingly in tandem with economic development. The most economically developed countries generally have below-replacement fertility rates, and are facing depopulation plus highly inverted population pyramids (high ratio of sick, dependent, aged people) if something doesn’t change. In places where fertility is below two children per woman, you could abolish death completely and population will plateau, since each generation produces fewer children than the one before, so you get a geometric series with r < 1. Thus, the only study I’m aware of that analyses the demographic consequences of rejuvenation therapies predicts only slow and limited population growth (at least in Sweden).
Of course, it all depends how fertility changes. Ending reproductive senescence could allow people to postpone childbirth even further than they already do, lowering fertility rates further, or it could lead to people having more children throughout their greatly extended lifetimes. I suspect the former will occur, but time will tell.
the biologist in me is very confident that it’s terribly misleading to think biological ageing as a simple accumulation of damages
Oh, I hope I didn’t give the impression that this was simple. One of my core points is that ageing cannot be reduced to any single target for intervention, there is no way around the fact that ageing is many-faceted and will require complex combination therapies to address. I only assert that there will be some combination of damage which, if eliminated, will elicit healthy life extension, and that we have some well educated guesses as to what those are and how to go about fixing them, and that doing so can only be good for people even if damage identified by SENS platform is not in fact comprehensive.
Counter-counter-points:
Yes, cancer is hard and I glossed over it. Partly it will be mitigated by immune system rejuvenation, since immune surveillance plays a huge role in cancer suppression. Dealing with mitochondrial dysfunction should also reduce cancer incidence, by lowing the amount of free-radical assault on DNA. Aside from that though, there are exciting things happening in actual cancer therapy: cancer immunotherapy (works very well in blood cancers, might scale to solid cancers), THIO by MAIA Biotechnology (rapidly kills cells that express high levels of telomerase, which most cancers do by necessity), Lou Hawthorne’s NaNots (can safely remove stuff from the blood, can be used to soak up the soluble immune system inhibitors that tumours have to use to protect themselves from the immune system, and finally, Michael Levin’s lab have managed to suppress cancers using ion channel drugs, along with opening a whole new perspective on what cancer fundamentally is.
The effects rejuvenating effects of parabiosis are due to either good things in the young blood, or dilution / removal of bad things (e.g. inflammatory signalling molecules) in the old blood. The debate continues, but I believe consensus is leaning towards the latter, since you can achieve similar results just by diluting the blood with saline-albumin [1,2].
Function follows form. If there are diseases that aren’t associated with any obvious damage, then that only means we’ve yet to identify what the damage is. Machines do what they do because their structure; if they malfunction, then there must some change somewhere in the structure that causes it, and whatever that change is, we call it damage. Remember that damage is defined very broadly, it doesn’t have to be what you’d think of as “wear-and-tear” necessarily—it could be something like over-sensitisation of a particular type of neuron in the case of CRPS, for example.
Progeria (assuming you mean Hutchinson-Gilford syndrome) is a mutation in a critical nuclear envelope protein which causes rapid and widespread damage of many forms, since it’s such a critical component that’s defective. It’s thought of as “premature ageing” since its symptoms resemble accelerated normal ageing, probably because “widespread damage of many forms” is what normal ageing is, but that doesn’t mean that the damage involved in progeria is the same as in normal ageing (though there’s probably some overlap). As for menopause, it’s kind of a moot point to me whether a change is “under genetic control” (difficult to define since everything is ultimately determined by genes anyway) or not. Thymic involution appears to be a programmed developmental change, but we still call it damage, because it causes a loss of function (inability to make new naive T-cells). We don’t much care how damage happens, only how to reverse it.
You may well turn out to be correct—biology is indeed fiendishly complex and still very poorly understood, and it may turn out that clearing the damage types identified by the SENS platform is insufficient for comprehensive rejuvenation. However, I’ve tried to separate that particular claim from a different, more defensible, more immediately relevant one: that damage repair should become our main approach to medical research in a world where infectious disease is largely conquered, and most suffering is due to (mostly age related) chronic illnesses. Most bodily disorders these days are caused not by infection by an external pathogen, but by damage to the body itself, most of which is self-inflected by ordinary metabolic processes. Yet for some reason, we continue to treat the (often arbitrarily defined) diseases themselves, as though they were infectious diseases, instead of even trying to repair the damage that causes them. This must change, and it is changing. To me, damage repair is a first and foremost a new, fundamentally different approach to medicine, one that emphasizes fixing things that are obviously broken, which I expect to work much better than the old paradigm of treating diseases separately. The whole “ageing” thing is almost secondary to me. We’ve literally not been fixing people’s bodies this entire time, and now people are finally trying to fix a bunch of obviously broken things, that’s why I’m excited.
If we define damage as material defects, which I think is closer to your original intent, then maladaptative memories (of the affective system in PTSD, of the pain system in CRPS, of the immune system….) can cause a loss of function without a lasting role for material damages
“Material”? I don’t think it’s useful to distinguish between material and immaterial change, the point is that the change is maladaptive. If your hard drive gets corrupted preventing your computer from booting then we still call that damage, even though it seems less material than, say, shooting the PSU. And ultimately, data and code are material anyway, whether they’re represented by patterns of magnetic polarisation or changes to neuronal sensitivity or connectivity or whatever.
Yes we’ve always been trying to fix damage—anything that restores function must fix damage somehow—but it’s a matter of what we consider to be damage, i.e. “bad stuff that we should try to fix because it would restore function”. Historically we’ve focused on trauma, infection and cancer, and although we’ve known about age-related changes like lipofuscin accumulation for a long time, it’s only recently that we started thinking of them as potential targets. Gerontology has historically been a field of basic science, with few gerontologists willing to venture that we could even in principle do something about age-related changes until very recently. They were too afraid to challenge the deeply held notion that ageing is normal, natural and fundamentally immutable. I’m not without sympathy though—anyone who had made that claim would have been attacked and likely would have lost their livelihood—Aubrey de Grey was only able to do it because he was financially independent.
If you look at things like atherosclerosis and AMD, both of which have long been known to be driven by the accumulation of toxic metabolites, I think we could and should have started working on the root causes of these diseases much earlier than we did. Sure, we’re only now figuring out how to remove 7KC and A2E, but that’s because we only just started working on it! Likewise, we’ve known about thymic involution for a long time and it’s always been clear that it damages your immune system in the long run, so why is it that only Greg Fahy is working on fixing it, and why does he get so little funding that his trials have to be funded by their own participants? We declared war on cancer long ago, when our models and tools were still hopelessly crude, and because of that we are further advanced now than we would have been otherwise.
Point taken that this is all extremely difficult and ambitious, and it’s not entirely unreasonable that researchers have been intimidated by that. Rejuvenation is indeed highly dependent on very recent technological breakthroughs, and those new capabilities are a huge part of why this shift is occurring. In my defense, there is a bit where I say
I don’t mean to imply that prior researchers had been stupid here; humans don’t come equipped with enzymes capable of degrading A2E so it wasn’t exactly obvious how to get rid of it.
Perhaps I could have stressed that more. To whatever extent there is blame, I place it more on grant committees than on medical researchers. The stuff SENS works on is all high-risk high-reward, which public purse-holders and investors are both cagey about—that’s why it all has to be philanthropically funded. For example, the idea to copy mitochondrial DNA into the nucleus where it is safer (which I ignored completely in the post) has been around for a long time, but SENS had to work on it because no one else would.
Anyway, I don’t want to fixate on whose fault things were. The point is that for whatever reason, we were pursuing a broken paradigm of “one disease one target one drug”, which was never going to work because the diseases were not naturally separable things and their root causes mostly lay in fundamental age-related changes, which we are now beginning to target, and it seems likely to work.
Always glad to see Greg Egan referenced. The important thing to me is that although population growth could be exponential (for the reasons cousin_it gave), it’s going to be very slow relative to the rate of technological progress. Unless fertility rises significantly, it’s likely to be hundreds of years before population would grow by 10x, by which point we’re well into Greg Egan territory and all bets are off anyway. So population could be a concern, but we’ll have plenty of time to address it via methods that don’t involve literally everyone dying.
It’s both prevention and cure—prevent disease by reversing damage before it gets bad enough to cause problems, but if you already have a chronic disease, then reversing the damage that causes it will be the only way to cure it (though prevention is better of course).
I agree the preventative medicine angle is a good one that people will buy easily, but you can make the same argument against it—that we’ve always been trying to prevent disease just as we’ve always been trying to fix damage.
It’s important to note that statins, blood thinners and rapamycin are not damage repair—they’re not useless, but I think damage repair will make them obsolete. These approaches focus on slowing the buildup of damage rather than reversing damage that’s already there. The problem is that a) this tries to modify human metabolism to “run more cleanly”, which is super difficult and prone to unforeseen consequences and b) you have to take these medicines every day, which makes it all the more dangerous. Statins are well known to have side effects, and rapamycin is an immunosuppressant, which unsurprisingly has a lot of side effects too. You don’t want to take this stuff every day.
The reason people associate things like statins with “being sick” is that they don’t actually make you less sick, they just slow the progression of a disease while causing side effects. Damage repair is far less prone to side effects because it targets stuff that’s definitely not supposed to be there instead of trying to change the way the body works. You won’t get side effects from removing atheromatous plaques (so long as that’s all the treatment does do), but you do get side effects from mucking around with liver chemistry. And because ageing damage accumulates so slowly, you’ll only have to take these therapies every 10 years or so (eventually, once they’re mature). And of course, because it reverses damage instead of merely slowing it, you’ll actually feel, look and be healthier and fitter after the treatment. That’s why I say it’s a new kind of medicine—the public are absolutely not used to medicine that makes them feel younger after they take it.
100% agree the messaging should focus on health rather than lifespan—not only because it’s far less controversial (most people want to be healthy), but because it’s true: we work directly on health, of which longevity is but a side effect. Glad you picked up on the multi-morbidity part too, tackling age-related sickness as a whole by focusing on fundamental ageing damage rather than treating diseases separately is crucial. Probably we’ll be talking about morbidity compression for a while yet; this is a crux that allows us to discuss medicine that actually works without having to acknowledge the Biblical-scale consequences of it. At some point though, laboratory results will become so compelling that the delusion collapses, and then all hell will break loose.
If it turns out healthspan aligns well with longevity
It can’t not. People don’t die because they get old, they die because they get sick.
The Senolytics section comes across a bit like all senescent cells are bad so if we could just kill them all we’d be doing a good thing. But my understanding is that they also play a role in healthy body functions. Is that also your understanding here?
From what I’ve read, the healthy roles they play are a) during early/embryonic development and b) in wound healing. So, don’t give senolytics to embryos or people healing from wounds I guess. Also, there’s a paper (summary here) showing that it’s possible to regain the wound-healing benefit of SCs in mice that are engineered not to create SCs, just by manually applying a single growth factor that SCs secrete into wounds. So it seems unlikely that the beneficial roles of SCs are a show-stopper, particularly because those beneficial effects are more associated with short-lived SCs rather than the ones that linger and accumulate.
Largely related to the above, the framing of the post seems that of body as a machine but I’m wondering if that is the best framing. Perhaps body as a ecosystem might be better. I don’t think it changes any of the main points you make (that I’ve read) but perhaps suggests following the machine metaphor might be limiting, and in some cases point in the wrong direction. Does that seem right to you?
I find it fascinating to think of the body as an ecosystem (e.g. cells competing for resources, occupying niches), but I use the machine analogy because it emphasizes that there’s an overall structure and function that can be degraded. What does it mean for an ecosystem to age or decline, how do you define the health of an ecosystem?
You can imagine an ecosystem <---> machine axis, where it’s more like an ecosystem when you have independent entities competing for their own survival, and more like a machine when they work more closely together to act as one. In this sense, I think metazoans are both ecosystem and machine. But much if not most of the damage that drives ageing is at the level of individual cells rather than whole organism, and it makes more sense to think of individual cells as machines.
One way in which the ecosystem view is helpful is that it emphasizes that dead or damaged cells (e.g. senescent cells) can be destroyed and replaced by the division of healthy cells. That’s something you couldn’t do in a non-ecosystem-like-machine, such as a car, and it’s a major benefit—replacement of old cells by new is really the best form of damage repair since you “repair” all the damage at once that way.
Indeed—people are finally thinking “what if ageing has something to do with all the age-related disease?” This is great, so long as you remember that “ageing” is not just one single root cause of age-related disease; rather, it’s a multitude of self-inflicted injuries the body slowly accumulates, which combine to make us frail and disease-prone.
Simulations of that fidelity level would indeed be ridiculously powerful tools, but I don’t know how long it’ll take to reach that level. Also, with a true full-body molecular simulation you’d have the ethical problem that a simulation at that level of detail is for all intents and purposes a human being, and may no longer be ethical to experiment on. The strength of damage repair as a medicinal paradigm is that it exposes a whole host of targets that we can safely go after, confident that doing so will improve function without having a full understanding of how the body works. Often we try to treat disease by changing the way the body works (e.g. statins), but this is very hard to do without side-effects because of how complex and inter-connected the body’s systems are. But the things I call “damage” are age-related changes that are unambiguously bad for you and can in principle be reversed. Doing so might not cure all age-related diseases, but it should prevent them all, since by definition they tend not to occur in people who don’t have a lifetime of age-related changes.
People are absolutely applying the centaur-intelligence thing to ageing already (e.g. Gero, Altos, In Silico), indeed I think all current applications of AI are working in synergy with human intelligence, and I hope it stays that way for as long as possible. It’s good that we are indeed able to benefit from (what currently passes for) AI long before it reaches the level where we can just ask ChatGPT “how do I cure ageing” and it just tells us.
What matters here is what we want. For now, I quite enjoy being made out of meat, and I think most other people do too. Our desires are what they are, and they don’t require justifying, to AGI or to anyone else. Of course, such transformative alterations/extensions to our selves will no doubt be unlocked one day, and I could happily live in a world where people modify themselves in all sorts of different ways, one step at a time, on their own terms, as and when they feel like it.
It’s almost certainly me that’s wrong here—there have been lots of different IFRs quoted so I picked 2% as the highest I could remember hearing, just to be conservative. Lower IFR would strengthen the argument I was making there, and indeed the true number is almost certainly much lower due to e.g. asymptomatic infections.
Not all cells in the adult body do divide, most of them I think divide only rarely if ever. Cells that divide more regularly generally express telomerase to keep their telomeres from running out. Telomeres running out may still be an issue though, so there are people looking at ways of lengthening them via telomerase expression.