With Charity Entrepreneurship team after spending considerable time on creating the best system we could for evaluating animal welfare, we applied this system to 15 different animals/breeds. This included 6 types of wild animal and 7 different types of farm animal environments, as well as 2 human conditions for baseline comparisons. This was far from a complete list, but it gave us enough information to get a sense of the different conditions. Each report was limited to 2-5 hours with pre-set evaluation criteria (as seen in this post), a 1-page summary, and a section of rough notes (generally in the 5-10 page range). Each summary report was read by 8 raters (3 from the internal CE research team, 5 external to the CE team). The average weightings and ranges in the spreadsheet below are generated by averaging the assessments of these raters.
The goal of Charity Entrepreneurship is to compare the different charitable interventions and actions so that new strong charities can be founded. One of the necessary steps in such a process is having a way to compare different animals in different conditions. We have previously written both about our criteria for evaluating animals and about our process for coming to that criteria. This post explains our process and how the results for this system are being applied to different animal conditions.
One of the goals of our system was to be applicable across different animals and different situations. We ended up comparing 9 animals (Humans, Hens, Turkeys, Fish, Cows, Chimpanzees, Birds, Rats, Bugs). These animals are not based on consistent biological taxonomy due to limited information being available on certain types (e.g. there was enough information on rats specifically to do a report on them, but for wild birds we had to look at a variety of birds to get sufficient data). We are not concerned about this limitation, as most of the interventions we are considering would hit a wide range of animals (e.g. a humane insecticide would most likely not be target-specific, so the most relevant data here is an index for bugs as a whole as opposed to an index on a specific species.)
The reports are formatted so that it is easy to quickly grasp the main information connected with the specific rating. Each report is a summary page with the key information and a short description as to why the given rating, and thus, should be polished and readable to all. Each report was time capped at 1-5 hours, so they are limited in both scope and depth. We are keen to get more information on any of these areas (particularly information that is numerically quantified or related to wild animals, as this information was the hardest to find).
Sample report:
After each of the reports were drawn up, each summary report was read and evaluated by 8 raters. We tried to get a diverse set of raters but all with a broadly utilitarian and EA framework. Three raters were from our internal CE research team (the staff who created or contributed to the reports) and five raters were external to the CE team, but involved in the animal rights’ research space (e.g. working or interning for EA animal organizations). The CE research team talked over ratings and disagreements openly, but the external raters did not see or disclose any CE ratings until after they had put in theirs. Ethically, people were best described as classical utilitarians, but with some slight variation (e.g. some more prioritarian, some negative leaning utilitarians). We liked the concept of multiple independent raters as there are many soft judgment calls and increasing the numbers of people doing ratings seems to mitigate specific biases and fallacies. This system has also been used before, and to good effect, by GiveWell.
Ultimately, we ended up with a wide range of ratings going from 81 (strongly net positive) to -57 (strongly net negative). Some of the reports were pretty surprising and ended up changing our intuitions (for example, many wild animals were worse off than in our initial views). Others were not that surprising (for example, the rankings of factory farmed hens).
Our full spreadsheet, with all the ratings as well as links to the 1-page reports, gives specific descriptions as to why certain animals and situations received certain ratings. We feel as though there is lots of room to improve these numbers, particularly with deeper investigation into the lives of wild animal. But we limited our time on these reports due to finding that, historically, within our CEAs, factors like these did not end up carrying the most weight or being the source of highest variability. For example, the cost of an intervention can vary by several orders of magnitude, and more logistical factors were more often the deciding factor when deciding between the most promising looking interventions.
If you want to receive information about our latest reports, subscribe to Charity Entrepreneurship's newsletter. Once a month we will send you a summary of our progres.
This is really neat, I'm a big fan of the comprehensive approach and the documentation style. Will spend more time later looking into the details; I'm not an expert in the field and can't comment on the specific methods, but the high-level work seems very reasonable.
Side note that it seems kind of dismal that wild chimps are apparently rated with higher welfare than average humans in India, though I guess the chimp lives may actually be pretty nice, especially because there aren't many of them. On that note, are there other animal species you think are particularly happy, but didn't include in this report?
Thanks!
In terms of other animals that could be quite net positive, large herbivores and predators at the top of their food chain with relative abundance of food (e.g. elephants, moose, whales and dolphins) would be my guess, but we did not go deep into any of those animals. Some domestic animals (e.g. well treated dogs and cats) also seem plausible to have pretty net positive lives.
Strong upvote for posting original research conducted with a model that (as far as I know) is both novel and coherent, and also for pulling in raters from outside your organization to assist in the work.
I have a few questions -
Can you elaborate on what difference species makes for the wild bug, wild fish, etc reports? I could imagine that a spider has a pretty different welfare on average than an ant, for example, so it seems hard to know what kind of animal these particular scores represent when they cover big categories.
Also, it seems like some of the welfare considerations for factory farmed animals don't account fully for regional laws, etc. (e.g. a country where debeaking is banned probably would have a different score for laying hens in battery cages than in ones where it isn't). Do you think that the average life tends to be close enough to some minimum that these sorts of differences don't end up mattering?
Also, as a small side note - you note in the broiler report that broilers are debeaked - that isn't accurate generally, except for breeding stock, since broilers are killed at a pretty young age now, before pecking negatively impacts the flock.
When we looked at larger groups like fish or bugs we looked for species that were a) more studied and b) more populous. For example, for bugs this tended to be ants, bees, flies, and beetles. Overall though we tried to get a score that we felt would be consistent with “a random unknown bug is killed by an insecticide. What was the welfare score of that bug?”
We only set aside enough time to cover a certain number of animals, and we did not think looking at most regional differences was as important as covering more animals. We will be releasing a table with some specific welfare changes (e.g. animals raised without any physical alterations) which will shed a bit of light on some regional differences. That being said, I expect the broadest level conclusions (e.g. prioritizing fish) to hold across different locations.
Thanks. It indeed looks like that was taken from a report on the breeders.
Thanks!
Thanks for pushing the frontier of interspecies comparisons!
I understand there is not much variation in the total welfare score, but this may not apply to the moral weight (which varies a lot based on the number of neurons, for instance). So species can potentially be a major factor for prioritisation.
Some questions:
I am particularly interested in the answers for terrestrial arthropods (i.e. the "wild bug").
This is really cool! One thing which stuck out to me: you list that there are the same number of bugs as there are factory farmed fish. Is that really correct? I would have thought that there would be many more bugs than fish.
Thanks Ben! Corrected: we certainly agree that there are many more bugs than fish factory farmed fish.
Some pretty unintuitive results for some of these. I would not have assumed that a dairy cow would have a worse estimate for welfare score than a beef cow. The method seems pretty logical so I think it is more accurate than just my intuition. I guess my concern would still be with inter-species comparisons of utility, given their possible varying levels of sentience. How is CE approaching this problem? With the usual neuron amount or is there a better way of doing it? I suppose that would just have to be something you have to concede a large margin of error for when comparing between species.
We pulled the data on odds of feeling pain from Open Phil’s report on consciousness and moral patienthood. The probability of consciousness (as loosely defines by examples in the report) for a given species were estimated based on proxies like last common ancestor with humans, neurobiological features, nociceptive features and other behavioral/cognitive features. In our system, we based weighting of different criteria based on multiple factors including proxying ethical value accuracy (metric and ethical value, encapsulation, directness and gamability) and cross-applicability, including cross-animal applicability. You can read more on that in our previous post.