[EDIT: This partial retraction has been updated and itself partially retracted, see here]
Commenter Tibbar used Mechanical Turk to replicate my survey on how people thought about the moral weights of animals.
After getting 263 responses (to my 50), he reports different results:
Chicken: 25
Chimpanzee: 2
Cow: 3
Elephant: 1
Lobster: 60
Pig: 5
Human: 1
On the one hand, Mechanical Turkers sometimes aren’t a great sample, and some of them seem to have just put the same number for every animal so they could finish quickly and get their money. They also probably haven’t thought about this that much and don’t have much of a moral theory behind what they’re doing. This makes them a different demographic than the people I surveyed, who were a mix of vegetarians and principled non-vegetarians who had thought a lot about animal rights. For example, 80% of my sample answered yes to a question asking if they were “familiar with work by Brian Tomasik, OneStepForAnimals, etc urging people to eat beef rather than chicken”.
On the other hand, this makes it pretty hard for me to claim my results are some kind of universal intuitive understanding of what animals are like. So I am partially retracting them (only partially, because of the consideration above) and adding this to my Mistakes page.
The best thing to do here would be to re-run my survey with a larger sample of a similar population, but unfortunately I’ve lost my chance to do that now that I’ve told you all this, so darn. Maybe I’ll include it on next year’s survey anyway and hope you’ve forgetten by then.
For what it’s worth, I’ve tried to take the annual survey the last couple of years and always given up because starting the survey gives me a strong feeling that your [where “you” is the survey author] values and general worldview are so alien to mine that I just can’t get past it.
I have no suggestions and I don’t even mean to imply that you’re doing anything wrong, but this has left me with an intuition that your survey results are probably even more highly selected than you might expect.
Imagine two surveys on the very popular, accessible topic “are you a good person?”
One was composed by an 18th-century Catholic cardinal, and the other was composed by someone who passed the Qing imperial examinations at the highest level as a 进士 [the title given to those who passed at the highest level]. Both make free use of their respective native extensively-developed moral jargon.
I suspect that Catholics would have a difficult time with the Confucian survey, and Chinese intellectuals would have a hard time with the Catholic survey, even assuming each was translated perfectly.
Would be very curious to hear more — can you give any examples of questions that were particularly alienating? How would you characterize the difference between the survey’s implied worldview and yours? I also felt a sense of value-alienation when looking at the survey, and I wonder if it was if from the same direction as you.
Personally, I tend to be in the camp where I don’t think chickens are likely to have a large enough brain to have any moral value at all, while I’m more uncomfortable about eating pork, and would never ever eat a chimp or an elephant.
Likewise. I would add that I think I distinguish separately or in different ways between suffering and death – I see animal deaths as incommensurable with human deaths, and would accept any amount of pig slaughter to save one human life, but would accept some human suffering to prevent (proportionately very large amounts of) pig suffering.
Suffering and death are both QALY reductions. The former is fractional and allows remediation while the latter is total, but that only matters if you are assessing the status going forward. If you are assessing things looking into the past then the overall impact is the same.
Put differently, would you like to live in a society that did let an unbounded number of humans live in abject misery just to save one individual?
I’m confused by your retraction.
The order is basically the same, and the trends for both lists are exponential on some scale. It’s not hard to squint at the two lists and see the same functional relationship with different parameters (which is easily understandable because of you sampled two different populations). I’d find these results encouraging for your hypothesis, not discouraging.
Also, you haven’t included error bars, so who’s to say that they’re really that different? You have one small sample (50) and another larger one where you claim lots of variance. Looking a single-number summaries and concluding “Well, they’re not identical, so I guess we’re done here!” seems incredibly naive.
The numbers in this survey are so wildly divergent from actual human behavior that it makes me want to ignore them. I guess there’s still some information in the relative values, but mostly this survey just tells me that people are bad at answering this kind of question. I felt that way to a much, much smaller extent with your first survey (though given that survey’s population, I figured it was more about that demographic having atypical norms).
This is really scary. Given a situation where someone was holding a human hostage, threatened the human with certain death unless a ransom was paid, and the ransom was X freshly slaughtered cows, are we really saying that if the kidnapper made any demand where X>3, most people would say, “No thanks; go ahead and slit the guy’s throat–we’ll keep the cows alive”?
That is incomprehensible to me. MAYBE you could argue on utilitarian grounds that you should limit the number of dead cows sacrificed to the hostage to [the point at which killing one more cow would probably cause an additional 2 human deaths due to lack of food], but even then I’d hesitate. THREE COWS for one human life? Horrifying. (Even the number from Scott’s survey, 50, is horrifying.)
Screw the cows. Every KFC restaurant is apparently a genocide-hub.
I concur that these numbers are completely insane and show no consistency with anything anyone does ever. That, or murder is apparently not a big deal.
Yea, the only reasonable way anybody could believe these numbers is if they just think that people are pretty shitty and it’s not morally a big deal to kill random people.
If the average person is willing to sell me out for three cows, they clearly suck.And so I’ll feel ok selling ’em out for 2 or 2.5 cows.
Ok, but seriously, 3 cows or 50 chickens? Even the top 50% of survey of variance across animals Is only up to 20 cows and 200 chickens.
If you care about nothing but money you shouldn’t sell a human out for less than the rough monetary value of a human life. IIRC humans live are worth on the order of a million. A head of cattle is worth ~1000. So about 1000 cows per human. A chicken’s worth about 25 bucks. So 40,000 chickens per human.
So what I’m getting is that the amoral homo economicus thinks humans are worth more than the average human. Maybe this explains something about why capitalism works…
That or people aren’t good at pegging numbers to morality.
I don’t think human lives are worth $1000000 in the same way that a chicken is worth $25.
(In particular, if I own a chicken, I can exchange it for $25 or so. I cannot exchange my life for $1000000 or so, except in the sense that I could promise all my spare income to someone for the rest of my lifespan, which is neither credible nor at all the same kind of exchange as selling a chicken.)
What you say is certainly true, thankfully you can’t buy and sell humans. But I think it gets roughly the right order of magnitude value for a human life for the economic comparison.
With uncertainty maybe it’s 4,000 to 400,000 chickens per human and 100-10,000 cows per human. I wouldn’t quibble too much in that range. Either way, it’s a lot more chickens and cows than the typical answers that people gave in these surveys.
This somehow brought to mind the traditional pastoral cultures where it was common to ‘buy’ brides with livestock.
Perhaps that would be an useful analogy. Would you consider it appropriate if your neighbour traded his daughter away to be married in a far off place for 3 cows? Would you consider it a fair trade if your family traded you away in slavery to save 25 or 50 cows from slaughter?
I agree that cow number is bonkers. But to be fair, the actual moral value our society assigns to cows, chickens, and pigs, as revealed by consumer behavior, is essentially zero.
Another scenario: a guy grabs a pig and says “I will keep this pig in a tiny pen so constricting that the pig literally goes insane; it’s entire existence will be pure suffering until it is slaughtered – unless you give me $2 to pay for a larger pen.” Most people would pay the $2, I would think. And then most people would go out and buy industrially produced pork products, even though much greater well-being for animals could be ensured at minimal cost. Which of those actions reveals the real moral weight they assign to pigs?
It would be interesting to add dogs and cats to this list.
Lots of amazon Turk respondents (especially at that very very low rate) are Indian and cows are seen as sacred in India, so that’s perhaps something to consider.
Looking at old german weregild lists: 1 woman of a free leaseholder = 5 pounds. 1 cow = 4 schilling. 1 pound = 20 schilling. So 1 woman = 25 cows.
And that’s not even accounting for serfs commanding no weregild, just property damage.
Capital punishment also used to be doled out for livestock theft. So moral systems where livestock is close in value to some humans are not inconceivable, they used to exist after all.
The list looks qualitatively the same imho. The only surprising difference is that these turks value elephants more than chimps? That seems so strange. Where are they from, do you happen to know?
Elephants might be a bit more charismatic than chimps? My availability heuristic generally has entirely positive thoughts on elephants whereas with chimps there’s a fair bit of “those things are dangerous” and “like humans in a lot of the bad ways.”
Is that not a factor of their rarity? I’m no expert but elephants and whales are well documented to be in danger.
While I believe the same is true of some gorillas, chimps are not facing extinction, afaik/afa most of us would know?
Put it on next year’s survey and just disguise the question, something like: A trolley loaded with chickens is heading down a track that will cause the trolley to fall off a cliff killing all the chickens. You can throw a switch that will change the track and make the trolley run over a single human but save the chickens. How many chickens must be on the trolley before you will throw the switch?
And I guess you’d also have to have the same question with people on both tracks, as a baseline.
But honestly, I think of chickens and other animals as having non zero moral value, and try to minimize the amount of meat I eat as a result, but I would rather a trillion chickens die than 1 human die (assuming that wouldn’t make chickens extinct). I’m not sure if this is human centric bias or something, but it makes me doubt that normal utilitarianism even correctly captures my moral intuitions with regard to animals. I am not sure there is any number of chickens for which their suffering would equal that of humans, nor any number of ants for which their suffering would equal that of chickens.
Surely, in this modern connected world, there are more alternatives for this than Mechanical Turk and a survey on your blog?
Someone mentioned r/samplesize.
> Chimpanzee: 2
> Elephant: 1
I could explain those by people wanting to protect rare species. I certainly care more about elephants than I would if there were a lot more of them.
> Cow: 3
> Pig: 5
I can’t explain this though.
A cow is worth more than a pig because it’s bigger and more valuable as a commodity. That’s one explanation.
Pigs are jerks and dirty. Cows are nice and cuddly.
Look, these people think that if it’s a choice between your life and that of 26 chickens, you should die. They are morons.
That, or they count “generic human” and “specific human” differently. “Generic human” could be a serial killer so far as we know, after all. Probably not, but median generic real humans aren’t known for their firm grasp of probability.
Or they might reject the framing and be making up their answers. If less than half of people use utilitarian morality, it’s entirely plausible for the median to be a useless indicator. And I’m pretty sure that utilitarians are a minority.
I concur with the legion of other commenters pointing out that these results are insane and completely inconsistent with how people appear to actually live. People very much do NOT act like 25 chickens equal a human life.
The original results are also insane and completely inconsistent with how people actually live. These are just slightly lower (= more insane) numbers.
If 4000 lobsters were actually comparable to one human life, fisheries would have been burned down long ago. By that metric, Maine alone would be hosting an industry that murdered more than 600 people a year for no other purpose than to eat them.
Here’s one of the (more extreme) responses from the last survey:
Chicken: 50
Chimp: 0.3
Cow: 0.2
Elephant: 0.05
Human: 1
Lobster: 40
Pig: 5
Female, vegetarian, from India. There’s no reason to think this is trolling – the relative ratios are pretty well calibrated, and the magnitudes align with lots of other Indian respondents. (Cows, elephants, and monkeys are sacred in Hindu.) There are many modern incidents of violence and murder in India over the treatment of sacred animals: “Between May 2015 and December 2018, 44 people suspected of killing or transporting cows for slaughter, or even just eating beef, were killed in vigilante attacks, according to Human Rights Watch. That number included 36 Muslims.” (https://www.cnn.com/2019/02/20/asia/india-cow-vigilante-hrw-report-intl/index.html).
Ok, so if some people actually believe this, then there’s no reason why a lot of them can’t. A lot of people seem to have sincerely different intuitions about the moral order.
If your goal is to establish how humans intuitively value animals, why are you interested in a larger but similar sample, not a sample representative of the general population?
These numbers seem off by orders of magnitude to me. If you eat meat, you probably cause the death of many, many animals over the course of your life. For example, while I don’t eat lobster terribly frequently, I’d guess I’ve had around 60 of them over the course of my nearly 40 years on the planet. I’ve also probably eaten at least a few pigs and cows over that period too. I probably eat 25 chickens or more every year.
I’m just spit-balling, but here are my numbers:
Chicken: 200,000
Chimpanzee: 500
Cow: 20,000
Elephant: 500
Lobster: 250,000
Pig: 30,000
Human: 1
Full disclosure – my numbers for chimps and elephants are low because they’r endangered and I think there’s value to keeping them around. My numbers for livestock are based on my estimate of what they cost and how much I think it’s worth paying to save a generic human life (as in not somebody I know, but somebody picked at random). I have a fairly anthropocentric view of morality.
>If you eat meat, you probably cause the death of many, many animals over the course of your life.
Why? This is the root idea behind ethical vegetarianism and I never got it. That animal was already dead, it was not killed specifically upon my request. Yes I do contribute to the aggregate demand but that is a far more vague thing. It is not a direct causality that would involve moral responsibility. I don’t understand this idea how indirect causality means moral responsibility.
This seems to me to be evidence that people do not use purely arithmetical linear utilitarianism in their moral intuitions.
Rather, there’s a certain amount of dictionary-sorting going on: chickens may have moral worth, but a kind that ranks below human moral worth. Mortal peril to chickens may merit consideration and inconvenience to humans, but it does not merit mortal peril to humans.
To complete the dictionary analogy, moral acts go from A-Z for both humans and chickens, but humans get the first letter of the word, chickens the second (or third, or fourth, behind cows but ahead of criminals, etc.).
(Just-so evo-psych-ish hypothesis ahead)
I expect this system is transmitted culturally/behaviorally/etc., since different cultures clearly assign different moral weights – and likely also different moral ranks – to various entities. The cross-cultural
trolleyself-driving car problem survey comes to mind. This makes the “which is more moral” check much quicker in most cases: you don’t actually have to sum anything up, just compare “who did the worse-ranked thing”, or if they’re the same rank, “who started it”.This breaks down where people assign things different moral ranks, as you might expect from different upbringings or cultures; it also takes more time to figure out when there are extenuating/aggravating circumstances (“stealing to avoid starving” being the canonical example, so let’s add “stealing from the poor to avoid starving” to illustrate the ambiguity better) – and this is also what we would expect.
(mentioned this in a reply on the initial post, thought it’d make sense as a top-level comment here now that I’ve thought about it a bit more)
Having orchestrated some MTurk experiments, the offered ~$1.20/hr seem much too low. Bad data can be worse than no data, and at least when I tried that rate of payment, there was no good reason to assume that most participants were reasonable, able to read English, or even human. I mean, sure, it’s a different demographic, but maybe more along the lines of “people who barely read the instructions” than “people who didn’t spend years deliberating the moral dilemmas of animal rights”. I would want to know what a person who spent at least 5 seconds per animal thinks, and I’m not sure if the current survey should be taken as strong evidence to answer that.
[Also, Tibbar – sorry for criticizing the results. It is unfair of me, especially after the fact and with no backing with data. I assume you used Qualifications?]
I’m redoing the survey, attempting to fix various issues. I clarified the instructions, requiring workers to explicitly agree with the utilitarian premise to continue. I added questions on age, country, gender, and vegetarianism.
I’m offering $0.60/survey this time (max 200 results). The survey now takes an average of 167 seconds to complete, so this works out to $12.93/hr. I’m now requiring workers to be “Masters,” which is a quality control status implemented by Amazon. I have no way of knowing how useful the status is, but I’m paying extra for it.
I also added more instructions on what constitutes a valid response, and I’m being more aggressive in removing responses that are illegal (e.g. those that answer 0 for some animal).
I have about 50 results so far, and I’m encouraged that they look significantly cleaner. The vast majority of these workers are from the USA. Notably, they’re turning in much higher medians than the Indian workers, though there’s not enough Indian responses yet to make this precise (will update with demographic splits when the data allows it). The overall medians are still mostly lower than Scott’s currently, and I expect that the differences are due to different demographics.
Lobster 500
Chicken 85
Pig 47
Cow 10
Chimp 6
Elephant 2
Human 1
While I’m sure Scott’s original group was ‘better’ than this one, I doubt it represents a universal median of thoughtful people, and I think the problem of determining whose opinion ought to count here is so fuzzy as to possibly invalidate the original premise anyway.
Still crazy.
No one (but Hindus) exchange 10 cows for 1 human life. I’m onboard with human life (except for me and mine) being 0.05 elephant life as the Hindu lady said, though… 🙂
Honest question: do Mechanical Turkers skew Hindu? It seems plausible that they might.
If these mechanical turks are disproportionately drawn form the rural poor or more likely rural poor in cultural memory than the high value on animals could be including their material value at least implicitly. A horse is valuable and should be protected is a statement one could hold to even if one thinks it has only material and not moral value. Rankings of different animals seem to roughly track market prices.
“Mechanical Turkers… some of them seem to have just put the same number for every animal” Surely this is because they read Animal Farm.