Disclaimer: This post isn't a critique of any grant-making or research organisation I know of - I imagine most of them think about cost-effectiveness in a nuanced way, rather than falling into the traps I describe below. However, I do think some individuals fall into these traps, and I thought it could be valuable to the community to capture these ideas in writing.
The potential problem with cost-effectiveness (in animal advocacy)
We can all agree that we want to get the most impact for our buck when we direct resources. A central EA practice, therefore, is to seek out cost-effective projects - those which maximise impact per dollar spent.
In principle, this is great. In practice, if you're an animal advocate, it leads to tricky territory.
It's one thing to assess projects' cost-effectiveness when you are clear on what "impact" actually means. If we are optimising for "lives saved" - as is common in the global health space - then it's sensible to compare different projects and choose the ones which save the most lives per dollar. Likewise, if we're optimising for the reduction of near-term animal suffering, then we can compare welfare improvement projects using a standardised measure of suffering like Ambitious Impact's "Suffering-Adjusted Days" (SADs) and invest in the ones which reduce the most SADs per dollar.
It's another thing entirely to carry out this exercise when we're working towards bigger outcomes and "impact" is therefore more fuzzily defined. Many advocates are working towards a broader project of ending industrial food animal production (IFAP). Here, the ultimate "impact" is, I suppose, "progress towards ending IFAP", which is too broad to practically work with. Therefore, that impact needs to be broken down into more tangible elements. Most advocates agree that ending IFAP will require many different approaches working towards multiple suboutcomes - for example, delegitimising consumption of animal products, increasing demand for alternatives, reducing the economic viability of IFAP, changing individuals' attitudes towards animals, building movement power, and so on. These are the kinds of high-level changes organisations aim to create in the world in order to end IFAP. When organisations are working towards such a wide range of sub-outcomes, it's much harder to come to a standardised definition of "impact".
Imagine I needed to decide which of these is most cost-effective: a project lobbying a national government to reduce barriers to alternative proteins; a project campaigning for default plant-based catering in universities; a project building a grassroots base of autonomous activists; or a project taking legal action against a mega-corporation for engaging in illegal yet common animal farming practices. Each of these is attacking IFAP from a different angle. Therefore it's extremely hard - if not outright impossible - to compare their cost-effectiveness, because each project's "unit of impact" is completely different.
A second issue is that cost-effectiveness will always be easier to calculate for projects focused on short-term, highly measurable outcomes. It will therefore also bias our attention towards these kinds of projects. The problem is that it's highly unlikely that those types of outcomes just also happen to encompass all the most impactful things we could be doing to end IFAP.
When cost-effectiveness clearly makes sense
I think there are two situations where cost-effectiveness is clearly a highly useful tool, even when we are working on bigger, fuzzier outcomes.
The first situation is when you're comparing projects which are all situated within the same strategy. For example, if I'm comparing five different projects all aimed at increasing alt-protein market share in Europe, then it makes sense to direct more resources to the projects which most cost-effectively increase European market share. Here, because the projects share the same strategy, their "unit of impact" is the same, and we can therefore make an apples-to-apples comparison.
The second situation is when we are comparing broad sets of approaches where it's evidently clear that there's a huge gulf between those approaches. For example - and these are purely fictitious examples, I'm not saying they're true - imagine a world where we realised projects seeking to change young peoples' attitudes through school outreach achieve virtually nothing, even when tried from many different angles; whereas we do have some idea that projects to launch pro-animal political parties do make some difference. In that case, we can clearly say the latter is more cost-effective, and therefore worthy of investment, than the former, even when their "units of impact" are totally different.
So what about when things aren't so clear-cut?
This post aims to raise awareness about pitfalls around how we use cost-effectiveness to make decisions, rather than provide a clear way forward. However, I will close this post by offering some rough heuristics.
When you deal in the realm of bigger, long-term change, such as ending IFAP, you will always be gambling. But I think we can be good gamblers by asking a few key questions, such as:
- Is it strategic? In other words, does the project produce higher-level outcomes that we think need to happen to effect bigger change?
- Is it needed? Is this project filling a gap in the ecosystem, or are we already overinvested in this sort of thing?
- Is it likely to succeed? Is the environment conducive to this project getting somewhere? What's the track record of the project team?
- Is it efficient? Compared to other projects aiming at the same outcome, will these project create more progress towards that outcome with the same resources?
Projects that fare best against these questions are the ones we should invest in.
Thanks for the post, Dilan.
1 SAD corresponds to 1 human-day of disabling pain. So it is not time-bound. It can be used to measure reductions in suffering even if they happen far into the future (although I guess more than 90 % of the overall effect materialises in the 1st 100 years for the vast majority of interventions).
There is a sense in which reducing animal suffering (as measured by SADs) is a broader project than ending large-scale farming? Farmed animals account for a tiny fraction of all animals and neurons.
Since you mention "if not outright impossible", do you think ending industrial farming is intrinsically good? I believe what matters is just decreasing suffering, and increasing happiness. So I would ultimately assess interventions decreasing industrial farming in these terms. I worry about using non-welfare final outcomes because animals in large-scale farming can have positive lives, with more happiness than suffering, under some conditions. In this case, they would prefer existing over not existing. I also think positive lives are very much possible in practice. I estimate that the welfare per chicken-year of slower growth broilers is 92.9 % higher than that of fast growth broilers, and that of layers in barns 80.4 % higher than that of layers in battery cages, which suggests lives in the improved conditions that are negative, but close to neutral (estimates of 100 % higher welfare would suggest neutral lives). I guess cows, broilers, and layers in organic production have positive lives.