Director of Operations at Rethink Priorities.

I previously co-founded and served as Executive Director at Wild Animal Initiative and one of its predecessor organizations, Utility Farm. I also ran corporate animal welfare campaigns for Mercy For Animals.

I run a monthly newsletter on invertebrate welfare. Archives are here: https://www.invertebratewelfare.org/newsletter


Best places to donate?

In these cases, it's likely that you're getting better returns on credit card fees than giving directly to 22 charities, but marginally worse efficiency on processing costs, since it is probably around the same processing  cost for all 22 charities, and also a processing cost at The Life You Can Save, etc.

Based on this, from a pure cost-to-programs view, I'd guess that if it is split up among at least 3 or 4 charities or more, the credit card fee benefits will outweigh the lower efficiency from the processing, so it is probably usually worth giving to something like the GiveWell maximum impact fund or TLYS, or the EA Funds, etc.

Also, I think getting all the benefits  you also get from giving via those funds, like the ones esentorella describes, makes it especially worthwhile to continue giving via those funds (e.g. their research and understanding about how to optimally redistribute the funding).

Best places to donate?

So for us currently, the processing time doesn't change much depending on the frequency of the gift. But I will say that we don't invest a lot in improving this process, because overall we don't put much time into it, so maybe at an organization with a way higher volume of small donations, they would have automated some of what I described. That being said, I've worked at a handful of fairly large nonprofits that are still doing this manually, so it just is going to depend on the organization, and 5min / transaction seems like a safe bet. There are other costs besides just processing - e.g. a bit of time spent on end-of-year donation receipts potentially, etc., that you might counterfactually cause the nonprofit to incur. 

If you're doing an ACH transfer monthly, the costs will vary by provider, but they are a lot cheaper than credit cards. I think Plaid, which is a really common ACH service, costs around 0.80%, capped at $5, or something like that. There is more variation here across providers though, unlike credit cards, where there seem to just be industry standard rates.

I do think there are some genuine advantages to giving monthly, or on a recurring basis, to some organizations. A lot of EA organization keep pretty large amounts of reserves in the bank (at least, compared to similarly sized other nonprofits I've seen, it seems like). But if you're giving to a small organization that doesn't have large reserves, the cashflow certainty that comes with monthly donations might offset to a large degree the extra processing costs, etc.

If I have to guess the best way to give to reduce costs, stress, etc. on an organization, I'd guess it would be a lump sum given in maybe March-May, when the fundraising season is in a lull, but I really think that's probably only very marginally better than giving in whatever way works best for you, monthly, etc., so if monthly helps someone give with less stress, it's probably still worth doing. And if it is a small organization, monthly can really help.

Best places to donate?

I believe the typical nonprofit credit card rate (for Visa and Mastercard) is 2.2%+0.30 USD. So for 5 x $50 donations, it's costing around $1.4*5 = $7 to process your credit card payments across all organizations. For my organization, entering a donation in whatever systems we enter it in probably takes around 5 minutes. I'd guess that with taxes, etc., the average EA nonprofit ops person costs around $40 / hour, so that's another $3.33 per donation = $16.67 across all donations. So of your $250, around $23.67 is going to overhead costs.

If you gave it in one gift, the fees would be $5.80, and you'd only have $3.33 in other ops costs, so $9.13 total.

So basically giving to one organization would increase the amount of your donation going to non-overhead things by probably around $14 / mo, or $168 / year, assuming the organization has other things for the operations staff to do in the 20 minutes they are saving by not processing the extra donations.

Silk production: global scale and animal welfare issues

Hey Michael,

Yeah it's a good point that if you're heavily discounting silkworms, the fact that the population isn't several orders of magnitude larger than chickens might be a reason to not prioritize them, though I am pretty uncertain about, if discounting ought to happen, how we ought to approach (I'm definitely not confident we should just look at neuron counts, but am not very bullish on any other approaches).

I do think that it might be significantly easier to reduce silkworm farming than chicken farming though. For example, I don't know if there are any major example of something similar to the ASOS ban on silk for chickens, like a major retailer going fully vegetarian, or just not selling chicken for animal welfare reasons. It seems plausible that silk bans would be much easier to achieve than these, so it might still be more cost-effective to work on silk. I'm just speculating here though - I don't really have any empirical information, but would definitely support a group trying targeted silk campaigns to compare the ease to current animal welfare campaigns.

Silk production: global scale and animal welfare issues

I didn't research it in detail, and mention in the introduction that I basically assume that they are for the purposes of looking into this, and that this assumption shouldn't be taken for granted if one was prioritizing work based on this. I don't really think this is an answerable question right now with currently available information, and I don't know how to discount on the basis on that uncertainty. I personally am fairly sympathetic to treating many kinds of insects as capable of having valenced experiences, on the basis of Rethink Priorities' work in that space (though they didn't look at larval silk specifically), but when I do research in this space, part of the purpose is just purely fact finding. There are a bunch of industries that use billions and trillions of animal, and very little work has been done to study them from an animal welfare lens. At a minimum it seems worth someone spending a few hours considering each of these industries from an animal welfare lens, so I've been doing that.

However, I will note that I think that silk bans seem to be possibly also be net-good for humans - silk production seems to involve a fair amount of human rights abuses, including slavery, child labor, etc., and has been campaigned against by human rights groups extensively (see that historical advocacy section for a bit more detail). It seems that several human rights groups explicitly have worked on securing silk bans. I'm not certain of the scale of these harms, so I am not recommending those campaigns from an EA-perspective (or animal welfare silk campaigns for that matter), but I do think that it's fairly plausible that the benefits of industrial silk do not outweigh the harms to humans.

I think that promoting silk alternatives for its industrial / commercial uses you mention, like the Material Innovation Initiative does, is a pretty promising route to both reduce human and potential animal suffering.

Silk production: global scale and animal welfare issues

Thanks for the question - I didn't look into it in too much detail, but my impression is that India is actually the largest importer of silk (mostly from China), and not a very large exporter, suggesting that there and in China are the largest markets. I believe the EU, Japan, and South Korea are fairly large as well.

I didn't look into interventions directly outside of the speculation on them listed here, but I'd be interested in the reasoning / evidence that bans in areas with low use being more tractable. I assume that this means there is some ideal level of use / cultural relevance that means both maximal impact vs highest tractability.

How we averted 130,000 animal deaths (in expectation) with a volunteer campaign.

This is great! Thanks for the write up!

One other assumption that jumps out to me (represented in your model under "School meals affected per year (190 days)")

If I recall correctly, HSUS in the US originally sought Meatless Monday commitments, but found that many school districts, etc., that committed, didn't actually reduce their purchasing  that much (or at all) - they ended up just making their normal orders for meat, and added some veg stuff on top of that. This likely meant that these districts ended up serving more meat on non-Mondays. So, they changed their ask to a "20% overall reduction in meat purchases". This might mean the effectiveness is unfortunately a bit lower, if this is generally the case (though for HSUS it was for US school districts, so purchasing might work differently in the UK).


I wonder if there are a lot of low-hanging fruit for these campaigns around the world. I imagine there are a fair number of local animal advocacy groups who are really well positioned to do this advocacy, and my impression is that some of these might be really easy to get to change - e.g. a teacher in a district having a few conversations with the right people and bringing them some information.

New Top EA Causes for 2021?

Out of curiosity I stuck an episode into the Wub Machine.  It's genuinely mildly listenable. Also takes no time so the cost-effectiveness here might be high. Original audio: 80,000 Hours.

New Top EA Causes for 2021?

Working title: Reversetermism

Longtermists have pointed out that we've often failed to consider the interests or wellbeing of future beings. But an even more neglected space is the past.

If we think that existential risk is sufficiently high in the near future, there is a good chance that the vast majority of moral value is in the past. Just considering humans, there are at least 300,000 years of experiences, all of which we ought to consider just as important as present day ones. If we consider non-humans' interests, there are billions of years and countless individuals who we ought to expand our moral circle to include.

The scale here is obvious, as is the neglectedness - as far as I am aware, there are no groups focused on ensuring that the past is as good as possible. So, how tractable is it?

Immediately, a handful of interventions come to find:

  • Cultivating expert backcasting:
    • Written history is just a few thousand years old, and unfortunately, a lot of it is incredibly sad. But prior to around 5,500 years ago, we have little data on what human lives were like. By improving our backcasting ability, we can ensure that documentation of these lives in the prehistoric world states they were as good as possible.
  • Making sure there were no existential catastrophes
    • If a x-risk is bad right now, it stands to reason that it might have been even worse had it occurred in the past. We might be able to verify that existential catastrophes did not happen previously, preventing the flourishing of both present day and future humans.

One immediate advantage of reversetermism is that cost-effectiveness can actually be estimated relatively accurately. Here's a simple test:

"On May 5th (Gregorian calendar), 10,560 BC, at 2:00pm Eastern, everything was chill for an hour for everybody."

This expert backcasting took around 12 seconds to produce. Assuming a human population of 2 million, and that you pay expert backcasters $30 USD / hour, this cost $0.10, and created around 228 years of good experiences. With an average lifespan of say 30 years, it costs around $0.013 to save a life. And even more expert backcasters might achieve more efficient results through further work in the field, driving down the cost-effectiveness further.

Load More