July 27, 2020

Introduction

The future is hard to predict. Studies show that experts’ ability to predict important political events decreases noticeably just a few years into the future (Tetlock 2005). This can make it very hard to figure out what actions will produce the best long-run consequences. One way to partially circumvent this problem is to use indicators, or signposts (Bostrom 2014), of near-term developments which look like they will lead to positive long-term outcomes (but see Tarsney 2019). This post outlines my thoughts on the reliability of indicators of generally positive trajectories and proposes signposts for progress toward improving wild animal welfare over the long term.

Two types of signposts

Given uncertainty about what the best future will look like and how exactly to get there, it might not make sense to chart specific pathways to the endpoint - or it might not be possible to imagine those pathways at all. Signposts can aid our thinking by directing our focus towards clear intermediate objectives closer in time, about which it is easier to make predictions.

One kind of signpost is a proximate measure of a trajectory we expect to be positive in almost all cases. The National Science Foundation (NSF) aims to promote the progress of science and relies on governmental funding to accomplish this mission. Thus, all else equal, increasing amounts of funding going toward the NSF can be thought of as a signpost for improving scientific progress. Even if we don’t know exactly what scientific progress will look like, supporting more research will almost certainly help achieve it.

Or consider a new movement with ambitious goals for social change that could take decades to realize. In the early days, it might not be clear exactly which strategies the movement should pursue. But increasing the number of people involved in the movement should be a reliable signpost for success, because most approaches to social change would benefit from a broader base of support.

Another kind of signpost is the reinforcement of a generally useful mechanism of change. This kind of signpost can set us up to achieve a global optimum even if it means forsaking a local optimum. For instance, chickens in battery cages on factory farms live in horrible conditions, and many animal advocates have pushed for a transition to cage-free operations. There has been a debate, however, over whether the aggression between overcrowded free-roaming chickens will lead to even more suffering than is experienced in the battery cages (Cotra 2017).

If it turned out that free range were actually worse for chicken welfare, we could take this as an indication that we should not advocate for cage-free operations, and that when forced to choose between caged and cage-free, you should choose the caged chicken. But if we examine the mechanisms of positive long-term development, it is unlikely that locking up chickens will constitute a step toward the end of animal farming and the suffering it causes (Cotra 2017; see also this interview with Lewis Bollard for a thorough discussion of the long-term project of ending animal farming).

In other words, it is only a local optimum rather than a global optimum: free range is a step on the way toward providing chickens with the freedom to express and act on their preferences, while battery cages are not. Furthermore, the abolition of battery cages will generate momentum in the animal advocacy community, whereas a begrudging acceptance of them would not. A shift to cage-free farming is a signpost for generally useful mechanisms for abolishing animal farming.

Reasons to be cautious when applying signposts thinking

Signposts are only as useful as they are accurate. All they can really tell us is our progress toward the signposts themselves. Whether that translates into progress toward the ultimate goal depends on whether our assumptions about the signposts are true. Because it’s hard to get direct feedback on the accuracy of signposts, we have to be extra vigilant against cognitive biases and motivated reasoning, and we have to be willing to incorporate uncertainty into our judgment.

As a first step, we need to rigorously examine the assumptions we make about the causal relationship between a signpost and the ultimate goal (see Bostrom 2014 on crucial considerations). In the case of the NSF, are we justified in believing that increasing scientific funding does in fact accelerate scientific progress? Maybe scientific progress is mainly constrained by the amount of available talent?

When our signpost refers to a specific mechanism of change, we need to ask ourselves whether that mechanism will continue to operate in the future. If it turns out that cage-free chickens are worse off than caged chickens, then cage-free reforms can still be useful if they lead to more and more reforms in factory farming. But if that momentum is lost, and future reforms don’t follow, then the path-setting elements of the intervention become irrelevant, and the effects of the cage-free reforms will be limited to their impact on chicken welfare in the short term.

Positive signposts for wild animal welfare

Below I will list a few tentative examples of positive signposts for wild animal welfare.

One example that is core to Wild Animal Initiative’s mission is widespread interest in wild animal welfare research, especially if we currently are able to influence whether or when that interest starts to grow. The proposed mechanism is fairly straightforward: 

1. We currently know very little about how to improve wild animal welfare, and rigorous research seems like the only tool that will allow us to eventually help wild animals in a reliable way.

2. Early research on how to help wild animals might lead to a positive feedback loop, by reducing the feeling of helplessness that can stop people from thinking about it or working to solve it.

It is possible that attracting more research to questions of wild animal welfare will do more harm than good, by missing a key factor (for instance, if the research community makes false assumptions about what is valuable, or implicitly assumes that animals can’t have net negative lives). But it seems like the potential upsides of such research are large and positive, while strong negative consequences are quite unlikely.

Another example of a possible positive signpost is the safe development of methods that are likely to be very useful in the future, such as genetic tools to improve welfare and fertility control for wild animals. In general, I’m less concerned with making the applications of the tools as good as possible for a certain target species in the short term (although we should obviously avoid causing net harm in the short term), and more concerned with making sure that these methods are developed safely, so that they can be put to use once we have perfected them. Seen this way, the top priorities become acting prudently and responsibly, and finding a target population and implementation scheme that together have a small chance of failure and will increase the likelihood of public acceptance of the methods.

Here, as in the earlier example of fostering interest in wild animal welfare research (and almost any other action we take), we cannot be completely sure that our efforts won’t do more harm than good. For instance, it is possible that the safe development of the tools is good, but making the argument for their implementation on the basis of wild animal welfare turns out to hamper progress. However, I don’t think this argument is strong enough to reverse the expected value of advocating for safely developing such methods to help wild animals, as long as the arguments are put forth in a non-polemic and cautious way.

What would change my mind about the value of using signposts in this way?

The main form of evidence that would change my mind on the usefulness of these specific examples of signposts would be indications of public or scholarly opinion shifting away from concern for wild animal welfare. If we were confident that humans wouldn’t care noticeably more about wild animal welfare in the future, and that this fact was very hard to change, I would be much more skeptical of using the aforementioned signposts to guide our actions. Given such evidence, the relative importance of short-term effects would be much higher. But given the fact that humanity’s moral circle seems to have slowly expanded, although with a few noticeable setbacks, I find it more likely than not that the appropriate considerations will be extended to wild animals in the future, and that our descendants will be willing to help them to a significant extent.

References

Bostrom, N. (2014). Crucial considerations and wise philanthropy. [Lecture].

Cotra, A. (2017). New report on the welfare differences between cage and cage-free housing. Open Philanthropy. 

Tarsney, C. (2019). The Epistemic Challenge to Longtermism. [Working paper].

Tetlock, P. E. (2005). Expert political judgment: How good is it? How can we know? Princeton University Press.