Tentative Thoughts on the SENS Foundation
Edit from fucking 2019: this is all obviously wrong, EA is a scam, do all your own analysis of everything if you want to have an impact
This post has been cross-posted to LessWrong.
It should be emphasized that back-of-the-envelope calculations, such as the one given in this post, ought to be adjusted to account for the fact that interventions can look much more cost-effective than they are, especially when the interventions were only shallowly investigated.
Previously, Givewell has looked into the cost-effectiveness of life sciences funding, as well as publishing a simple estimate of the impact of the average dollar spent on cancer research, which suggested that, in the past, each $2790 spent on cancer-relevant biomedical research in the US added one year of life lived (YLL) to the life of a US resident. Givewell has also interviewed Aubrey de Grey of the SENS foundation. Owencb has previously estimated the cost-effectiveness of funding SENS/ anti-aging research as being around $50 per QALY. Aubrey de Grey has previously been averse to giving explicit cost-effectiveness estimates regarding how many QALYs would be gained per unit of funding supplied to SENS, though he has been clear that SENS’s funding needs are “$100 million per year for each of the next ten years”.
This part of the post will consist of me using lots of best guesses to produce something vaguely resembling a cost-effectiveness estimate for SENS. You should not take this cost-effectiveness estimate literally.
If SENS needs one billion dollars to ensure that rejuvenation technologies that give individuals 30 extra years of healthy life are available to the public in 30 years, we might (completely arbitrarily) assume that someone else will come along and fund SENS in ten years if we don’t contribute to funding SENS today. This means that if we fund SENS today instead waiting for it to be hypothetically funded in ten years from now, about ten times the number of people who die each year would live 30 years of healthy life that they wouldn’t have lived otherwise. Given that there are about 57 million deaths per year worldwide, this translates to about 17 billion YLLs lost by waiting ten years to fund SENS; since SENS ostensibly requires only 1 billion of philanthropic funding, this implies that $0.059 of funding for SENS produces a YLL.
Of course, regenerative medicine won’t be free to the people receiving it, and I have no idea how to account for this, given that I don’t have a good idea of how much regenerative therapies will initially cost. The above estimate hasn’t been adjusted to account for the fact that there is a time-delay between when funding is provided, and when the benefits of regenerative therapies are available to the public. Perhaps Aubrey isn’t well-calibrated, and the “$100 million per year for ten years” figure is entirely wrong. It may be the case that starting work on SENS’s research agenda earlier rather than later would allow certain people who would have otherwise died to live until aging escape velocity is reached, which would have lots of utility. There are plenty of other issues with this cost-effectiveness estimate which I am sure that readers could point out.
The point I wanted to make, though, was that maybe, possibly, SENS is competitive with GiveWell’s top charities—I’m legitimately not sure whether I would fund SENS or GiveWell if I were making a charitable donation today. Does anyone have any further thoughts on this topic?
I see two questions:
(1) Can QALYs be bought very cheaply (in expectation) through this research?
(2) Assuming they can, is this a good use of resources?
Most of the comment discussion so far is on (1), although Lila brings up something of (2).
On (1), I think your estimate looks significantly too optimistic overall, by erring on the side of optimism in most of the components. However, that isn’t nearly enough to sink the idea. Even a small chance of success could be enough to provide a high cost-effectiveness in expectation, which would justify putting more effort into seeing if we can be confident that the chance of success is very close to zero.
On (2), QALYs are an abstraction that work well because they provide a basis for comparison between different health states—different situations, but ones which often have substantial similarities. The effect of successful anti-ageing research would be different enough in kind that we might want to think about it from first principles rather than just using the QALY framework (although it’s an informative starting point).
A solution to ageing would increase the length of lives, and increase the length of productive lives. It isn’t obvious what the effect on total population would be: the direct effect would be to increase it, but the existence of this technology might well change the birth rate. Overall this seems to be good for the people living in this society.
A second effect might be to lengthen people’s time horizons. If global decision-makers expected to live another 50-100 years, we might see better coordination on global problems. This is also a very speculative effect, but seems to be positive.
Lila mentions cancer, which I think is very instructive in this context. A “war on cancer” was declared about 45 years ago. Since then, 100-300 billion dollars have been spent on cancer R&D. Very little progress has been made. Why should we expect a SENS-inspired “war on aging” to make lots of progress, on all seven causes of aging (of which cancer is just one), in one third of that time, with one hundredth of that budget?
EDIT: The paragraph above overstretches the analogy bewteen cancer and aging. See Gwern’s comment below.
Saying ‘very little progress’ seems to considerably understate it; many cancers are now treatable which were untreatable, and even former death sentences can be cured. As well, much of that research was spent in the past on expensive but obsolete methods or on building knowledge bases and tools which are now available for anti-aging research. (While Apollo may have cost $26b to put a man on the moon in 1969, it should not then cost another $26b in 2017 to put another man on the moon.)
Comparing with cancer is interesting in part because they’re so different. Cancer is a hostile self-reproducing ecosystem which literally evolves as it is treated; aging and senescent cells, however, appear to be none of those. For example, it appears to be a lot easier to trick a senescent cell30246-5 “‘Targeted Apoptosis of Senescent Cells Restores Tissue Homeostasis in Response to Chemotoxicity and Aging’, Baar et al 2017”) into committing suicide than a cancer cell.
Do you really need progress on all 7? Mortality with age follows a Gompertz distribution which has an exponential term increasing mortality risk and a baseline hazard/risk; interventions on the aging process itself, as opposed to tinkering with improved fixes for symptoms like cancer, would seem like they would affect the exponential term and not the hazard term. Since the Gompertz mortality curve is dominated by the exponential term, not the baseline hazard ratio, even small reductions in the aging rate lead to large changes in life expectancy. (In contrast, large reductions in the hazard ratio, like halving, only add a few years.)
Thanks for this. Your estimate is around three orders of magnitude more optimistic than mine, but not fundamentally different in kind. I tried to correct figures for something like an outside view, so that I was worried about being too pessimistic as well as worried about being too optimistic. I agree that the area is plausibly very high value and probably doesn’t receive enough attention from society. However it is speculative, and there may be even better opportunities if you are open to speculative areas.
Specific notes:
I read the $100 million per year figure as the funding required to give SENS a decent shot at it. We shouldn’t assume that this translates to success. In the interview where he provides that figure, de Grey talks about a 50% chance of success given this level of support. So I think you should at least reduce your estimate by a factor of 2.
While de Grey is one of the people most informed about this question, there is an effect where the people who are most optimistic about this line of research are the ones most likely to go into it. So I guess his estimate is more likely to err on the side of optimism than pessimism (this issue comes up with catastrophic risks, too …)
I was almost an order of magnitude more pessimistic than you on the benefits of success.
The biggest driver of difference between our estimates is the amount of relevant funding I guessed there already was in the area. This really was a guess on my part and I am not confident about it. Still, I am wary of assuming all the relevant work is coming through SENS.
I agree with everything in your two replies to my post.
You know, I’m probably more susceptible to being dazzled by de Grey than most—he’s a techno-optimist, he’s an eloquent speaker, he’s involved in Alcor, and I personally have a stake in life-extension tech being developed. I’m not sure how much these factors have influenced me in subtle ways while I was writing up my thoughts on SENS.
Anyhow, doing cost-effectiveness estimates is one of my favorite ways of thinking about and better understanding problems, even when I end up throwing out the cost-effectiveness estimates at the end of the day.
I’m really really skeptical of the claim that SENS can give every person on earth 30 additional years of healthy life for a billion dollars. Billions are spent annually on cancer research, and we still haven’t cured cancer.
There are two types of benefits of SENS’s research. First is the more mundane disease reduction stuff, which is a valuable way to promote quality of life, as GiveWell points out. However, there’s no need to focus on SENS specifically in finding cures for diseases.
Second is the life-extension stuff. Another way of understanding life-extension stuff is that it increases the number of people on earth at a given time, all else being equal. But the more obvious way to increase the number of people on earth is to promote births. Of course, there are transition costs to death: people really don’t like dying. On the other hand, there may be diminishing returns to life, and people might prefer to improve their chances of being born, whatever that means. I am floating in the ether and am offered a tradeoff: I can increase my probability of existing, but this will decrease the length of my existence if I receive an existence. I’m not sure what probability-length tradeoff I’d choose as optimal.
Wouldn’t it change the nature of life? What about the distribution of resources / elites—if elites could live longer than ordinary people, they would have a ridiculous advantage and may ensure their survival as a group above the needs of those without access to this therapy? Isn’t there something good about mortality—helps you appreciate the life you have etc.? Doesn’t aging serve a purpose in terms of a wisdom or responsibility marker / confidence reducer? Don’t all these questions assume that we can organise society rationally and either ensure that only people worthy of anti-aging get anti-aging, rather than what happens with most crucial technologies involving capital to produce: they are available disproportionately to those with the most resource in society and change the underlying power structure?
This sounds like the movie In Time: “In a future where time is literally money, and aging stops at 25, the only way to stay alive is to earn, steal, or inherit more time. Will Salas lives life a minute at a time, until a windfall of time gives him access to the world of the wealthy, where he teams up with a beautiful young heiress to destroy the corrupt system.”
The awful economics of ‘In Time’
Does that article deserve the title you gave it? It only would if the movie were presenting itself as a credible economic scenario, but its more like Swiftian satire.
Do you know of a link with a justification for thinking that SENS spending $1 billion over ten years would have a decent chance of curing ageing?
I haven’t found any such breakdown, even after looking around for a while. The 80,000 Hours interview with Aubrey, as well as a number of Youtube interviews featuring Aubrey (I don’t remember which ones, sorry) note that Aubrey thinks SENS could make good use of $1 billion over the next ten years, but none of these sources justify why this much money is needed.
My impression of the elite consensus on SENS is that many of their claims and strategies are highly over-optimistic, possibly so much so as to be totally impractical. (See e.g. the three debunking papers here, along with de Grey’s rebuttals and the authors’ counter-rebuttals.) I don’t know enough biology to vet people’s actual opinions, but is my impression of the elite consensus accurate?
A complication is that elites are in this context sometimes susceptible to what De Grey calls the “pro-aging trance” (a silly name for a real phenomenon), and are as such predisposed against his proposal for reasons that have little to do with its actual scientific merits (philosopher C. D. Broad describes a parallel case here). Maybe a better strategy for non-experts is to look for relevant historical analogies and try to extract a lesson from those. I do this in another comment, and the conclusion seems to be that De Grey is way over-optimistic.
Have you looked up Preston W. Estep III, the lead author of the judges’ pick from that contest?
I don’t think he (or probably the other gerontologist authors of those papers) could be said to be a victim of the “anti-aging trance.”
I agree that Estep doesn’t fit that description. In this particular case, however, I’d raise other concerns. I recall he described SENS as “agenda-driven pseudoscience”, which inclines me to think that he isn’t an objective critic of de Grey’s proposal.
I don’t really see your reasoning there. Plenty of objective critics of various things end up calling them agenda-driven pseudoscience.
I think Estep’s behavior is typical of an academic who is faced with an outsider making outrageous claims about how much better they’re going to be able to do than the rest of that field. I’ve seen multiple other examples of this, and the academics have usually been right.
My point wasn’t that calling something “agenda-driven pseudoscience” is in itself evidence that the critic is biased, but that this is so when there is sufficient independent evidence that the target doesn’t merit that criticism. De Grey has published dozens of papers in respectable scientific journals, is the editor of one such journal himself, has organized successful conferences attended by many field experts, has received praise from leading researchers, etc. I cannot think of many other people with comparable indicators of expertise whose research could be justifiably dismissed in such harsh terms.
(All of this, needless to say, is compatible with de Grey’s estimates being wildly optimistic, as I in fact believe they are.)
For one thing, the authors are talking about the SENS plan for curing aging, not de Grey’s past research; for another, they had $20,000 riding on whether they could successfully argue that the plan was “not even worthy of discussion” or something to that effect, so I think the harsh words are understandable.
Parapsychologists? Scientific racists? Climate change deniers? Tobacco deniers? Each of these fields has boosters as prominent as de Grey, if not far more (e.g. Daryl Bem, James Watson) and has been called out as “pseudoscience” by fairly objective-seeming critics.
The two are related, though I agree this limits the relevance of that particular indicator of expertise. (The successful conferences organized by de Grey, by contrast, are explicitly about SENS.)
The phrase I quoted was made by Estep in a paper that was unrelated to the Technology Review challenge, so I don’t think this point is relevant. Furthermore, if the financial incentives are operating in the manner you suggest, that would give us reason to distrust, to some degree, the claims made by Estep et al. in their submission to the challenge.
Our disagreement can probably be traced to some deeper epistemological difference between us: I suspect that many of the researchers you have in mind are scientists that I’d also think shouldn’t be dismissed in the manner Estep dismissed de Grey. But I would prefer not to spend my scarce weirdness points discussing publicly my views about these controversial figures.
Thanks for the link to the technology review! My reading is that the consensus is somewhere between “highly speculative, neither proven nor disproven” and “wildly over-optimistic”. Getting a better idea of which is more accurate could be quite valuable, as in the former case further work may well be worth funding, but in the latter case it probably isn’t.