Thank you for spending time analyzing our methods. We appreciate those who are willing to engage with our work and help us improve the accuracy of our recommendations and reduce animal suffering as much as possible.

Based on previously received feedback and internal reflection, we have significantly updated our evaluation methods in the past year and will be publishing the details next Tuesday when we release our charity recommendations for 2024. From what we can tell from a quick skim, we think that our changes largely address Vetted Causes’ concerns here, as well as the detailed feedback we received last year from Giving What We Can (see also our response at the time) as part of their program that evaluates evaluators. Our cost-effectiveness analyses no longer use achievement or intervention scores, but rather directly calculate cost-effectiveness by dividing impact by cost, as you suggest. That being said, our work will never be perfect so we invite anyone reading this with the expertise to improve the rigor of our work to reach out, now or in the future.

Although your comments are related to methods that we no longer use, we’d like to spend more time understanding and engaging with them, learning from them, and potentially correcting any misconceptions. Unfortunately, we won’t have the opportunity to do so until after our charity recommendations are released next week. Additionally, it might be a comfort to know that for the past few months, Giving What We Can has been assessing ACE’s new evaluation methods along with a panel of other experts and that they intend to publish the results later this month.

Thank you.

- The ACE team

Did you ask ACE to review this before publishing? It seems like the kind of thing that would be worth getting feedback on before publishing. I didn't look at this for more than a couple minutes, but I saw immediately that there might be some conceptual disagreements between you and ACE - for example, I noticed that in your first example, you assume in your example (I believe), that if LIC didn't spend 200k on the lawsuit against Costco, they wouldn't spend it on anything else. It's unclear to me that this is the counterfactual, or how ACE is conceptualizing those funds. There might be reasoning behind their decisionmaking that would be useful to your critiques they could share.

I also felt like this felt pretty politically motivated. Not sure if that is your intention, but paragraphs like this:

ACE's recommendations determine which animal charities receive millions of dollars in donations.^[1] Thus far, we have reviewed 5 of ACE's "Top 11 Animal Charities to Donate to in 2024" and only one of them (Shrimp Welfare Project) appears to be an effective charity for helping animals. ACE's poor evaluation process leads to ineffective charities receiving recommendations, and many animals are suffering as a result.

Without any evidence feels pretty intense. ACE is kind of low hanging fruit to pick on in the EA space, so this read to me like more of that, without necessarily the evidence base to back it. Reading your report, I felt kind of like "oh, there are interesting assumptions here, would be interested to learn more", and not "ACE is doing an extremely bad job." 

E.g. I think the questions that would be good to ask in a critique of ACE might be:

• If ACE didn't exist, how would the funds the direct be spent otherwise? Would that be better or worse for animals?
• Is historical track record / cost-effectiveness the only lens on which to evaluate charities?
  • If the answer is yes, seems very hard to start new things!
  • I don't know if the LIC legal case is this, but celebrating the potential impact of promising bets that didn't pan out seems good to me.

I also think getting feedback on statements like this would be really helpful:

The correct formula for calculating cost-effectiveness is simply impact divided by cost. Rather than using this simple formula, ACE has elected to create a methodology that does not properly account for impact or cost.

I think ACE has wanted to do this at points in their history — my impression is just that it is incredibly difficult, so they've approached it from other angles instead. I also don't think it's clear to me that ACE's goal is to report cost-effectiveness. I think clarifying this with them, and getting a sense of why they don't do what you see as the simple approach would be useful for making this critique stronger. And, I don't think people should make giving decisions based only on historic cost-effectiveness - just because an opportunity was impactful doesn't mean the organization needs more funds to do that work, that it will scale, work in the future, etc.

I don't disagree that ACE might be directing funds to ineffective charities! I don't really think non-OpenPhil EA donors should give to farmed animal welfare, for example. But, I don't think it is obvious to me that ACE going away means money going to more effective charities - I expect it would mostly be worse - people giving to animal charities with basically no vetting.

That being said, critique of critical organizations is great in my opinion, so appreciate you putting this out there!

Thank you for doing this work. I’m very supportive of productive criticism on the Forum. As a moderator, I’d like to recommend this post for tips on how to make criticism more productive. EA is a collective project, and I think that steps such as sharing this feedback with ACE directly and writing a less aggressive title for your post would improve the outcomes of this work.

Their evaluation process has been updated (e.g. here), and I'm inclined to wait to see their new evaluations and recommendations before criticizing much, because any criticism based on last year's work may no longer apply. Their new recommendations come out November 12th.

FWIW, I am sympathetic to your criticisms, as applied to last year's evaluations. I previously left some constructive criticism here, too.

Thanks for writing this. I feel like the following is the crux of your criticism of LIC:

ACE acknowledges the lawsuit was dismissed, but still celebrates this achievement. They note that this achievement would inspire similar lawsuits. Would it be good to inspire more lawsuits that cost $200,000 and are dismissed?

You state this as though the answer is "obviously no" but the answer feels extremely nonobvious to me. I note that you excluded the some key things when quoting ACE:

LIC’s first lawsuit, a shareholder derivative case against Costco’s executives for chicken neglect, was featured on TikTok and in multiple media outlets, including CNN Business, Fox Business, The Washington Post, and Meatingplace... We thought the achievement has strong potential for indirect impact, and it received a high amount of media attention. - ACE

The Facebook fan page still as of this writing has a post about the lawsuit pinned to the top because apparently the owner decided to boycott after learning about the cruelty.

It sounds like the Costco board also had to take official action:

In a letter dated August 15, 2023, Costco’s board stated that it had “formed a Board committee to review and investigate the demand’s allegations.” LIC’s shareholder clients then met with investigators retained by the committee. - LIC

Is it worth $200k to get a bunch of bad publicity for Costco, force the board to form a committee and hire an investigator, etc.?

I don't know, I'm pretty willing to believe that the answer is "no", but it doesn't seem obvious to me. I could pretty easily believe that the CEO of the next company they sue would to change their policies instead of having to deal with the embarrassment of asking the board to form a committee to investigate.

So, I have some mixed views about this post. Let's start with the positive.

In terms of agreement: I do think organizational critics are valuable, and specifically, critics of ACE in the past have been helpful in improving their direction and impact. I also love the idea of having more charity evaluators (even in the same cause area) with slightly different methods or approaches to determining how to do good, so I’m excited to see this initiative. I also have quite a bit of sympathy for giving higher weight to explicit cost-effectiveness models when it comes to animal welfare evaluations.

I can personally relate to the feeling of being disappointed after digging deeper into the numbers of well-respected EA meta organizations, so I understand the tone and frustration. However, I suspect your arguments may get a lot of pushback on tone alone, which could distract from the more important substance of the post and concepts (I’ll leave that for others to address, as it feels less important, in my opinion).

In terms of disagreement: I will focus on what I think is the crux of the issue, which I would summarize as: (a) ACE uses a methodology that yields quite different results than a raw cost-effectiveness analysis; (b) this methodology seems to have major flaws, as it can lead to clearly incoherent conclusions and recommendations easily; and (c) thus, it is better to use a more straightforward, direct CEA.

I agree with points A and B, but I am much less convinced about point C. To me, this feels a bit like an isolated demand for methodological rigor. Every methodology has flaws, and it’s easy to find situations that lead to clearly incoherent conclusions. Expected value theory itself, using pure EV terms, has well-known issues like St. Petersburg Paradox, optimizer's curse, and general model mistakes. CEAs in general share these issues and have additional flaws (see more on this here). I think CEAs are a super useful tool, but they are ultimately a model of reality, not reality itself, and I think EA can sometimes get too caught up in them (whereas the rest of the world probably doesn’t use them nearly enough). GW, which has ~20x the budget of ACE, still finds model errors and openly discusses how softer judgments on ethics and discount factors influence outcomes (and they consider more than just a pure CEA calculation when recommending a charity).

Overall, being pretty familiar with ACE’s methodology and CEAs, I would expect, for example, that a 10-hour CEA of the same organizations would be quite a bit further from the truth of the actual impact or effectiveness of an organization. It's not clear to me that spending equal time on pure CEAs versus a mix of evaluative techniques (as ACE currently does) would lead to more accurate results (I would probably weakly bet against it). I think this post overstates the importance of discarding a model due to a flaw that can be exploited.

A softer argument, such as “ACE should spend double the percentage of time it currently spends on CEAs relative to other methods” or “ACE should ensure that intervention weightings do not overshadow program-level execution data,” is something I have a lot of sympathy for.

If you're correct in the linked analysis, this sounds like a really important limitation in ACE's methodology, and I'm very glad you've shared this!

In case anyone else has the same confusion as me when reading your summary: I think there is nothing wrong with calculating a charity's cost effectiveness by taking the weighted sum of the cost-effectiveness of all of their interventions (weighted by share of total funding that intervention receives). This should mathematically be the same as (Total Impact / Total cost), and so should indeed go up if their spending on a particular intervention goes down (while achieving the same impact).

The (claimed) cause of the problem is just that ACE's cost-effectiveness estimate does not go up by anywhere near as much as it should when the cost of an intervention is reduced, leading the cost-effectiveness of the charity as a whole to actually change in the wrong direction when doing the above weighted sum!

If this is true it sounds pretty bad. Would be interested to read a response from them.

Of course, the other thing that could be going on here, is that average cost-effectiveness is not the same as cost-effectiveness on the margin, which is presumably what ACE should care about. Though I don't see why an intervention representing a smaller share of a charity's expenditure should automatically mean that this is not where extra dollars would be allocated. The two things seem independent to me.

Great analysis, Isaac! I worry the Animal Welfare Fund (AWF) has similar problems (see below), but they are way less transparent than ACE about their evaluations, and therefore much less scrutable. Instead of mostly deferring to AWF, I would rather have donors look over ACE's evaluations, discuss their findings with others, and eventually publish them online, even if they spend much less time on these activities than you did.

AWF only runs cost-effectiveness analysis (CEAs) for a minority of applications. According to a comment by Karolina Sarek, AWF's chair, on June 28 (this year):

In the past, we tended to do CEAs more often if: a) The project is relatively well-suited to a back-of-the-envelope calculation b) A back-of-the-envelope calculation seems decision-relevant. At that time, a) and b) seem true in a minority of cases, maybe ~10%-20% of applications depending on the round, to give some rough sense. However, note that there tends to be some difference between projects in areas or by groups we have already evaluated versus projects/groups/areas that are newer to us. I'd say newer projects/groups/areas are more likely to receive a back-of-the-envelope style estimate.

Comparisons across grants also seem to be lacking. From Giving What We Can's (GWWC's) evaluation of AWF in November 2023 (emphasis mine):

Fourth, we saw some references to the numbers of animals that could be affected if an intervention went well, but we didn’t see any attempt at back-of-the-envelope calculations to get a rough sense of the cost-effectiveness of a grant, nor any direct comparison across grants to calibrate scoring. We appreciate it won’t be possible to come up with useful quantitative estimates and comparisons in all or even most cases, especially given the limited time fund managers have to review applications, but we think there were cases among the grants we reviewed where this was possible (both quantifying and comparing to a benchmark) — including one case in which the applicant provided a cost-effectiveness analysis themselves, but this wasn’t then considered by the PI in their main reasoning for the grant.

GWWC looked into 10 applications:

Of the 10 grant investigation reports we reviewed, three were provided by the AWF upon our general request for representative grants; two were selected by us from their grants database; two were selected by the AWF after we provided specifications; and three were selected by the AWF based on our request for grant applications by organisations that applied to both the AWF and ACE’s MG.

Karolina also said on June 28 that AWF has improved their methodology since GWWC's evaluation:

However, since then, we've started conducting BOTEC CEA more frequently and using benchmarking in more of our grant evaluations. For example, we sometimes use this BOTEC template and compare the outcomes to cage-free corporate campaigns (modified for our purposes from a BOTEC that accompanied RP's Welfare Range Estimates).

I do not doubt AWF has taken the above steps, but I have no way to check it. I think donating to ACE over AWF is a good way of incentivising transparency, which ultimately can lead to more impact.

I strongly upvoted this post because I'm extremely interested in seeing it get more attention and, hopefully, a potential rebuttal. I think this is extremely important to get to the bottom of!

At first glance your critiques seem pretty damning, but I would have to put a bunch of time into understanding ACE's evaluations first before I would be able to conclude whether I agree your critiques (I can spend a weekend day doing this and writing up my own thoughts in a new post if there is interest).

My expectation is that if I were to do this I would come out feeling less confident than you seem to be. I'm a bit concerned that you haven't made an attempt at explaining why ACE might have constructed their analyses this way.

But like I'm pretty confused too. It's hard to think of much justification for the choice of numbers in the 'Impact Potential Score' and deciding the impact of a book based on the average of all books doesn't seem like the best way to approach things?

It feels like this needs a response from both ACE and Legal Impact for Chickens. (I'm not suggesting it should be a quick one, some things are important enough to warrant careful review. I agree with @abrahamrowe  it would probably have been better to ask for their comments before publishing)

• I think it is possible for a charity focusing on taking legal action to be impactful without [consistent] legal success, which the review doesn't really acknowledge. A large part of the theory of change around suing corporate bad behaviour is the idea that it will deter bad behaviour in future, by making standards compliance more cost effective than defending lawsuits
• Deterrent effects however are a more complicated theory of change than actually winning cases and forcing actors to change. And it may be very difficult to have a deterrent effect if cases are typically dismissed.
• To that extent I'm quite surprised to learn that Legal Impact for Chickens apparently hasn't yet had any victories, based on what I had heard about that organization. I don't think this necessarily reflects badly on the organization, which is a young charity focused on a legal process which inevitably takes time. But it does mean the error bars for their impact are rather large, and could mean a nonzero possibility they aren't [yet] having an impact at all. It would be interesting to hear more about metrics used (both by LIC and ACE, and other charities with similar theory of change for that matter) to evaluate the impact of an unsuccessful lawsuit, and how substantial those are.
• Some of the questions raised about ACE's weightings are quite independent from the example given. It would be interesting to hear from ACE if and how evaluation criteria for their [apparently mostly subjective] impact scoring takes into account the idea that a charity could achieve a higher score by subdividing campaigns, and if and how they intend to update impact assessments in cases like the example of books either failing to reach a non trivial number of people or being phenomenally successful even if the case they make for veganism was not originally assessed as particularly evidence-based.

I think this would have been an interesting contribution to the Animal Welfare vs GHD debate week. From the limited amount I read of it, it seemed that even people (on different sides of the debate) whose analysis was very thorough weren't taking account the more straightforward possibility that some of the highlighted top animal advocacy charities simply weren't close to being as effective [yet] at achieving their goals as suggested, regardless of philosophical positions and empirical claims about welfare levels.