How to compare theories of change in animal welfare?

By PabloAMC 🔸 @ 2025-12-28T11:29 (+22)

Almost all the charities recommended by Giving What We Can and Animal Charity Evaluators base their core strategy on corporate outreach, which has been very successful in shifting egg-laying hen production from cages to cage-free methods. A prototypical example of a charity doing this is The Humane League.

A different theory of change is that of the Good Food Institute. Here, the goal is to create a market for alternative proteins that may hopefully replace animal products. My perception is that lately, they have been exploiting the climate change perspective more, perhaps because they intend to attract government funding and sell this more broadly. For example, they sought certification as excellent charity by Giving Green but not Animal Charity Evaluators.

This raises an interesting question: can we create parametric models that would allow us to estimate, at least in principle, how well these theories of change work? Then, we might be able to estimate these parameters from empirical evidence, or at least explore what the effectiveness looks like under different assumptions of those parameters.

Some thoughts on this:

1. The best cost-effectiveness analysis I have been able to find so far is this spreadsheet created by FarmKind and sourced from MieuxDonner, but which is hard for me to follow. Perhaps a good answer to this question would be just explaining that spreadsheet, allowing people to have an intuitive understanding of what model we should be using.

2. There is also this post in the EA forum on whether it is better to donate to The Humane League or invest in a specific alternative protein startup. The model we should be using looks something like $$E\left[M\right(t\left)\right]={\int }_0^1{\int }_0^{\infty }\underset{\text{Adoption Curve}}{\underset{}{\left(\frac{c}{1+e^{-k(t-\tau )}}\right)}}\cdot \underset{\text{Probability of Parity}}{\underset{}{P(T_{parity}=\tau )}}\cdot \underset{\text{Probability of Ceiling}}{\underset{}{P(C=c)}}d\tau dc$$ where $E\left[M\right(t\left)\right]$ represents the expected market share of alternative proteins. The most important limitation of the blog post is that it assumes $c=1$, but there is a chance this parameter depends on culture or regulation, too.

3. My understanding is that the largest open question on corporate outreach is how well this model will replicate beyond egg-laying hens. In fact, in Europe, the labelling of egg production is perhaps the most reliable indicator of animal welfare in animal food production, but nothing similar exists for other animals.

4. Similarly, cultivated meat faces significant technological and even legal barriers in some places. It is also not clear what factors are necessary to make people shift to alternative proteins: perhaps only social norms will work on this problem, and they are annoyingly slow to change. A good starting model might be the Bass diffusion model.

5. I must acknowledge that I have a strong desire and bias for systemic change work (GFI) instead of corporate outreach (THL). But the strong focus of the EA community on corporate outreach and animal advocacy makes me wary of being biased. Because of this, I would like to have a good model of how likely it is for this to work out and when.

Feel free to leave your opinions and thoughts. If we receive some good answers, perhaps we could write a blog post that adds clarity to the community's decisions.

Many thanks in advance!