This piece is part of a short series on how the Wild Animal Initiative research team thinks about uncertainty. These posts represent the opinions of individual staff members, and not necessarily the organizational position of Wild Animal Initiative on uncertainty. This summer, we are raising $50,000 to further our research—you can learn more here and support our work here. This post is crossposted from the WAI blog.
Introduction
Many critiques of improving wild animal welfare focus on the nontarget uncertainty of our actions. We will describe what we mean by “nontarget” below, but in general the term refers to unintended consequences of an intervention. These critics argue that because we are unable to predict the nontarget effects of interventions, we can’t reasonably say that interventions will actually reduce wild animal suffering. For example, if we introduce a vaccine to a disease that we know causes raccoons to suffer, we might change the population size or change the dynamics of the ecosystem in a manner that has consequences for other animals’ wellbeing in unpredictable ways.
Critiques based on nontarget uncertainty should not be viewed as unique to wild animal welfare. If there is a strong case for dismissing a cause area on the basis of nontarget uncertainty, then we’d have good reason to dismiss many large-scale charitable projects, including addressing global health and factory farming. Although this critique fails to apply exclusively to wild animal welfare, we should still strive to improve our ability to predict nontarget consequences of our interventions in all cause areas, and work to resolve some types of uncertainty that are particularly relevant to wild animal welfare.
The argument for caution in the face of nontarget uncertainty
The argument against taking action due to nontarget uncertainty is usually framed as follows:
1. We are uncertain, and in some cases clueless about the nontarget effects of major wild animal welfare interventions on ecosystems.
2. Nontarget effects might have negative ecological or welfare consequences.
P. We should not pursue major interventions if we want to avoid negative ecological or welfare consequences.
This argument generally seems to assume that wild animal suffering matters morally when posed as an objection to wild animal welfare projects. It takes the form of “I agree we should do something about that but…”. The argument might still apply if one does not think wild animal welfare matters morally, but would instead focus on the nontarget effects for humanity of changing an ecosystem.
Target and nontarget effects
Before discussing this argument, we need standardized terminology. Unintended consequences, in various fields, have been called downstream, knock-on, or flow through effects. However, a more general way of conceptualizing the effects of interventions when thinking about uncertainty is to divide them into “target” and “nontarget” effects.
Target effects are both primary and intended. For example, an animal rights activist showing a documentary might have the immediate target effects of causing someone to reduce their meat intake and feel more compassion for animals, and have less immediate target effects like a reduction in the number of animals born on factory farms. The animal rights activist intended for all these things to happen, and they were the direct targets of the intervention.
Nontarget effects are the unintended consequences of interventions. The same documentary might cause a well-intended but less thoughtful watcher to eat more chicken instead of beef, causing more animals to be born into factory farms, due to chickens’ smaller size. Another nontarget effect might be that fewer cows being fed means that some soy farms fail, and are converted to lettuce farms, resulting in a more rabbits to be born due to an abundance of new food (note that this is just a theoretical example). These nontarget effects might happen simultaneously with a target effect—despite causing more animals to suffer overall, the watcher might have decreased their intake of animals calorically, and might feel more compassion for farmed animals.
In the case of interventions to reduce wild animal suffering, the target effects tend to impact the welfare of animals. The nontarget effects might also be related to welfare, or could be related to ecological preservation. Depending on our moral values, we might weigh these nontarget effects differently. For example, a project to make insecticides less painful would, in a perfect implementation, not change the number of target insects killed by the insecticide. However, if the insecticide leaches into the soil, there might be nontarget effects worth considering—perhaps the original insecticide is safe for annelids in the soil, and the less painful one is toxic. Then, we might expect annelid populations to decrease, leading to less food for local birds and a variety of uncertain outcomes for local animals, both from an ecosystem stability and a welfare perspective.
Nontarget uncertainty: not just for wild animal welfare
Of course, wild animal welfare interventions aren’t the only projects to make the world better that involve massive uncertainty. Eradicating the raising and slaughtering of animals for food would shift the use of 36% of crops (by calories). This shift would be a massive change in land use globally, and given the inherent inefficiency of animal agriculture, might cause much of this farmland to rewild. The change in land use could have huge ecological consequences. Populations of local animals would change dramatically. Likewise, addressing global poverty could introduce large-scale industrialization or extraction industries into new regions, having similarly dramatic consequences for local ecosystems. If we care about wild animal welfare, or accept that wild animal suffering is an issue, we should be equally concerned about the nontarget effects of all global development.
Additionally, other charitable projects face their own major nontarget uncertainty independent of wild animal welfare. Global development risks spillover effects. Any effort to impact the far-future might involve a high degree of cluelessness. Efforts to expand cage-free egg supply might lead to individuals being more comfortable with eating eggs, causing more harm in the long run to animals.
Of course, we should still work to reduce the suffering caused by factory farming and should try to reduce global poverty. And for wild animal welfare interventions, it is true that we have comparatively less research on even target effects than we do for farmed animal welfare—farmed animals live in highly regulated environments that humans observe on a daily basis, while we know next to nothing about the lives of most wild animals and the impact of changing environments on their welfare. However, for many interventions, we can make some claims regarding the target outcomes, and we can use insights from conservation and other fields to assess the potential short-term nontarget effects.
Appeals to nature and uncertainty
Many of the problems faced by wild animals can be distinguished from some of the other forms of suffering that have received attention from the effective altruist community—humanity is probably not the proximate cause of most wild animal suffering. In theory, the causes of most wild animal suffering are entirely natural.Yet arguing that we should treat wild animal welfare issues differently from other morally urgent issues because they are natural is an instance of the “appeal to nature” fallacy, if we are assuming the conditions wild animals experience are good because they are natural. But it is not clear that naturalness is a morally relevant feature in any sense.
Similarly, we must also be careful not to resort to appeals to nature when critiquing interventions on the basis of nontarget uncertainty. When we say “an intervention has uncertain nontarget effects and that is bad” it seems that we may sometimes be masking the fallacious belief that “an intervention might change an ecosystem fundamentally, and that is bad in and of itself.”
Of course, we expect our interventions to change ecosystems. That’s the point. To make an intervention resilient, and last through many generations, we should expect it to change fundamental features of that ecosystem. If we are concerned about ecosystems changing for aesthetic purposes (i.e. we find nature to be very beautiful), that’s unrelated to ethical questions. And if we are concerned about ecosystems changing for reasons related to human survival (i.e. climate change will be harmful), that is an important moral consideration, but not the only one. In all of these situations, wild animal welfare still matters.
Our expectation cannot be that ecosystems will not change. Climate change has already eliminated many species that were alive only a few decades ago. Animal species are going extinct on a daily basis. Human development has completely changed many wild spaces, and human activity has enormous effects on ecosystems globally. The hope that these systems will not change is not only morally irrelevant, it is extremely unlikely to be realized regardless of what actions we take on behalf of wild animals. The future of conservation and wild animal welfare requires active intervention and stewardship of wild places.
Finally, it is worth noting that while in the case of addressing wild animal welfare the appeal to nature might seem especially relevant, we can see many existential risks as being naturogenic, such as biosecurity threats, or major asteroid impacts. Similarly, many diseases that global health initiatives work to prevent are themselves naturogenic. Given that appeals to nature aren’t applied as objections to working on these causes, we should also not take them seriously for wild animal welfare work.
Further research on outcomes should still happen
Of course, we need to continue researching the nontarget effects of our actions to improve wild animal welfare. Fortunately, fields like restoration ecology have already made significant advances on collecting evidence related to and predicting the effects of human intervention in nature. We can decrease our uncertainty about the target and nontarget effects of our interventions by continuing to fund research on predicting the outcomes of human interventions in ecosystems, especially as it applies to wild animals.
We will never have absolute certainty about the nontarget effects of our interventions, but we can work to mitigate negative effects by developing better tools to monitor and react to unexpected effects as they arise. Although the cost-effectiveness of an intervention would be negatively impacted by continuous monitoring, building monitoring plans into our interventions can help to mitigate negative nontarget effects as they arise. Such monitoring has already been incorporated into Environmental Impact Assessments that are used to predict the effects of human development on local ecosystems for regulatory purposes. Although monitoring welfare is probably much more complicated, we (a small, underfunded group of researchers) don’t need to pursue it alone. We can adopt and use tools developed by other fields to improve our approach, and develop collaborations to pursue mutually beneficial goals.
An argument that focuses exclusively on uncertainty is not a good reason to dismiss addressing wild animal welfare as an important and urgent project. Although we should be trying to improve our understanding of nontarget effects, we shouldn’t dismiss wild animal welfare as a cause area because there are nontarget effects that we can’t predict.
Uncertainty in wild animal welfare we should care about
The uncertainty that we should be most concerned about is uncertainty that impacts our current prioritization work. Right now, the most pressing open questions are those of invertebrate sentience, and the population mean or median sign of animal lives.
These two problems in particular will directly inform what interventions are most pressing to pursue. If we are confident that there is a nontrivial chance that many or most invertebrates are moral persons, the focus and complexity of intervention design might increase dramatically. Similarly, if we determine that most animals have net-positive lives, we might focus on interventions that reduce pain instead of ones that impact populations. This is because reducing populations of animals might be preferable if we think that they have net-negative lives, but reducing the painfulness of existence for a population is good for both animals with net-negative and net-positive lives.
It is worth noting the possibility that we do not make progress on either of these questions, if they are really hard or simply impossible to answer without some amount of subjective guessing. If this is the case, it might be better for advocates to be risk-averse. Perhaps we should assume that animals have net-positive lives or that most invertebrates suffer because actions on the basis of these views might be good regardless of the truth of those claims, even if those actions aren’t the most cost-effective. Additionally, we might only want to pursue interventions that are moderately to completely reversible in case new information changes our beliefs after we begin an intervention. These considerations need further research. For suffering-focused advocates, assuming that animals have net-positive lives might actually be a barrier to doing the most impactful work.
Note that wild animal welfare is important to work on even if animals generally have net-positive lives. Charities that work on human poverty usually assume that humans have net-positive lives, yet still see their interventions as making the world even better. Similarly for wild animals, there might be inexpensive ways to reduce large amounts of suffering for animals with already positive lives, such as eliminating a painful disease. Because there are so many wild animals in relatively malleable ecosystems, we can reasonably expect it to be inexpensive to improve their lives. So although determining that these lives are net-positive might change what interventions we pursue, it doesn’t change the urgency of working on improving wild animal welfare.
Conclusion
If we accept that wild animal suffering is a problem, but think that nontarget uncertainty is a good justification for not pursuing interventions, we should be skeptical of all large-scale charitable endeavors on the same grounds. Of course, we aren’t concerned about the impact of reducing factory farming or ending global poverty on wild animals lives, in part because the short-term consequences of those projects are so beneficial. This is the right approach to take with wild animal suffering as well. Instead of focusing so much on nontarget uncertainty, we should be working to develop tools to minimize unpredicted outcomes of our interventions, and design interventions to be both reversible and malleable in the face of new information. Although we should be trying to improve our understanding of nontarget effects, and improve our ability to predict the outcomes of interventions, we shouldn’t dismiss the project of reducing wild animal welfare outright on the basis of the inherent unknowability of the future.
Nevertheless, uncertainty about invertebrate welfare and the sign of animals’ lives is important to consider early. We need to know which animals to target with our interventions, and how our interventions should function to make the world better. In the face of uncertainty, it might be important to assume that most invertebrates are moral patients, and have net-positive lives, so that we neither proceed down irreversible and ultimately bad paths with intervention projects, nor spend the limited resources available to address these issues helping fewer rather than more animals.
Uncertainty and Wild Animal Welfare
This piece is part of a short series on how the Wild Animal Initiative research team thinks about uncertainty. These posts represent the opinions of individual staff members, and not necessarily the organizational position of Wild Animal Initiative on uncertainty. This summer, we are raising $50,000 to further our research—you can learn more here and support our work here. This post is crossposted from the WAI blog.
Introduction
Many critiques of improving wild animal welfare focus on the nontarget uncertainty of our actions. We will describe what we mean by “nontarget” below, but in general the term refers to unintended consequences of an intervention. These critics argue that because we are unable to predict the nontarget effects of interventions, we can’t reasonably say that interventions will actually reduce wild animal suffering. For example, if we introduce a vaccine to a disease that we know causes raccoons to suffer, we might change the population size or change the dynamics of the ecosystem in a manner that has consequences for other animals’ wellbeing in unpredictable ways.
Critiques based on nontarget uncertainty should not be viewed as unique to wild animal welfare. If there is a strong case for dismissing a cause area on the basis of nontarget uncertainty, then we’d have good reason to dismiss many large-scale charitable projects, including addressing global health and factory farming. Although this critique fails to apply exclusively to wild animal welfare, we should still strive to improve our ability to predict nontarget consequences of our interventions in all cause areas, and work to resolve some types of uncertainty that are particularly relevant to wild animal welfare.
The argument for caution in the face of nontarget uncertainty
The argument against taking action due to nontarget uncertainty is usually framed as follows:
1. We are uncertain, and in some cases clueless about the nontarget effects of major wild animal welfare interventions on ecosystems.
2. Nontarget effects might have negative ecological or welfare consequences.
P. We should not pursue major interventions if we want to avoid negative ecological or welfare consequences.
This argument generally seems to assume that wild animal suffering matters morally when posed as an objection to wild animal welfare projects. It takes the form of “I agree we should do something about that but…”. The argument might still apply if one does not think wild animal welfare matters morally, but would instead focus on the nontarget effects for humanity of changing an ecosystem.
Target and nontarget effects
Before discussing this argument, we need standardized terminology. Unintended consequences, in various fields, have been called downstream, knock-on, or flow through effects. However, a more general way of conceptualizing the effects of interventions when thinking about uncertainty is to divide them into “target” and “nontarget” effects.
Target effects are both primary and intended. For example, an animal rights activist showing a documentary might have the immediate target effects of causing someone to reduce their meat intake and feel more compassion for animals, and have less immediate target effects like a reduction in the number of animals born on factory farms. The animal rights activist intended for all these things to happen, and they were the direct targets of the intervention.
Nontarget effects are the unintended consequences of interventions. The same documentary might cause a well-intended but less thoughtful watcher to eat more chicken instead of beef, causing more animals to be born into factory farms, due to chickens’ smaller size. Another nontarget effect might be that fewer cows being fed means that some soy farms fail, and are converted to lettuce farms, resulting in a more rabbits to be born due to an abundance of new food (note that this is just a theoretical example). These nontarget effects might happen simultaneously with a target effect—despite causing more animals to suffer overall, the watcher might have decreased their intake of animals calorically, and might feel more compassion for farmed animals.
In the case of interventions to reduce wild animal suffering, the target effects tend to impact the welfare of animals. The nontarget effects might also be related to welfare, or could be related to ecological preservation. Depending on our moral values, we might weigh these nontarget effects differently. For example, a project to make insecticides less painful would, in a perfect implementation, not change the number of target insects killed by the insecticide. However, if the insecticide leaches into the soil, there might be nontarget effects worth considering—perhaps the original insecticide is safe for annelids in the soil, and the less painful one is toxic. Then, we might expect annelid populations to decrease, leading to less food for local birds and a variety of uncertain outcomes for local animals, both from an ecosystem stability and a welfare perspective.
Nontarget uncertainty: not just for wild animal welfare
Of course, wild animal welfare interventions aren’t the only projects to make the world better that involve massive uncertainty. Eradicating the raising and slaughtering of animals for food would shift the use of 36% of crops (by calories). This shift would be a massive change in land use globally, and given the inherent inefficiency of animal agriculture, might cause much of this farmland to rewild. The change in land use could have huge ecological consequences. Populations of local animals would change dramatically. Likewise, addressing global poverty could introduce large-scale industrialization or extraction industries into new regions, having similarly dramatic consequences for local ecosystems. If we care about wild animal welfare, or accept that wild animal suffering is an issue, we should be equally concerned about the nontarget effects of all global development.
Additionally, other charitable projects face their own major nontarget uncertainty independent of wild animal welfare. Global development risks spillover effects. Any effort to impact the far-future might involve a high degree of cluelessness. Efforts to expand cage-free egg supply might lead to individuals being more comfortable with eating eggs, causing more harm in the long run to animals.
Of course, we should still work to reduce the suffering caused by factory farming and should try to reduce global poverty. And for wild animal welfare interventions, it is true that we have comparatively less research on even target effects than we do for farmed animal welfare—farmed animals live in highly regulated environments that humans observe on a daily basis, while we know next to nothing about the lives of most wild animals and the impact of changing environments on their welfare. However, for many interventions, we can make some claims regarding the target outcomes, and we can use insights from conservation and other fields to assess the potential short-term nontarget effects.
Appeals to nature and uncertainty
Many of the problems faced by wild animals can be distinguished from some of the other forms of suffering that have received attention from the effective altruist community—humanity is probably not the proximate cause of most wild animal suffering. In theory, the causes of most wild animal suffering are entirely natural.Yet arguing that we should treat wild animal welfare issues differently from other morally urgent issues because they are natural is an instance of the “appeal to nature” fallacy, if we are assuming the conditions wild animals experience are good because they are natural. But it is not clear that naturalness is a morally relevant feature in any sense.
Similarly, we must also be careful not to resort to appeals to nature when critiquing interventions on the basis of nontarget uncertainty. When we say “an intervention has uncertain nontarget effects and that is bad” it seems that we may sometimes be masking the fallacious belief that “an intervention might change an ecosystem fundamentally, and that is bad in and of itself.”
Of course, we expect our interventions to change ecosystems. That’s the point. To make an intervention resilient, and last through many generations, we should expect it to change fundamental features of that ecosystem. If we are concerned about ecosystems changing for aesthetic purposes (i.e. we find nature to be very beautiful), that’s unrelated to ethical questions. And if we are concerned about ecosystems changing for reasons related to human survival (i.e. climate change will be harmful), that is an important moral consideration, but not the only one. In all of these situations, wild animal welfare still matters.
Our expectation cannot be that ecosystems will not change. Climate change has already eliminated many species that were alive only a few decades ago. Animal species are going extinct on a daily basis. Human development has completely changed many wild spaces, and human activity has enormous effects on ecosystems globally. The hope that these systems will not change is not only morally irrelevant, it is extremely unlikely to be realized regardless of what actions we take on behalf of wild animals. The future of conservation and wild animal welfare requires active intervention and stewardship of wild places.
Finally, it is worth noting that while in the case of addressing wild animal welfare the appeal to nature might seem especially relevant, we can see many existential risks as being naturogenic, such as biosecurity threats, or major asteroid impacts. Similarly, many diseases that global health initiatives work to prevent are themselves naturogenic. Given that appeals to nature aren’t applied as objections to working on these causes, we should also not take them seriously for wild animal welfare work.
Further research on outcomes should still happen
Of course, we need to continue researching the nontarget effects of our actions to improve wild animal welfare. Fortunately, fields like restoration ecology have already made significant advances on collecting evidence related to and predicting the effects of human intervention in nature. We can decrease our uncertainty about the target and nontarget effects of our interventions by continuing to fund research on predicting the outcomes of human interventions in ecosystems, especially as it applies to wild animals.
We will never have absolute certainty about the nontarget effects of our interventions, but we can work to mitigate negative effects by developing better tools to monitor and react to unexpected effects as they arise. Although the cost-effectiveness of an intervention would be negatively impacted by continuous monitoring, building monitoring plans into our interventions can help to mitigate negative nontarget effects as they arise. Such monitoring has already been incorporated into Environmental Impact Assessments that are used to predict the effects of human development on local ecosystems for regulatory purposes. Although monitoring welfare is probably much more complicated, we (a small, underfunded group of researchers) don’t need to pursue it alone. We can adopt and use tools developed by other fields to improve our approach, and develop collaborations to pursue mutually beneficial goals.
An argument that focuses exclusively on uncertainty is not a good reason to dismiss addressing wild animal welfare as an important and urgent project. Although we should be trying to improve our understanding of nontarget effects, we shouldn’t dismiss wild animal welfare as a cause area because there are nontarget effects that we can’t predict.
Uncertainty in wild animal welfare we should care about
The uncertainty that we should be most concerned about is uncertainty that impacts our current prioritization work. Right now, the most pressing open questions are those of invertebrate sentience, and the population mean or median sign of animal lives.
These two problems in particular will directly inform what interventions are most pressing to pursue. If we are confident that there is a nontrivial chance that many or most invertebrates are moral persons, the focus and complexity of intervention design might increase dramatically. Similarly, if we determine that most animals have net-positive lives, we might focus on interventions that reduce pain instead of ones that impact populations. This is because reducing populations of animals might be preferable if we think that they have net-negative lives, but reducing the painfulness of existence for a population is good for both animals with net-negative and net-positive lives.
It is worth noting the possibility that we do not make progress on either of these questions, if they are really hard or simply impossible to answer without some amount of subjective guessing. If this is the case, it might be better for advocates to be risk-averse. Perhaps we should assume that animals have net-positive lives or that most invertebrates suffer because actions on the basis of these views might be good regardless of the truth of those claims, even if those actions aren’t the most cost-effective. Additionally, we might only want to pursue interventions that are moderately to completely reversible in case new information changes our beliefs after we begin an intervention. These considerations need further research. For suffering-focused advocates, assuming that animals have net-positive lives might actually be a barrier to doing the most impactful work.
Note that wild animal welfare is important to work on even if animals generally have net-positive lives. Charities that work on human poverty usually assume that humans have net-positive lives, yet still see their interventions as making the world even better. Similarly for wild animals, there might be inexpensive ways to reduce large amounts of suffering for animals with already positive lives, such as eliminating a painful disease. Because there are so many wild animals in relatively malleable ecosystems, we can reasonably expect it to be inexpensive to improve their lives. So although determining that these lives are net-positive might change what interventions we pursue, it doesn’t change the urgency of working on improving wild animal welfare.
Conclusion
If we accept that wild animal suffering is a problem, but think that nontarget uncertainty is a good justification for not pursuing interventions, we should be skeptical of all large-scale charitable endeavors on the same grounds. Of course, we aren’t concerned about the impact of reducing factory farming or ending global poverty on wild animals lives, in part because the short-term consequences of those projects are so beneficial. This is the right approach to take with wild animal suffering as well. Instead of focusing so much on nontarget uncertainty, we should be working to develop tools to minimize unpredicted outcomes of our interventions, and design interventions to be both reversible and malleable in the face of new information. Although we should be trying to improve our understanding of nontarget effects, and improve our ability to predict the outcomes of interventions, we shouldn’t dismiss the project of reducing wild animal welfare outright on the basis of the inherent unknowability of the future.
Nevertheless, uncertainty about invertebrate welfare and the sign of animals’ lives is important to consider early. We need to know which animals to target with our interventions, and how our interventions should function to make the world better. In the face of uncertainty, it might be important to assume that most invertebrates are moral patients, and have net-positive lives, so that we neither proceed down irreversible and ultimately bad paths with intervention projects, nor spend the limited resources available to address these issues helping fewer rather than more animals.