The first impression though is that animal charities should be accepted as more effective until proven otherwise by some large positive AMF flow-through effect that outweighs saving a life (maybe reducing insect populations?) Until then it seems much more straightforward to donate to ACE charities, specifically the cage-free ones.

Thanks for this, hadn't seen that link before.

One point made there is that "likely interventions in human welfare, as well as being immediately effective to relieve suffering and improve lives, also tend to have a significant long-term impact... By contrast, no analogous mechanism ensures that an improvement in the welfare of one animal results in the improvements in the welfare of other animals." An important long-term consideration for the effects of welfare reforms is whether they generate more momentum for further reforms for animals and for expansion of the moral circle, or whether they generate complacency. I'm currently very uncertain on this, though lean slightly towards momentum. See here for relevant considerations and evidence.

Some other posts related to considering the long-term effects of animal advocacy interventions:

1) Jacy Reese, "Why I prioritize moral circle expansion over artificial intelligence alignment"

2) Me, "How tractable is changing the course of history?" (see especially some of the considerations in "How tractable are trajectory changes towards moral circle expansion?")

3) Brian Tomasik, "Charity Cost-Effectiveness in an Uncertain World" (not necessarily specific to animal issues, but I think there is some v useful theoretical discussion)

You are relying quite heavily on the 18 animals/dollar figure, which seemed very high to me. Does it really seem likely that spending 6 cents on corporate outreach can save a life that brings several dollars in profit per month?

I don't really see a strong connection between these two numbers. To start, it's primarily not sparing animals from factory farming but changing the conditions in which they live. That being said, welfare reforms do tend to increase food prices, and assuming animals don't produce any less food per animal, the number of animals used per year would tend to decrease, too (but not necessarily the number of animals alive at any time, e.g. if companies switch to slower growing broiler chickens, there may be more of them alive at any moment because they have to be alive longer before they're slaughtered). A company has to consider what happens to their profits if they're viewed as particularly unethical. However they respond, their expected profits will suffer because of a campaign.

There's also been recent research by Rethink Priorities and Charity Entrepreneurship on this.

In fact, it seems that ACE has updated this to less than one for THL in 2018. Granted, ASF, another ACE top charity, does maintain a very high figure, however, I'm concerned by this fluctuation year-over-year that we saw with THL. It seems driven by one-off wins, not a sustainable, predictable pattern.

Is this just because of data on new campaigns, or did they also change their model? Or has THL's work focus changed?

I'm not sure if this is an implicit assumption in your comment, but it does sound like you think credit should not sum past 100%, but this need not be the case. Abstractly, if A and B are together necessary and sufficient causes for C, then A and B both deserve 100% of the credit for C, if C happens. For example, biological parents are 100% responsible (in causal terms) for everything that happens to their children and that their children do, because if they hadn't had their children, they wouldn't exist for anything to happen to them or for them to do anything. Similarly, you are 100% responsible for your own veganism, but others are also partially responsible for it, too.

Credit should be thought of in causal terms: what would have happened otherwise. If, without the involvement of a given charity, a given company would not have made (with certainty) the switch but did in fact make the switch, then that charity is 100% responsible. When you consider the possibility that they might have made the switch regardless (at the same time or later) or to a smaller extent, things get much more complicated.

I think that when someone puts a number on an unknown value, the only good response is to say whether it's too high or too low. Merely describing the uncertainty doesn't get us anywhere closer to knowing where to donate. Animal charities could easily be better than the OP suggests.

Since there are less than 1 million elephants alive today, even if each elephant has modestly more moral value than each human, elephant welfare is still very unlikely to meet the importance criteria.

The idea of basing the calculation on a simple neuronal count is flat out wrong, because humans aren't even at the top in an even, 1:1 weighting in that regard, see: https://en.wikipedia.org/wiki/List_of_animals_by_number_of_neurons .

Although I suspect this is more likely to be false than true, it is not inconceivable that less intelligent animals of a given species could matter more than humans, individually. For example, their experiences, good or bad, could be more intense than ours, or they could experience life more quickly*. They don't need to have more neurons for this to be true, either, and I am skeptical of the importance of neuron count, too, in part because of this.

*if the rate didn't matter, you'd run into problems with the theory of relativity: if you are moving very fast compared to another person, you each will see the other as aging more slowly, all else equal. If the rate didn't matter, then you'd each see the other as mattering more, all else equal, because the other would live longer. Then ethics would have to depend on the frame of reference, which is pretty weird, but perhaps not fatal.