Random Indians, not rich Indians. I would appreciate it if you could represent my post fairly.

"Random Indians" is a group which includes poor Indians (i.e. recipients of anti-poverty measures, which have non-random targeting) and rich Indians (typically not recipients of GiveWell or AIM charitable interventions). The assumption you make by using a mean consumption figure is that poor Indians and rich Indians alike consume ~11 shrimp per year. That's what the text you quoted said, and a perfectly fair representation of your post

I actually think it's an unfair representation of my post to accuse me of misrepresenting you simply because I spelled out the logical implications of your choice of figure, especially when I have also presented multiple reasons why I believe zero would be more representative of the amount they were likely to consume, and even more representative of the marginal impact of a typical GiveWell/AIM recipient surviving on Indian aquaculture production.

I explicitly said in the post "The harms would be smaller for a random person helped by such GiveWell’s grants or Ambitious Impact’s organisations", and then argued why it is unclear this changes my conclusions.

My argument is that the median survivor due to GiveWell/AIM aid causes zero harm via aquaculture, and even the small minority of survivors who do consume shrimp are unlikely to have any impact upon numbers of shrimp culled in factory farms. I'm aware your post above argues that meat consumption may be linearly related to welfare via the common factor that is GDP, but I don't think the relatively small diminution in self-reported welfare from lower incomes you've considered here is anywhere near enough to doubt that the survival of Indians without access to aquaculture products might be net positive in the welfarist framework you presented!

I feel we're going in circles here, so I'll wish you a happy Christmas and am unlikely to continue the discussion.

Even no consumption of shrimp would lead to saving lives in India being harmful neaterm if I kept all my other parameters constant, and there would always be significant uncertainty even if my point estimate suggested the benefits to humans are larger than the harms to animals nearterm.

Yes, but I don't see any reason to assume that the uncertainty skews in favour of humans dying rather than humans surviving.  Particularly not when the assumptions you used to reach this conclusion were that that poor Indians receiving nutrition supplementation have access to the same ~11 farmed shrimp per year in their diet as rich Indians, and that the positives of 1 Indian human living for 1 year are no more than the negatives of four shrimp being farmed. 

I think in an area of high uncertainty we should default to the idea that humans should survive (and maybe change their dietary preferences) and not to the idea that they should die

The consumption of poultry birds and farmed aquatic animals per person helped in India in 2022 would have to be less than 19.3 % (= 1/5.17) as large as that of a random person for extending human lives to increase welfare nearterm

Sure. But since food choices are skewed heavily by budgets and aquaculture is a premium export market, and the supply chains to send 11 farmed shrimp per year to every man, woman and child in interior villages don't exist, I don't think the evidence points to the median GiveWell beneficiary consuming any of the ~13 farmed aquatic animals per year you've attributed to them. The idea that GiveWell donations have a non-zero effect on the size of the aquaculture industry at the margin is even more dubious, given that the economics of farming in a region which exports nearly all of its aquaculture products are highly unlikely to factor in a few thousand GiveWell non-beneficiaries dying to their demand calculations and reduce production accordingly.

Others have suggested that proposing that people shouldn't be allowed to survive on the basis of things they might choose to eat in future. I think it would be worse to condemn them for things they are statistically unlikely to even get the opportunity to eat. 

Either way, it's a view you're perfectly entitled to and have clearly done some research into, but I don't think it's a glaring omission that an organization that considers it axiomatic that human lives are worth saving hasn't invested time in doing their own "so actually, under what set of assumptions can we conclude humans shouldn't be saved" calculations.

I do not ask GiveWell for "massive amounts of time" in the context of effects on animals.

I revised this during the time in which you were replying to say "use researcher and writer time" instead, precisely because I didn't want to give the appearance of misrepresenting your post. Though it would certainly take a massive amount of time for GiveWell to address every detailed contrarian argument that actually humans surviving is bad or that welfare ought to be weighted by country or nationality or potential economic output or religion or proximity to the donor or any other mechanism dreamed up by people who have totally different axiomatic beliefs. It seems rather pointless for an organization focused on quantifying how effective organizations are at preserving human life to spend time explaining the precise nature of their disagreements with people holding the contrary belief that it would be better for humans not to live. Clearly there is no agreement to be found there.^[1]

I don't think animal welfare charities should feel obliged to quantify possible negative impacts on human welfare of their activity in the interests of supposed impartiality either. 

I  would say animals matter a lot on non-hedonic views

I agree it is possible to conclude that animal welfare should be prioritised over human welfare using other moral frameworks. I actually find deontological arguments for prioritising animal welfare more convincing than ones filled with arbitrary utility estimates. But the specific framing of the argument you want GiveWell to address is based on total welfarist hedonism with valence symmetry and multiple ancillary assumptions about the relative moral weight of animal lives they clearly don't agree with. Adding the expectation they engage with even more forms of argument that human life is bad would make the proposed standard of reasoning transparency more unreasonable to impose upon them, not less.  

I don't think GiveWell should apportion any more time to the question "what if it's better for humans to die" than the Shrimp Welfare Project should to "what if shrimps don't have welfare and all we're doing is making human lives a little worse". Both are clearly and transparently incommensurate with their beliefs, and people that think that humans don't deserve to live given their dietary habits or that shrimps don't have meaningful welfare ranges can always find and advocate different causes.

I also think one should be very wary of supporting interventions which can easily increase suffering a lot nearterm in the hope that integrating moral uncertainty makes it worthwhile.

I think one should be wary of diverting funding which fairly unambiguously decreases well-established sources of nearterm suffering on the basis of a set of guesses about welfare impacts on aquatic animals with pretty dissimilar biology.^[2] 

I relied on the shrimp supply (production plus net imports) per capita in 2022

Nothing indicates that this supply statistic subtracts out exports. You linked to some total shrimp production stats, and an ambiguous source you caveated which suggests Indians consume half a kilo of shrimp per capita which would be the majority of their domestic product (despite the fact India is the world's largest shrimp exporter and exports nearly all their produce). Elsewhere, I've seen it suggested India's per capita consumption of shrimp is of the order of 100 grams (1.6% of overall seafood consumption). That's less than a restaurant serving per person: it's clearly not being transported to remote villages to be dished out to kids whose diet is so poor vitamin supplementation meaningfully increases their survival chances!

I suspect most recipients of GiveWell funding in India have never even seen a shrimp.

acknowledged the people helped by GiveWell and AIM would cause less harm to animals than random people, but I do think this resolves the meat-eater problem.

If the majority of people in India helped by GiveWell and AIM eat no aquaculture products which is likely true, then by the estimates you posted above, it resolves the meat eater problem as presented in this post for those people (specifically the net benefit in human DALYs is greater than the net harms caused to chicken welfare from egg and occasional meat consumption).^[3]  I agree that you acknowledged that poor people consumed less, but only in the context of dismissing it as a relevant factor^[4]

And yes, your estimates don't factor in future economic growth which could reasonably be expected to increase meat and shrimp consumption by a lot. But you also assume the Shrimp Welfare Project's India-centred campaigns have no impact on the aquaculture practises they're campaigning about despite heartily recommending them as effective!^[5]

I did not follow. I estimate saving lives in India would be harmful nearterm even for no consumption at all of shrimp

At the RP midpoint used in the table in your post, 3.05 of the 4.55 animal harms per year you estimate surviving Indians responsible will be responsible for are attributed to shrimp welfare ergo most^[6] of the quantifiable harm is based on [apparently incorrect] assumptions about shrimp consumption. I agree that this still estimates saving lives as net harmful, though it becomes net positive for the average Indian to live (yay!) under your estimates as soon as the rest of aquaculture is omitted from the equation. Which is not insignificant when there's no scientific consensus about whether sea life experiences welfare at all, never mind how to quantify the impact of aquaculture.

1. ^{^}

   if they were going to enormously expand the scope of their impact calculations to engage with people with fundamentally different starting axioms, it would probably be more productive to do additional calculations to engage with supporters of QALYs or "values of a statistical life"with lower weighting for the poor...

2. ^{^}

   bearing in mind that there isn't even scientific consensus on whether they experience welfare at all, never mind the welfare impact of aquaculture

3. ^{^}

   to be fair there are other farmed animals like goats (which are considerably more likely to be consumed by poor non-vegetarians than farmed shrimp) left out of the calculations whose consumption animal welfare enthusiasts can reasonably attach negative weight to, although I don't think those animals are often factory farmedin India.  

4. ^{^}

   I found the argument that as meat consumption was broadly correlated with GDP per capita and life satisfaction is also broadly correlated with GDP per capita unconvincing given the relatively small impact of GDP on satisfaction. If the beneficiaries are typically consuming negligible amounts of animal produce you can't just write off the net benefit by assuming their lives can't be worth living!

5. ^{^}

   there seem to be quite a few reasons to believe that shrimp stunning practices actually will change over time in India (it's not obviously costly, the SWP isn't the only organization pushing it and seems to be getting at least some positive responses, and India is characterised by influential religions unusually receptive to arguments about invertebrate welfare)

6. ^{^}

   (FWIW I also edited to removed the word "majority" whilst you were drafting your response to try to avoid confusion, though I don't think it's incorrect in the sense I've used it)

GiveWell's moral weights do not depend on the country, which is highly controversial, in the sense the vast majority of people value humans in their country more. However, I would disagree with GiveWell changing their approach to align more closely with societal views, because I do not think welfare intrinsically depends on the country you are born (although the country one is born influences welfare).

You don't seem to expect them to devote researcher and writer time towards "reasoning transparency" justifying treating all human lives of approximately equal value though? So I'm not sure why you would expect them to devote more time and effort to coming up with BOTECs to prove that this value is non-negative, an axiomatic GiveWell assumption which is no more hidden than your own longer list of axiomatic assumptions (hedonic welfare characterised by valence symmetry and the approximate accuracy of a long list of assumptions about how to quantify it is the one true way to allocate resources). If you think the meat-eater problem is a compelling reason to avoid averting human DALYs, you're not going to GiveWell's list of human DALY averting charities for inspiration any more than someone who believes that true charity begins with looking after people closest to you or involves religious salvation.

FWIW even if we accept your axioms there's substantial reason to doubt your own conclusion (as you acknowledge with your "low confidence" statement) that saving human lives is bad. Most obviously, you appear to have imputed that the average Indian reliant on donations for lifesaving (typically in poverty) will grow up to consume shrimp (a relatively high-value foodstuff) roughly in proportion to the number of shrimp farms in India (which exports up to 95% of its shrimp production). That estimate that's probably off by at least one order of magnitude accounts for a large fraction of your estimated negative impact of saving an Indian child's life

I don't see any reason to believe that the same amount of gambling would take place in aggregate. Most entertainment businesses grow the market and this one is promoting new motivations for potentially different people to participate in an activity which is often addictive. And if you're running a bricks and mortar casino you're facing the same high operating costs as the competition: I don't see any reason to believe you'd reach profitability without putting in similar amounts of effort to entice new players, encourage people to return on days they weren't planning to gamble and encouraging people to shovel more money into machines after they've already lost more than they planned. 

You could have a great night in which you win hundreds or thousands of dollars, but even if you lose, they know that your losses are helping to dramatically better the world. 

A cynic reads this as "you could have a great night in which you deprive a few hundred people of malaria nets, but at least in the long run they and also random unrelated and typically obnoxious corporations might stand to benefit from the gambling addiction this has instilled in you....". Possibly the first part of the proposition is slightly less icky if the house is simply taking a rake from a competitors in a game of skill, but still.

Maybe I just know too many people broken by gambling.

Thanks for the very interesting post.

I don't work in commercial aviation any more, but can offer a few pointers

• Eurocontrol are exactly the people you want taking this seriously - they regulate European airspace. So whilst I think it probably is neglected relative to other climate proposals in terms of funding vs estimated impact, it may not be neglected by the right people.
• For related reasons, I think it's way more tractable than most interventions: changing altitude under certain conditions is a lot easier than dissuading people from flying or consuming. And there is an established track record of regulators enforcing environmental rules and costs like noise restrictions and NOx emissions charges (along with sticks governments haven't beat them with yet like carbon taxing jet fuel)
• On the other hand it seems like it's actually true the current state of scientific consensus hasn't resolved the important question of when and where to divert yet (see the variability factors in your infographic) and the diversion usually does result in increased fuel burn (and some contrails are even cooling!) And flight directions are a complex multidimesional problem
• Airspace controllers will need to be involved because airlines are unlikely to do anything voluntarily that impacts their profit margins (which are on average small anyway) regardless of how settled the science. In general, being "greener" through lower fuel consumption actually saves them money; this is an obvious exception.
  • An indirect "stick" approach like levying fines or additional charges on airlines causing contrails whilst passing through particular airspace sounds neat, but whilst theoretically contrails observed from the ground or orbit can be matched to ADS-B readings of aircraft that recently passed through that space, systematically validating that in a legally-valid way in congested airspace seems tricky...
• I can't see it being practical to achieve via consumer pressure and wider public awareness campaigns run the risk of getting mixed up with "chemtrails" conspiracy theories
  • If you want a possible exception to airline lack of sympathy, a UK startup airline Zeroavia is owned by eco-activist billionaire Dale Vince. Their hydrogen powered fleet claims they already intend to capture water emissions to release at lower altitude ^[1] for the stated purpose of avoiding contrails. Zeroavia are a very atypical airline, currently have zero flights and I'm not sure how much aviation industry executives actually respect Dale, but if you wanted to outreach to an airline that actually might be sympathetic and see PR benefits of shouting about contrails, they'd be a starting point

So I think there's definitely something to be worked on here, but its going to take industry experts more than grassroots campaigning. I think there are probably some really interesting algorithm development projects there for people with the right skillsets too... 

(For anyone interested in space, an analogous situation is the aluminium oxide deposited in the mesosphere by deorbiting spacecraft. This used to be negligible. It isn't now that constellations of 10s of 1000s of satellites with short design lives in LEO are a thing. The climate impact is uncertain and not necessarily large but probably negative; the impact on ozone depletion could be much more concerning. Changing mindsets on that one will be harder)

1. ^{^}

   which sounds seriously expensive to me....

Hi FWI. I have actually worked on both Earth Observation projects and projects looking at other forms of remote sensing for assessing water quality in aquaculture (but don't have the technical skillset to participate in your challenge). A few (hopefully helpful) points:

• any sort of useful model is going to require calibrating against background data on the variables you select (particularly as you appear to be working with small, shallow freshwater pools which have different appearance in visual spectrum imagery that probably represents differences in factors other than those you're quantifying e.g. mineral content and depth)
  • since obviously any past water quality shared is going to be highly correlated with the current water quality in a given pool, it would make sense to evaluate models on their ability to pick up change from previous levels in future samples using future images rather than simply on predicting which pools have the most ammonia....
  • the more observations you can share, the better the chance the model actually works
• the variables you've highlighted are theoretically detectable using EO, but they're relatively weak indicators perturbed by other stronger indicators and things like weather (at least it's normally sunny there!). Depending on how much water you collect at how many points in the farm, presumably there's some natural variation in the samples you collect too.
• free satellite imagery such as Copernicus typically has 1 pixel representing 10-20m on the ground so some of your lakes might be only about 2-4 pixels accross. Large pixel sizes don't necessarily stop major changes in water quality being picked up in multispectral imagery (and aren't necessarily an issue if you're measuring the sea surrounding a coastal fish farm, but it's going to significantly affect the fidelity of your results. Unfortunately, I suspect commercial imagery (spatial resolutions more of the order of 0.5m pixels) is outside your budget
• if you're currently only occasionally collecting data, satellite revisit rate should be fine 

I wouldn't be hugely optimistic about success in the short term as I suspect the scope of what you're looking at is a lot more subtle than "spot the effects of leachate on the massive lake" and the data you have so far may not be enough

The other problem with the "indirect enough" argument is that the donations are even more indirect

Sure, the meat people eat is usually killed long before it's ordered and eating a few dozen chickens per year doesn't individually shift an industry. But likewise, a $1000 donation doesn't meaningfully affect an advocacy charity's ability to win a court case. 

Both only work in aggregate, and on a causal basis the link between meat demand and factory farming is much more robustly-established than advocacy charity income and relative absence of factory farms^[1]

And standards for crediting impact need to be stricter here because multiple counting is a much more meaningful problem than when an altruistic donor is deciding where to donate

This is a good point too.  If you're using donations to prioritise in a counterfactual scenario, what part of the outcome is actually "your impact" is irrelevant. If you're using them to buy indulgences, that's less obviously the case.

1. ^{^}

   on a money basis it's less certain, but I still don't think vegan diets are dramatically more expensive than meat ones, and the DALY impact of eating half a chicken doesn't seem to be very different from favourable estimates of DALY impact of a dollar donation to Legal Impact for Chickens...