(I’m not at Rethink Priorities anymore, and I’m not speaking on their behalf.)
RP did in fact respond to some of these arguments, in the piece Do Brains Contain Many Conscious Subsystems? If So, Should We Act Differently?, on which I am a co-author.
If we’re talking about the need for more workers on timelines of decades, advancements in AI and automation will be relevant. Will immigrants be necessary in decades?
Also, fertility rates are falling pretty quickly globally, so immigration might only help so much for so long on its own.
There are multiple views considered “person-affecting views”, and I think the asymmetry (or specific asymmetric views) is often considered one of them. What you’re describing is a specific narrow/strict person-affecting restriction, also called presentism. I think it has been called the person-affecting view or the person-affecting restriction, which is of course confusing if there are multiple views people consider person-affecting. The use of “person-affecting” may have expanded over time.
It’s worth pointing out that ACE’s estimates/models (mostly weighted factor models, including ACE’s versions of Scale-Tractability-Neglectedness, or STN) are often already pretty close to being BOTECs, but aren’t quite BOTECs. I’d guess the smallest fixes to make them more scope-sensitive are to just turn them into BOTECs, or whatever parts of them you can into BOTECs, whenever not too much extra work. BOTECs and other quantitative models force you to pick factors, and scale and combine them in ways that are more scope-sensitive.
For the cost-effective criterion, ACE makes judgements about the quality of charities’ achievements with Achievement Quality Scores. For corporate outreach and producer outreach, ACE already scores factors from which direct average impact BOTECs could pretty easily be done with some small changes, which I’d recommend:
Score “Scale (1-7)” = “How many locations and animals are estimated to be affected by the commitments/campaign, if successful?” in terms of the number of animals (or animal life-years) per year of counterfactual impact instead of 1-7.
Ideally, “Impact on animals (1-7)” should be scored quantitatively using Welfare Footprint Project’s approach (some rougher estimates here and here) instead of 1-7, but this is a lower priority than other changes. Welfare improvements per animal or per year of animal life can probably vary much more than 7 times, though, and can end up negative instead, so I’d probably at least adjust the range to be symmetric around 0 and let researchers select 0 or values very close to it.
The BOTEC is then just the product of “Impact on animals (1-7)” (the average welfare improvement with successful implementation), “Scale”, “Likelihood of implementation (%)”, expected welfare range and the number of years of counterfactual impact (until similar welfare improvements for the animals would have happened anyway and made these redundant). Similar BOTECs could be done for the direct impacts of other interventions.
For groups aiming to impact decision-making or funding in the near term with research like Faunalytics, ACE could also highlight some of the most important decisions that have been (or are not too unlikely to be) informed by their research so that we can independently judge how they compare to corporate outreach or other interventions. ACE could also use RP’s model or something similar to get impact BOTECs to make comparisons with more direct work.
For other charities, ACE could also think about how to turn the models into BOTECs or quantitative models of important outcomes. These can be intermediate outcomes or outputs that aren’t necessarily comparable across all interventions, if impact for animals is too speculative, but the potential upside is high enough and the potential downside small enough.
For the Impact Potential criterion, ACE uses STN a lot and cites the 80,000 Hours’ article where 80,000 Hours explains how to get a BOTEC by interpreting and scoring the factors in specific ways. ACE could just follow that procedure and then the STN estimates would be BOTECs.
That being said, STN is really easy to misapply generally (e.g. various critiques here), and I’d be careful about relying on it even if you were to follow 80,000 Hours’ procedure to get BOTECs. For example, only a tiny share of a huge but relatively intractable problem, like wild animal welfare/suffering, may be at all tractable, so it’s easy to overestimate the combination of Scale and Tractability in those cases. See also Joey’s Why we look at the limiting factor instead of the problem scale and Saulius’s Why I No Longer Prioritize Wild Animal Welfare. STN can be useful for guiding what to investigate further and filtering charities for review, but I’d probably go for BOTECs done other ways, like above to replace Achievement Quality Scores, and with more detailed theories of change.
For example, you could do a BOTEC of the number of additional engagement-weighted animal advocates, which could be part of a BOTEC for impact on animals, but going from engagement-weighted animal advocates to animals could be too speculative, so you stop at engagement-weighted animal advocates. This could be refined further, weighing by country scores.
Per animal or per animal life-year, to match Scale.
It seems ACE did so for the Scale factor, but no specific quantitative interpretation for the others.
I think there are some interesting arguments here, but the argument in “4.3 Computational Equivalence” can probably still be saved, because it shouldn’t depend on any controversial parts of computational theories.
Instead, imagine two identical brains undergoing the same physical events, but one doing so at twice the speed and over a period of time half as long. Neural signals travel twice as fast, the time between successive neuron firings is halved, etc..
In my view, any plausible theory of consciousness and moral value should assign the same inherent hedonistic value to the two brains over their respective time intervals. Computational theories are just one class of theories that do. But we can abstract away different physical details.
On the other hand, I can imagine two people with identical preferences living (nearly) identical lives over the same objective time intervals (from the same reference frame), but one with twice the subjective rate of experience as the other, and this making little difference to the moral value on preference accounts. Each person has preferences and goals like getting married, having children, for there to be less suffering in the world, etc., and while how they care subjectively about those matters, the subjective rate of experience doesn’t make their preferences more or less important (at least not in a straightforward multiplicative way). Rather, we might model them as having preferences over world states or world histories, and their subjective appreciation of how things turn out isn’t what matters, it’s just that things turn out more or less the way they prefer.
Maybe the thought experiment is in fact physically impossible, except through relativity, which the author addresses, but I don’t think the right theory of consciousness should depend on those details, and we can make the brains and events different enough while still preserving value and having corresponding events unfold at twice the speed.
And I’d guess the same subjective value from the perspectives of those brains in most cases, but people can care about their relationships with the world in different ways that we might care about, e.g. maybe they want the timing of their subjective experiences or neural activity to match some external events. That seems like an odd preference, but I’m still inclined to care about it.
It seems quite likely to me that all the results on creatine and cognition are bogus, maybe I’d bet at 4:1 against there being a real effect >0.05 SD.
Unless I’m misunderstanding, does this mean you’d bet that the effects are even smaller than what this study found on its preregistered tasks? If so, do you mind sharing why?
Does this study tell us much about the counterfactual advancement of policies that pass the threshold by more significant margins, like a few percentage points or even double digit percentage points? Presumably those are more popular, so more likely to be passed eventually anyway. Some might still be popular but neglected because they aren’t high priorities in politics, though, e.g. animal welfare.
I think if there’s anything they should bother to be publicly transparent about in order to subject to further scrutiny, it’s their biggest cruxes for resource allocation between causes. Moral weights, theory of welfare and the marginal cost-effectiveness of animal welfare seem pretty decisive for GHD vs animal welfare.
There are other simple methodologies that make vaguely plausible guesses (under hedonism), like:
welfare ranges are generally similar or just individual-relative across species capable of suffering and pleasure (RP’s Equality Model),
the intensity categories of pain defined by Welfare Footprint Project (or some other functionally defined categories) have similar ranges across animals that have them, and assign numerical weights to those categories, so that we should weigh “disabling pain” similarly across animals, including humans,
the pain intensity scales with the number of just-noticeable differences in pain intensities away from neutral across individuals, so we just weigh by their number (RP’s Just Noticeable Difference Model).
In my view, 1, 2 and 3 are more plausible and defensible than views that would give you (cortical or similar function) neuron counts as a good approximation. I also think the actually right answer, if there’s any (so excluding the individual-relative interpretation for 1), will look like 2, but more complex and with possibly different functions. RP explicitly considered 1 and 3 in its work. These three models give chickens >0.1x humans’ welfare ranges:
Model 1 would give the same welfare ranges across animals, including humans, conditional on capacity for suffering and pleasure.
Model 2 would give the same sentience-conditional welfare ranges across mammals (including humans) and birds, at least. My best guess is also the same across all vertebrates. I’m less sure that invertebrates can experience similarly intense pain even conditional on sentience, but it’s not extremely unlikely.
Model 3 would probably pretty generally give nonhuman animals welfare ranges at least ~0.1x humans’, conditional on sentience, according to RP.
You can probably come up with some models that assign even lower welfare ranges to other animals, too, of course, including some relatively simple ones, although not simpler than 1.
Note that using cortical (or similar function) neuron counts also makes important assumptions about which neurons matter and when. Not all plausibly conscious animals have cortices, so you need to identify which structures have similar roles, or else, chauvinistically, rule these animals out entirely regardless of their capacities. So this approach is not that simple, either. Just counting all neurons would be simpler.
(I don’t work for RP anymore, and I’m not speaking on their behalf.)
Although we could use a different function of the number instead, for increasing or diminishing marginal returns to additional JNDs.
Maybe lower for some species RP didn’t model, e.g. nematodes, tiny arthropods?
They won’t be literally identical: they’ll differ in many ways, like physical details, cognitive expression and behavioural influence. They’re separate instantiations of the same broad class of functions or capacities.
I would say the number of times a function or capacity is realized in a brain can be relevant, but it seems pretty unlikely to me that a person can experience suffering hundreds of times simultaneously (and hundreds of times more than chickens, say). Rethink Priorities looked into these kinds of views here. (I’m a co-author on that article, but I don’t work at Rethink Priorities anymore, and I’m not speaking on their behalf.)
FWIW, I started very pro-neuron counts (I defended them here and here), and then others at RP, collaborators and further investigation myself moved me away from the view.
We have a far more advanced consciousness and self awareness, that may make our experience of pain orders of magnitude worse (or at least different) than for many animals—or not.
I agree that that’s possible and worth including under uncertainty, but it doesn’t answer the “why”, so it’s hard to justify giving it much or disproportionate weight (relative to other accounts) without further argument. Why would self-awareness, say, make being in intense pain orders of magnitude worse?
And are we even much more self-aware than other animals when we are in intense pain? One of the functions of pain is to take our attention, and it does so more the more intense the pain. That might limit the use of our capacities for self-awareness: we’d be too focused on and distracted by the pain. Or, maybe our self-awareness or other advanced capacities distract us from the pain, making it less intense than in other animals.
(My own best guess is that at the extremes of excruciating pain, sophisticated self-awareness makes little difference to the intensity of suffering.)
The old cortical neuron count proxy for moral weight says that one chicken life year is worth 0.003, which is 1/100th of the RP welfare range estimate of 0.33. This number would mean chicken interventions are only 0.7x as effective as human interventions, rather than 700x as effective.
700/100=7, not 0.7.
But didn’t RP prove that cortical neuron counts are fake?Hardly. They gave a bunch of reasons why we might be skeptical of neuron count (summarised here). But I think the reasons in favour of using cortical neuron count as a proxy for moral weight are stronger than the objections.
But didn’t RP prove that cortical neuron counts are fake?
Hardly. They gave a bunch of reasons why we might be skeptical of neuron count (summarised here). But I think the reasons in favour of using cortical neuron count as a proxy for moral weight are stronger than the objections.
I don’t think the reasons in favour of using neuron counts provide much support for weighing by neuron counts or any function of them in practice. Rather, they primarily support using neuron counts to inform missing data about functions and capacities that do determine welfare ranges (EDIT: or moral weights), in models of how welfare ranges (EDIT: or moral weights) are determined by functions and capacities. There’s a general trend that animals with more neurons have more capacities and more sophisticated versions of some capacities.
However, most functions and capacities seem pretty irrelevant to welfare ranges, even if relevant for what welfare is realized in specific circumstances. If an animal can already experience excruciating pain, presumably near the extreme of their welfare range, what do humans have that would make excruciating pain far worse for us in general, or otherwise give us far wider welfare ranges? And why?
There’s a related tag Meat-eater problem, with some related posts. I think this is less worrying in low-income countries where GiveWell-recommended charities work, because animal product consumption is still low and factory farming has not yet become the norm. That being said, factory farming is becoming increasingly common, and it could be common for the descendants of the people whose lives are saved.
Then, there are also complicated wild animal effects from animal product consumption and generally having more humans that could go either way morally, depending on your views.
The arguments in Reasons to doubt that suffering is ontologically prevalent by @Magnus Vinding have made me more skeptical of panpsychism, especially this excerpt:
Counterexamples: People who do not experience pain or sufferingOne argument against the notion that suffering is ontologically prevalent is that we seem to have counterexamples in people who do not experience pain or suffering. For example, various genetic conditions seemingly lead to a complete absence of pain and/or suffering. This, I submit, has significant implications for our views of the ontological prevalence (or non-prevalence) of suffering.After all, the brains of these individuals include countless subatomic particles, basic biological processes, diverse instances of information processing, and so on, suggesting that none of these are in themselves sufficient to generate pain or suffering.One might object that the brains of such people could be experiencing suffering — perhaps even intense suffering — that these people are just not able to consciously access. Yet even if we were to grant this claim, it does not change the basic argument that generic processes at the level of subatomic particles, basic biology, etc. do not seem sufficient to create suffering. For the processes that these people do consciously access presumably still entail at least some (indeed probably countless) subatomic particles, basic biological processes, electrochemical signals, different types of biological cells, diverse instances of information processing, and so on. This gives us reason to doubt all views that see suffering as an inherent or generic feature of processes at any of these (quite many) respective levels.Of course, this argument is not limited to people who are congenitally unable to experience suffering; it applies to anyone who is just momentarily free from noticeable — let alone significant — pain or suffering. Any experiential moment that is free from significant suffering is meaningful evidence against highly expansive views of the ontological prevalence of significant suffering.
One argument against the notion that suffering is ontologically prevalent is that we seem to have counterexamples in people who do not experience pain or suffering. For example, various genetic conditions seemingly lead to a complete absence of pain and/or suffering. This, I submit, has significant implications for our views of the ontological prevalence (or non-prevalence) of suffering.
After all, the brains of these individuals include countless subatomic particles, basic biological processes, diverse instances of information processing, and so on, suggesting that none of these are in themselves sufficient to generate pain or suffering.
One might object that the brains of such people could be experiencing suffering — perhaps even intense suffering — that these people are just not able to consciously access. Yet even if we were to grant this claim, it does not change the basic argument that generic processes at the level of subatomic particles, basic biology, etc. do not seem sufficient to create suffering. For the processes that these people do consciously access presumably still entail at least some (indeed probably countless) subatomic particles, basic biological processes, electrochemical signals, different types of biological cells, diverse instances of information processing, and so on. This gives us reason to doubt all views that see suffering as an inherent or generic feature of processes at any of these (quite many) respective levels.
Of course, this argument is not limited to people who are congenitally unable to experience suffering; it applies to anyone who is just momentarily free from noticeable — let alone significant — pain or suffering. Any experiential moment that is free from significant suffering is meaningful evidence against highly expansive views of the ontological prevalence of significant suffering.
Also, if we defer to people’s revealed preferences, we should dramatically discount the lives and welfare of foreigners. I’d guess that Open Philanthropy, being American-funded, would need to reallocate much or most of its global health and development grantmaking to American-focused work, or to global catastrophic risks.
EDIT: For those interested, there’s some literature on valuing foreign lives, e.g. https://scholar.google.ca/scholar?hl=en&as_sdt=0%2C5&q=“valuing+foreign+lives”+OR+”foreign+life+valuation”
Global catastrophic risks might already do the job.
I think that’s basically right, but also rejecting unitarianism and discounting other animals through this seems to me like saying the interests of some humans matter less in themselves (ignoring instrumental reasons) just because of their race, gender or intelligence, which is very objectionable.
People discount other animals because they’re speciesist in this way, although also for instrumental reasons.
On the other hand, if I could persuade ten people to donate to effective animal charities or one to devote their career to helping animals, I’d probably prioritize the latter.
These ten people donating to effective animal charities can, through their donations, bring in more people to devote their careers to helping animals. They could do that by paying enough for one more person to work full-time in the movement, or by paying for further outreach.
Difference-making risk aversion (the accounts RP has considered, other than rounding/discounting) doesn’t necessarily avoid generalizations of (2), the 50-50 problem. It can
just shift the 50-50 problem to a different place, e.g. 70% good vs 30% bad being neutral in expectation but 70.0001% being extremely good in expectation, or
still have the 50-50 problem, but with unequal payoffs for good and bad, so be neutral at 50-50, but 50.0001% being extremely good in expectation.
To avoid these more general problems within standard difference-making accounts, I think you’d need to bound the differences you make from above. For example, apply a function that’s bounded above to the difference, or assume differences in value are bounded above).
On the other hand, maybe having the problem at 50-50 with equal magnitude but opposite sign payoffs is much worse, because our uninformed prior for the value of a random action is generally going to be symmetric around 0 net value.
Assume you have an action with positive payoff x (compared to doing nothing) with probability p=50.0001%, and negative payoff y=-x otherwise, with x very large. Then
Holding the conditional payoffs x and -x constant, but changing the probabilities at 100% x and 0% y=-x, the act would be good overall. OTOH, it’s bad at 0% x and 100% y=-x. By Continuity (or the Intermediate Value Theorem), there has to be some p so that the act that’s x with probability p and y=-x with probability 1-p is neutral in expectation. Then we get the same problem at p, and a small probability like 0.0001% over p instead of p can make the action extremely good in expectation, if x was chosen to be large enough.
Holding the probability p=50% constant, if the negative payoff y were actually 0, and the positive payoff still x and large, the act would be good overall. It’s bad for y<0 low enough. Then, by the Intermediate Value Theorem, there’s some y so that the act that’s x with probability 50% and y with probability 50% is neutral in expectation. And again, 50.0001% x and otherwise y can be extremely good in expectation, if x was chosen to be large enough.
Each can be avoided if the adjusted value of x is bounded and the bound is low enough, or x itself is bounded above with a low enough bound.
I think the same would apply to difference-making ambiguity aversion, too.
y=-x if difference-making risk averse, any y< -x if difference-making risk neutral, and generally for some y<0 if the disvalue of net harm isn’t bounded and the function is continuous.