Recurring issue I find with your posts is that you state expected values with huge uncertainty around them as if theyâre certainties.
E.g. âThe Shrimp Welfare Project (SWP) has been 64.3 k times as cost-effective as GWâs top charitiesâ
These estimates are built on other estimates each having their own error bars, sometimes with some assumptions thrown in. These EVs are houses of cards that shouldnât be taken very seriously at all.
While this is true, I also think itâs worth considering that this is often a criticism of any CEA, period. To the average person, the suggestion that a GiveWell top-recommended charity is more cost-effective than, say, a local food kitchen similarly requires estimates with error bars.
Yes, there are more assumptions when dealing with animals given welfare ranges, but I am reluctant to dismiss the analysis entirely because of that.
It is not immediately intuitive to me on what grounds one should value a human life more than that of a cow or pig. The moral weights project tries to put a number to something difficult to quantify. Itâs not perfect, but it is better than nothing. (To be clear, I do value the human life more, but it seems unfair and speciest to do so.)
Jesse I donât think your example is correct because did the GiveWell error bars donât overlap with food kitchen ones. We can be 99+ percent sure malaria nets are more cost effective than soup kitchens. That just isnât the case here. Comparing the certainly of human intervention effectiveness vs. animals is like chalk and cheese
I donât agree with Henry that the huge error bars make RPs welfare ranges useless, probably because I value certainty a bit less than him. But I do think if we value certainly to any degree that can reasonably make us de-value animal welfare point estimates as RP demonstrate themselves in their moral parliament tool.
Alternative response: If someone told me that there was somewhere between a 0.00001 and 0.5 chance that I was to be struck by lightning tomorrow, it would not be reasonable for me to say âwell almost everywhere within that confidence interval I have a >1% chance of being hit by lightning tomorrowâ
Most of these CIs start at zero and they canât go below zero so shouldnât we consider these on a log scale? In which case the scale goes back to negative infinity and âalmost everywhere withinâ is meaningless.
I donât know of any reasonable justification for caring about expected log-welfare rather than expected welfare. For a welfare range estimate, the thing that matters is the expected value.
For a welfare range estimate, the thing that matters is the expected value.
Right. I have been using Rethink Prioritiesâ (RPâs) median welfare ranges, but I care about expected welfare. RP thinks their median welfare ranges are a better proxy for the actual means than the means of the distributions they got, and I tend to agree.
I think itâs very relevant that animal welfare interventions look better than global health interventions almost everywhere within the RP intervals.
I think this point is stronger than inferred from the graph because the 90 % confidence interval of the median is narrower than the range from the 5th to the 95th percentile.
Even if one takes the midpoint of the RP intervals as established fact, there are a lot of other assumptions Vascoâs arguments depend on, like the magnitude and duration of suffering a particular creature experiences with pain scales with thousands of points to cancel out the RP weights, and the cost-effectiveness of brand new charities in a field (campaigning) where marginal cost-effectiveness is relatively difficult to measure.
Unlike for RP we donât have published estimates of distributions or confidence intervals for these, but if we did theyâd also be extremely wide and Iâm not sure that animal welfare interventions would look better across most of the distribution for them.
That argument is weak to me because you could take any intervention we are clueless about and it would look better than global health interventions within most of the interval. If our interval spans zero to close to infinity then global health interventions are going to be a speck near the bottom of that interval.
That argument is weak to me because you could take any intervention we are clueless about and it would look better than global health interventions within most of the interval.
The overall effect of global health and development (GHD) interventions depends on effects of animals due to the meat-eating problem. So I think clueless about the benefits of helping animals implies cluelessness about whether GHD interventions are beneficial or harmful.
Besides not touching on all of these considerations, many of my modelled inputs are highly uncertain too. However, this means extending human lives globally, and in China, India and Nigeria may be, in the nearterm, not only beneficial, but also hugely harmful. Using RPâs [Rethink Prioritiesâ] 5th and 95th percentile welfare range of shrimp of 0 and 1.15, and maintaining all the other inputs, the harms caused to poultry birds and farmed aquatic animals as a fraction of the direct benefits of human life in 2022 would be:
Globally, 5.61 to 372.
In China, 12.3 and 841.
In India, 1.70 and 131.
In Nigeria, 1.12 and 45.3.
If our interval spans zero to close to infinity then global health interventions are going to be a speck near the bottom of that interval.
The cluelessness for GHD interventions would in that case be way more severe. It would go from minus to plus infinity instead of 0 to infinity. Less abstractly, one can be much more confident that humane slaughter interventions are beneficial than that saving human lives is beneficial. Humane slaughter interventions may have negligible benefits if the animals helped turn out to have a negligible welfare range, but it is very hard for them to be harmful in expectation, because they have minor effects on human- and animal-years. In contrast, saving human lives may well increase the animal-years with negative lives, thus potentially being harmful.
That might be true by your lights Vasco, but we are discussing a specific issue here (GiveWell vs. Animal Welfare confidence intervals) and I think its a bit disingenuous to bring adjacent arguments like the meat eating problem into this here.
If youâre clueless about an intervention and you use a fat-tailed prior, then the expected value might be very large but the median value will be very small, and most of the probability mass will be close to 0. For the RP welfare estimates, the median values make animal welfare interventions look highly effective.
Hi Henry. That graph represents the welfare range distributions (minimum, 25th percentile, median, 75th percentile, and maximum), not the confidence intervals of the medians. I think these are what matters, and that they would be much narrower.
The evaluations of human welfare interventions neglect the uncertainty of welfare ranges too. By not considering effects on animals, they are implicitly assuming all non-human welfare ranges are equal to 0. For plausible welfare ranges, some grants from GiveWell (GW), and organisations incubated by Ambitious Impact (AIM) may be harmful. Lots of uncertainty about the benefits of helping animals translates into lots of uncertainty about whether saving human lives, which tends to increase the population of animals nearterm, is beneficial or harmful.
I strongly endorse expectationaltotalhedonisticutilitarianism (increasing happiness, and decreasing suffering), but I think the takeaways are basically the same under desire theories, as beings want to be happy, and not suffer. For my conclusions to change significantly, I believe one has to strongly reject impartiality, or consider Rethink Prioritiesâ median welfare ranges dramatically overestimate animalsâ capacity for welfare.
If these animal welfare analyses keep concluding that all human development has been net negative because of our terrible impact on animals, then, reductio ad absurdum, perhaps these analyses arenât useful.
This reasoning should go in the basket of âhard to say itâs wrong but leads to impractical absurd conclusionsâ along with Ted Kaczynskiâs manifesto and Antinatalism
If these animal welfare analyses keep concluding that all human development has been net negative because of our terrible impact on animals
I said âmay be harmfulâ, but global health and development interventions may also be beneficial. My analysis looks into the nearterm effects, but one should also consider longterm effects, and even the nearterm ones are very uncertain.
Nevertheless, it is unclear to me whether saving human lives in China, India or Nigeria is beneficial or harmful to farmed animals. Even if it is harmful to farmed animals nearterm, it can still be beneficial overall:
I wouldsay at least chickensâ lives can become positive over the next few decades in some animal-friendly countries. Relatedly, I would ideally determine the welfare burden per animal per year by country, although it is unclear to me whether I am over or underestimating it. Furthermore, I guess better worsening conditions now imply a longer time until reaching positive lives, and therefore a longer time until increased consumption of farmed animals being beneficial.
I can see saving human lives being beneficial due to decreasing the number of wild animals with negative lives, although no one really knows whether this is the case or not.
It is unclear to me whether saving lives increases or decreases person-years. It increases these nearterm via increasing population, but may decrease them longterm, as lower child mortality is associated with lower fertility, which can lead to a smaller longterm population. Note human welfare may be decreased in this case.
I assume improved human conditions increase the success of animal welfare interventions, for example, via greater willingness to pay for higher welfare products. In any case, I expect more targeted approaches explicitly optimising for animal welfare to be much more cost-effective.
Recurring issue I find with your posts is that you state expected values with huge uncertainty around them as if theyâre certainties.
E.g. âThe Shrimp Welfare Project (SWP) has been 64.3 k times as cost-effective as GWâs top charitiesâ
These estimates are built on other estimates each having their own error bars, sometimes with some assumptions thrown in. These EVs are houses of cards that shouldnât be taken very seriously at all.
While this is true, I also think itâs worth considering that this is often a criticism of any CEA, period. To the average person, the suggestion that a GiveWell top-recommended charity is more cost-effective than, say, a local food kitchen similarly requires estimates with error bars.
Yes, there are more assumptions when dealing with animals given welfare ranges, but I am reluctant to dismiss the analysis entirely because of that.
It is not immediately intuitive to me on what grounds one should value a human life more than that of a cow or pig. The moral weights project tries to put a number to something difficult to quantify. Itâs not perfect, but it is better than nothing. (To be clear, I do value the human life more, but it seems unfair and speciest to do so.)
Jesse I donât think your example is correct because did the GiveWell error bars donât overlap with food kitchen ones. We can be 99+ percent sure malaria nets are more cost effective than soup kitchens. That just isnât the case here. Comparing the certainly of human intervention effectiveness vs. animals is like chalk and cheese
I donât agree with Henry that the huge error bars make RPs welfare ranges useless, probably because I value certainty a bit less than him. But I do think if we value certainly to any degree that can reasonably make us de-value animal welfare point estimates as RP demonstrate themselves in their moral parliament tool.
Thanks, Jesse.
I think it is fair to value human welfare more. Rethink prioritiesâ median welfare ranges are still lower than 1.
The Rethink Priorities Welfare Ranges have absurdly wide confidence intervals. So wide that I would argue theyâre almost worthless.
I think itâs very relevant that animal welfare interventions look better than global health interventions almost everywhere within the RP intervals.
Alternative response: If someone told me that there was somewhere between a 0.00001 and 0.5 chance that I was to be struck by lightning tomorrow, it would not be reasonable for me to say âwell almost everywhere within that confidence interval I have a >1% chance of being hit by lightning tomorrowâ
Most of these CIs start at zero and they canât go below zero so shouldnât we consider these on a log scale? In which case the scale goes back to negative infinity and âalmost everywhere withinâ is meaningless.
I donât know of any reasonable justification for caring about expected log-welfare rather than expected welfare. For a welfare range estimate, the thing that matters is the expected value.
Agreed, Michael!
Right. I have been using Rethink Prioritiesâ (RPâs) median welfare ranges, but I care about expected welfare. RP thinks their median welfare ranges are a better proxy for the actual means than the means of the distributions they got, and I tend to agree.
I think this point is stronger than inferred from the graph because the 90 % confidence interval of the median is narrower than the range from the 5th to the 95th percentile.
Even if one takes the midpoint of the RP intervals as established fact, there are a lot of other assumptions Vascoâs arguments depend on, like the magnitude and duration of suffering a particular creature experiences with pain scales with thousands of points to cancel out the RP weights, and the cost-effectiveness of brand new charities in a field (campaigning) where marginal cost-effectiveness is relatively difficult to measure.
Unlike for RP we donât have published estimates of distributions or confidence intervals for these, but if we did theyâd also be extremely wide and Iâm not sure that animal welfare interventions would look better across most of the distribution for them.
That argument is weak to me because you could take any intervention we are clueless about and it would look better than global health interventions within most of the interval. If our interval spans zero to close to infinity then global health interventions are going to be a speck near the bottom of that interval.
Hi Nick.
The overall effect of global health and development (GHD) interventions depends on effects of animals due to the meat-eating problem. So I think clueless about the benefits of helping animals implies cluelessness about whether GHD interventions are beneficial or harmful.
The cluelessness for GHD interventions would in that case be way more severe. It would go from minus to plus infinity instead of 0 to infinity. Less abstractly, one can be much more confident that humane slaughter interventions are beneficial than that saving human lives is beneficial. Humane slaughter interventions may have negligible benefits if the animals helped turn out to have a negligible welfare range, but it is very hard for them to be harmful in expectation, because they have minor effects on human- and animal-years. In contrast, saving human lives may well increase the animal-years with negative lives, thus potentially being harmful.
That might be true by your lights Vasco, but we are discussing a specific issue here (GiveWell vs. Animal Welfare confidence intervals) and I think its a bit disingenuous to bring adjacent arguments like the meat eating problem into this here.
If youâre clueless about an intervention and you use a fat-tailed prior, then the expected value might be very large but the median value will be very small, and most of the probability mass will be close to 0. For the RP welfare estimates, the median values make animal welfare interventions look highly effective.
Hi Henry! The reason why the intervals are so wide is because theyâre mixing together several models. Iâve explained more about this modeling choice and result here: https://ââforum.effectivealtruism.org/ââposts/âârLLRo9C4efeJMYWFM/ââwelfare-ranges-per-calorie-consumption?commentId=Wc2xksAF3Ctmi4cXY
Hi Henry. That graph represents the welfare range distributions (minimum, 25th percentile, median, 75th percentile, and maximum), not the confidence intervals of the medians. I think these are what matters, and that they would be much narrower.
Hi Henry,
The evaluations of human welfare interventions neglect the uncertainty of welfare ranges too. By not considering effects on animals, they are implicitly assuming all non-human welfare ranges are equal to 0. For plausible welfare ranges, some grants from GiveWell (GW), and organisations incubated by Ambitious Impact (AIM) may be harmful. Lots of uncertainty about the benefits of helping animals translates into lots of uncertainty about whether saving human lives, which tends to increase the population of animals nearterm, is beneficial or harmful.
I strongly endorse expectational total hedonistic utilitarianism (increasing happiness, and decreasing suffering), but I think the takeaways are basically the same under desire theories, as beings want to be happy, and not suffer. For my conclusions to change significantly, I believe one has to strongly reject impartiality, or consider Rethink Prioritiesâ median welfare ranges dramatically overestimate animalsâ capacity for welfare.
If these animal welfare analyses keep concluding that all human development has been net negative because of our terrible impact on animals, then, reductio ad absurdum, perhaps these analyses arenât useful.
This reasoning should go in the basket of âhard to say itâs wrong but leads to impractical absurd conclusionsâ along with Ted Kaczynskiâs manifesto and Antinatalism
I said âmay be harmfulâ, but global health and development interventions may also be beneficial. My analysis looks into the nearterm effects, but one should also consider longterm effects, and even the nearterm ones are very uncertain.