Hi Siobhan! I’m the Science Director at WAI. I'm stepping in to answer your questions here because Cam is on medical leave.
“What exactly counts as a self-sustaining academic field for wild animal welfare? Is that defined by number of labs? Funding sources? Course offerings? Publication volume? ‘Self-sustaining’ risks becoming an unending horizon.”
Ultimately, what we care about is generating scientific research that can guide action towards improving wild animal welfare and carries the credibility to persuade policymakers to implement it (or that enables wildlife managers to base decisions on welfare and not just population size - which we're working towards through projects like this). However, we think that both of those things require a scientific field to do effectively at scale. Operationally, we define a “self-sustaining academic field for wild animal welfare” primarily in terms of institutions and funding sources. We would be satisfied that there is one when it has most of the institutional trappings of a field — e.g. at least two out of three of a) a regular conference, b) a dedicated journal, and c) a professional society — with significant momentum coming from value-aligned scientists outside of Wild Animal Initiative, AND a majority of funding for the field coming from sources other than WAI (say >$1M per year). That needs to happen while keeping the priorities of the field aligned with benefiting as many animals as possible by as much as possible, which requires more careful progress than if we were satisfied with a field focused on the welfare of endangered species, for example.
We feel on track to achieve the ‘institutions’ part within the next 5 years at our current level of funding, as we’ve seen great signs of progress in just the last year as some of our earliest grantees (from 2022) have stuck with the field and begun to organize their own events and collaborations related to wild animal welfare.
We have made less progress so far on bringing new mainstream funders explicitly into the field. However, we expect this to follow “institutionalization” of the field, as it comes to their attention in a credible way. This year in particular we have seen some evidence of new funding outside of EA. There have also been early conversations starting this year around particular kinds of academic events and structures that we consider signposts of the field becoming institutionalized (some examples described here), so we’re hopeful to have clearer timelines on this in the near future.
Additionally, in the meantime, many of the projects we fund build on existing projects in some way, which represents indirect co-funding that increases cost-effectiveness.
It’s worth noting that the pathway we are on (of institutionalization → funding diversification) wasn’t/isn’t the only way things could proceed towards field-building. Without the support WAI has received from EA donors, we likely would have prioritized funding diversification first to start building the field, but that would have been much less likely to succeed at all and certainly would have taken much longer to build the field than we currently think it will (at current levels of funding) for us to reach a self-sustaining academic field for wild animal welfare.
Just as an anecdotal piece of evidence, I have worked at WAI since it started in 2019, coming from an academic research background, and I am very surprised and grateful for how the field has been able to develop over this timespan.
“What does ‘the long run’ mean in practice?”
You’re right that the honest answer is “as long as it takes,” as with most cause areas. The ultimate goal of improving wild animal welfare is something that we expect will require ongoing management, research and development, like public health for humans. However, academic field building is simply a means to that end, and we do expect it to be ‘achieved’ at some point in the not-so-distant future, as described in my answer to #1.
Overall, we’re basically hoping to take a process that generally occurs over about 50-80 years and shorten it to 15-20 (including the 6 years we’ve already operated) for full success, with the turning point of having mainstream funding being the primary funding source sooner than that.
“How much funding do you estimate is required to reach this self-sustaining point?”
If we’re right that institutionalization will catalyze the entry of more mainstream funders into the field, then I think we could reach that goal within another 5-10 years with a total operating budget of $5M/yr (closer to 5 years) to $3M per year (closer to 10 years; our current budget).
So overall, if you think farmed animal welfare can easily absorb an additional $5M per year and maintain high cost effectiveness, and you prefer immediate gains to long term strategies, WAI and wild animal welfare may not be the ideal giving opportunity for you! But if you are comfortable with funding longer-term theories of change, we think it is as cost effective or more so in expectation than farmed animal welfare — particularly at the margin, as there is some reason to think that the farmed animal welfare movement cannot currently absorb more funding than Coefficient can provide.
Good points. If I understand you correctly, these sorts of benefits from population size (independent of the amount of resources/habitat area available) would be essentially the converse of density-independent mortality factors. I've tried to use the term "density" quite broadly here so that effects like these could be accounted for in a single density-dependent welfare curve.
Hi Flemming, you're right that high survivorship would generally entail a long life expectancy. Sorry, this summary didn't adequately explain how "RWE" is to be calculated. In the RWE calculation, welfare expectancy is normalized around the average annual welfare across all the ages within an individual's maximum lifespan (i.e. the lifespan they might live if all their needs were met and they died of old age), so the average age-specific welfare == 1. This normalized welfare expectancy is then divided by the life expectancy, which always values every year of life as 1. This controls for differences in life expectancy, so in the pre-print linked above, species with life expectancies as different as 1 year and 40 years come out with RWE values pretty near 1 on either side.
RWE is intended to show whether the ages which most individuals live through are especially good or especially bad ones. For example, as kcudding pointed out in an earlier comment, some herbivorous insects seem like they may have higher welfare as juveniles than as adults. This would lead to RWE > 1. For many species, though, the juvenile period involves very high mortality, so most individuals only survive to experience desperate times. They would probably end up with RWE < 1. RWE always tends towards 1 as life expectancy increases towards the maximum lifespan, including in humans (which I can say with confidence since we have actual data on age-specific psychological wellbeing for humans!), which emphasises that it is about identifying a gap between welfare expectancy and life expectancy, not a welfare metric in itself.
I completely agree that the downstream effects on other species should be accounted for. The true ideal would be to manage a population in such a way that maximizes the total welfare expectancy of all sentient life! I hope we'll eventually have the information necessary to do so. In the near term, I'm advocating for populations that are currently managed in such a way as to maximize their net reproductive rate (i.e. 'fecundity expectancy') to instead be managed around total welfare expectancy.
Given the relevant information, I think the welfare expectancy approach makes sense of the example of an herbivorous insect where the juvenile stage offers the highest welfare, but where locking them into that stage would lead to rapid extinction.
(edited to correct numbers:)
For example, assume that the average survival rate and welfare during the juvenile stage is 0.9 and during the adult stage is 0.2; the juvenile stage lasts ~1.8 months while adults survive a further ~0.7 months; and adults produce an average of 5 offspring per month. Assume we could determine a proportion of the larvae to develop into adults while the rest remain juveniles, and that this wouldn't affect any survival/welfare/fecundity rates. The lifetime welfare expectancy of juvenile-locked individuals would be ~8.08 and of "normal" individuals ~1.56. "Normal" individuals would produce an average of 4.86 offspring over their lifetime. A higher proportion of individuals being limited to the juvenile stage increases average welfare expectancy per individual, but curtails the population size. A lower proportion of individuals being limited to the juvenile stage means average welfare expectancy is lower, but the number of individuals experiencing it is higher.
Plugging in the numbers above, I find the total welfare expectancy is maximized when 38% of individuals are limited to the higher-welfare (but non-reproductive) juvenile stage. Of course, the assumption that such an intervention wouldn't affect age-specific survival/welfare rates seems implausible, and this still isn't accounting for potential effects on other species. However, I hope this illustrates how the usefulness of this approach to wild animal welfare might scale with better ecological understanding!
I agree with your observation about scale. It's interesting to think about where the idea of parents having obligations to their children - or of individuals having a special obligation to their community members/fellow citizens - comes from. I think these might come partially from a notion of neglectedness. My child is not more important, morally, than any other, but I can assume most other children already have parents looking out for them, so my child is counterfactually the most neglected cause (and the most tractable cause among children I could care for).
Hi Siobhan! I’m the Science Director at WAI. I'm stepping in to answer your questions here because Cam is on medical leave.
So overall, if you think farmed animal welfare can easily absorb an additional $5M per year and maintain high cost effectiveness, and you prefer immediate gains to long term strategies, WAI and wild animal welfare may not be the ideal giving opportunity for you! But if you are comfortable with funding longer-term theories of change, we think it is as cost effective or more so in expectation than farmed animal welfare — particularly at the margin, as there is some reason to think that the farmed animal welfare movement cannot currently absorb more funding than Coefficient can provide.
Good points. If I understand you correctly, these sorts of benefits from population size (independent of the amount of resources/habitat area available) would be essentially the converse of density-independent mortality factors. I've tried to use the term "density" quite broadly here so that effects like these could be accounted for in a single density-dependent welfare curve.
Hi Flemming, you're right that high survivorship would generally entail a long life expectancy. Sorry, this summary didn't adequately explain how "RWE" is to be calculated. In the RWE calculation, welfare expectancy is normalized around the average annual welfare across all the ages within an individual's maximum lifespan (i.e. the lifespan they might live if all their needs were met and they died of old age), so the average age-specific welfare == 1. This normalized welfare expectancy is then divided by the life expectancy, which always values every year of life as 1. This controls for differences in life expectancy, so in the pre-print linked above, species with life expectancies as different as 1 year and 40 years come out with RWE values pretty near 1 on either side.
RWE is intended to show whether the ages which most individuals live through are especially good or especially bad ones. For example, as kcudding pointed out in an earlier comment, some herbivorous insects seem like they may have higher welfare as juveniles than as adults. This would lead to RWE > 1. For many species, though, the juvenile period involves very high mortality, so most individuals only survive to experience desperate times. They would probably end up with RWE < 1. RWE always tends towards 1 as life expectancy increases towards the maximum lifespan, including in humans (which I can say with confidence since we have actual data on age-specific psychological wellbeing for humans!), which emphasises that it is about identifying a gap between welfare expectancy and life expectancy, not a welfare metric in itself.
I completely agree that the downstream effects on other species should be accounted for. The true ideal would be to manage a population in such a way that maximizes the total welfare expectancy of all sentient life! I hope we'll eventually have the information necessary to do so. In the near term, I'm advocating for populations that are currently managed in such a way as to maximize their net reproductive rate (i.e. 'fecundity expectancy') to instead be managed around total welfare expectancy.
Given the relevant information, I think the welfare expectancy approach makes sense of the example of an herbivorous insect where the juvenile stage offers the highest welfare, but where locking them into that stage would lead to rapid extinction.
(edited to correct numbers:) For example, assume that the average survival rate and welfare during the juvenile stage is 0.9 and during the adult stage is 0.2; the juvenile stage lasts ~1.8 months while adults survive a further ~0.7 months; and adults produce an average of 5 offspring per month. Assume we could determine a proportion of the larvae to develop into adults while the rest remain juveniles, and that this wouldn't affect any survival/welfare/fecundity rates. The lifetime welfare expectancy of juvenile-locked individuals would be ~8.08 and of "normal" individuals ~1.56. "Normal" individuals would produce an average of 4.86 offspring over their lifetime. A higher proportion of individuals being limited to the juvenile stage increases average welfare expectancy per individual, but curtails the population size. A lower proportion of individuals being limited to the juvenile stage means average welfare expectancy is lower, but the number of individuals experiencing it is higher.
Plugging in the numbers above, I find the total welfare expectancy is maximized when 38% of individuals are limited to the higher-welfare (but non-reproductive) juvenile stage. Of course, the assumption that such an intervention wouldn't affect age-specific survival/welfare rates seems implausible, and this still isn't accounting for potential effects on other species. However, I hope this illustrates how the usefulness of this approach to wild animal welfare might scale with better ecological understanding!
I agree with your observation about scale. It's interesting to think about where the idea of parents having obligations to their children - or of individuals having a special obligation to their community members/fellow citizens - comes from. I think these might come partially from a notion of neglectedness. My child is not more important, morally, than any other, but I can assume most other children already have parents looking out for them, so my child is counterfactually the most neglected cause (and the most tractable cause among children I could care for).