Well, it's getting to be near the traditional giving season, and it's time to donate a portion of my meager college income. This year, it's looking like I'll be able to afford to donate $800 (~14%). So, now it's time to answer the question that's always on my mind... where should it go? Since I'm looking to make the highest impact possible, I wanted to give a chance to be persuaded one way or another.

Here are my candidates:

Veg Ads (Vegan Outreach / The Humane League)

VegAds are a compelling candidate. Between Brian Tomasik's How Much Is a Dollar Worth? The Case of Vegan Outreach (PDF) and "Donating toward Efficient Online Veg Ads", it seems somewhat well established that you can prevent more than 100 days of nonhuman animal suffering for $1.

The theory seems well established, with data in place supporting click rates and the amount people stay vegetarian. However I do have some concerns:
1.) I'm worried about to what extent the video truly causes vegetarianism, or if people would have gone veg for other reasons absent the video eventually, or if people who saw the video were already veg.
2.) I'm worried that the data, while existent, is very thin. I would want to see a few more intentional replications, perhaps making use of more robust though complex methods.
3.) I'm worried that while the "how long do people stick with it" data is there, it's still unsourced, and isn't specifically related to the video. I have a weak suspicion that those converted by a video alone are less likely to stick with it.

Furthermore, it's worth pointing out that I have committed 10% of my income to my Giving What We Can pledge, which must target global poverty specifically, and I thus cannot sincerely count a veg ads donation to this goal. Right now, I think maintaining GWWC membership and advocating it to others is more effective than my current donation amount, but this might become a more pressing problem in the future.

Lastly, I don't really buy Brian's argument that donating to Veg Ads will primarily impact our future concern for nonhumans, making us more likely to be concerned with wild animal suffering or less likely to spread wild animals to other planets.

~

Malaria Bed Nets (Against Malaria Foundation)

I like GiveWell, and want to support what they are doing. Where everything else is a "back of the envelope" calculation that seems error-prone, GiveWell is doing robust and thorough work that I have close to zero quality-based concern with. Thus, I'm satisfied to take it on their evidence that $500-$2000 will save a life (provide about 60 additional life-years of moderate happiness) via AMF.

The problem is GiveWell has admitted to not taking nonhuman animal welfare seriously, and thus I cannot trust that AMF is really the best donation target, all entities considered. I'd guess, loosely and uncertainly, that Veg Ads are about 10-1000x more effective.

So really a donation to AMF via GiveWell is going to be mostly a function of weighting based on my certainty (and perhaps irrational risk-aversion) and wanting to "reward' GiveWell and leverage my donations to persuade others. Also, I can sincerely donate to AMF for my GWWC pledge.

~

Meta-Charity (Center for Effective Altruism)

Another recently compelling candidate is funding the expansion of CEA (80K Hours / GWWC / EAA / TLYCS). Will Crouch's back of the envelope calculation estimates very speculatively somewhere between $9 to $51 donated to a "top charity" for every $1 donated to CEA.

My only concern here is basically that this is wrong -- the estimates might be erroneous for some unknown reason, or past growth not being able to adequately track future growth. The causal theory here seems plausible, but weaker than the case made for veg ads -- I'm still not quite sure exactly how future money will generate more members. The plan is kind of generically and vaguely focused on "hire more staff".

I'm attracted to this option, however, because it could be sincerely used toward my GWWC pledge, and if I earmarked the money to 80,000 Hours alone, there's a good chance I could be using my pledge money to funnel more money to veg ads.

Though that gives rise to a second concern -- I have no idea how much of the funneled money actually would go to Veg Ads. But I think I'd only have to achieve a rate of 12% to Veg Ads to break even at a 9x multiplier.

~

Existential Risk Reduction (Singularity Institute)

The Singularity seems moderately plausible as a pressing issue, and the ability to increase the chance of survival for massive quantities of future people and increase our time table on utopias sounds very worthwhile. However, I don't think there even is a back of the envelope calculation here, and definitely no working theory about how the money. Thus, this seems like a good cause in want of a demonstrably effective organization.

I could consider holding my donation until such an organization emerges.

~

Wild Animal Suffering

Wild animal suffering is also an amazingly pressing concern, though it currently strikes me as less so than the Singularity. Additionally, while I know some orgs are in the works, there is no current organization focusing on this issue, let alone a demonstrably effective organization. I could consider holding my donation here as well.

~

So where do you think my pot should go? I'm open but wary of dropping my GWWC membership, so I imagine it could easily turn out that 10% of my income should go somewhere and the remaining 4% of my income should go somewhere else.

Thanks for the help!

Thus, I'm satisfied to take it on their evidence that $500-$2000 will save a life (provide about 60 additional life-years of moderate happiness) via AMF.

I think the main value of saving these lives is that it prevents unpleasant deaths and additional replacement pregnancies? Obviously, you do not provide the resources for 60 years of life for $2000. Due to changes in fertility, there is no linear connection between saving children's lives and total life-years on the planet. This is one of the reasons why I think the common "saved lives" measure for ethical utility is less valuable than people assume.

Regarding vegan outreach, consider this: Even if no one changes their consumption of animal products, the video creates additional awareness of human-caused animal suffering. This probably has a beneficial effect on acceptance of future animal welfare policies and alternative production technologies for current animal products.

Did you consider investing your money and donating later? Or focusing on your career foremost, and the specifics of your altruism later?

Hedonic Treader wrote:I think the main value of saving these lives is that it prevents unpleasant deaths and additional replacement pregnancies? Obviously, you do not provide the resources for 60 years of life for $2000. Due to changes in fertility, there is no linear connection between saving children's lives and total life-years on the planet. This is one of the reasons why I think the common "saved lives" measure for ethical utility is less valuable than people assume.

Thanks for your feedback!

I agree that "saved lives" isn't useful. But a malaria net typically saves someone who would have died at 5 and instead allows them to have a 50% chance of making to age 60, thus having 27.5 expected life-years (27.5 QALY) added (Source). That's what I was referring to.

~

Hedonic Treader wrote:Regarding vegan outreach, consider this: Even if no one changes their consumption of animal products, the video creates additional awareness of human-caused animal suffering. This probably has a beneficial effect on acceptance of future animal welfare policies and alternative production technologies for current animal products.

Probably? I still feel like that's a stretch too far into the future, based on too little to be confident enough to donate. If people find the video unconvincing, then it seems unconvincing wholesale -- I'd be surprised if it didn't work for consumption change but did work for other policies. Then extrapolating that far into the future worries me more, just because so much can happen between then and now.

And again, there's the fact that this is just speculation.

~

Hedonic Treader wrote:Did you consider investing your money and donating later?

Yes. Right now I find it marginally more plausible that social returns of donations outpace investment opportunities, in general. I think that if investment opportunities were better, the non-profit could just invest it themselves anyway, perhaps at a more competitive rate since they could aggregate multiple donations together into an endowment. Though I'd want to see more research on this.

I think this is a moot point anyway, because right now my income is too small to have serious investment opportunities.

~

Hedonic Treader wrote:Or focusing on your career foremost, and the specifics of your altruism later?

My donation amount doesn't affect my opportunities for career. Luckily, I can focus on both!

peterhurford wrote:Existential Risk Reduction (Singularity Institute)

The Singularity seems moderately plausible as a pressing issue, and the ability to increase the chance of survival for massive quantities of future people and increase our time table on utopias sounds very worthwhile. However, I don't think there even is a back of the envelope calculation here, and definitely no working theory about how the money. Thus, this seems like a good cause in want of a demonstrably effective organization.

I could consider holding my donation until such an organization emerges.

Brian Tomasik wrote: I love a lot of things about transhumanism, but I'm concerned that development of artificial general intelligence might lead to astronomical amounts of suffering. Unless we create a benevolent singleton, the future will likely be dominated by Darwinian pressures on AIs, causing the trajectory of civilization to slip out of our control (http://www.nickbostrom.com/fut/evolution.html).

From: http://www.facebook.com/groups/437177563005273/

peterhurford wrote:Wild Animal Suffering

Wild animal suffering is also an amazingly pressing concern, though it currently strikes me as less so than the Singularity. Additionally, while I know some orgs are in the works, there is no current organization focusing on this issue, let alone a demonstrably effective organization. I could consider holding my donation here as well.

Hold it till January and you should be able to donate to an antispeciesist charity who care a lot about WAS !:D!

Ruairi wrote:Hold it till January and you should be able to donate to an antispeciesist charity who care a lot about WAS !:D!

I'm excited to see that organization get off the ground. What activities to you anticipate it doing that would be worth donating toward? I'm worried about another "great cause, but organization of unknown impact" problem (like with the Singularity).

Fb message sent

Anyone else seeking info please contact me or Brian Tomasik

Ruairi, I think for this year I'm going to have to wait for the RWAS group to get more of a track record first. I'd be happy to consider donating to the group sometime in 2013.

I would definitely avoid donating to global poverty because of the poor meat-eater problem. I don't see that maintaining your commitment to GWWC is particularly important. Advocating that people give more to charity is great, but I think it's not clear that your commitment to GWWC helps with that. Plus, GWWC is too focused on global poverty, which I think is a serious issue, especially in light of the poor meat-eater problem.

I'd donate to Vegan Outreach or The Humane League. While they don't have as strong evidence of effectiveness as AMF, they do have pretty good evidence, and they seem to be the most cost-effective charities. They use methods that are fairly well-researched and appear to work.

Edit: I'd also avoid existential risk charities. While averting existential risk is important, no such charity has ever provided even a moderate amount of evidence that it is actually effective (AFAIK).

MTGandP wrote:I would definitely avoid donating to global poverty because of the poor meat-eater problem.

I don't think the "poor meat-eating problem" is devastating. While meat eating does seem to go up, it doesn't seem to go up that much, and economic development also has an effect of reducing the population, which would tend to cancel out the effect. I'm less concerned.

The argument in favor of VO/THL, however, is there increased effectiveness, which I still peg to be 10x-1000x that of AMF.

~

MTGandP wrote:I don't see that maintaining your commitment to GWWC is particularly important. Advocating that people give more to charity is great, but I think it's not clear that your commitment to GWWC helps with that.

Right now, my actual donatable income is small, but I have the ability to make much larger changes through convincing others. I've already convinced at least 1 person to take the pledge who otherwise likely wouldn't have. GWWC also gives me access to connections and blogging that has so far been helpful.

I think I'd only want to drop my membership and commitment if I no longer found it useful for persuading, if VO/THL's evidence based increased, and/or my income increased. If I could estimate with moderate confidence that VO/THL were 100x more effective than AMF, I'd drop my GWWC membership immediately. That's my current threshold.

~

MTGandP wrote:Plus, GWWC is too focused on global poverty

I think that makes it a fairly strong candidate for convincing otherwise, however. Hopefully Effective Animal Activism could take off and convince people. But I can see strong strategic reasons for GWWC to focus on "global poverty" (it's a significantly easier sell and still considerably more effective than "folk" donation targets).

I also think enough people "walking away" from GWWC over speciesism concerns could be effective.

~

MTGandP wrote:Edit: I'd also avoid existential risk charities. While averting existential risk is important, no such charity has ever provided even a moderate amount of evidence that it is actually effective (AFAIK).

Agreed.

Re THL/VO I think you can ask for your donation to be used on fb ads, which are really effective

peterhurford wrote:

MTGandP wrote: I'd also avoid existential risk charities. While averting existential risk is important, no such charity has ever provided even a moderate amount of evidence that it is actually effective (AFAIK).

Agreed.

Didn't you see the article in the latest issue of the Journal of Risk Research? For non-subscribers, here's the abstract:

To date, there has been no experimental evidence that levels of research spending on global catastrophic risks (GCRs) are related to reduced rates of global catastrophic events (defined as the death of 99.9% of the human population). In this pilot study, earthlike worlds (n = 60) were randomly assigned to zero, low (0.05% of GDP), or high (5% of GDP) levels of GCR-research spending. Thirty-three global catastrophic events were observed over 100 years. The frequency of such events was lower in the high-spending group than in the zero-spending group, but this difference was not significant (p = 0.11). Further research is necessary to determine whether GCR research is effective.

Pat wrote:Didn't you see the article in the latest issue of the Journal of Risk Research? For non-subscribers, here's the abstract

I'm not a regular reader of risk journals. That article is really interesting, though I wonder how they arrived at those conclusions / ran that model.

Ruairi wrote:Re THL/VO I think you can ask for your donation to be used on fb ads, which are really effective

If FB ads are the most effective, why don't THL / VO already automatically do that with my donation if I donate to them in an unrestricted fashion?

That was a joke. I should have put a smiley face somewhere. This isn't the first time I've made that mistake.

I'm not sure how an existential-risk charity could provide evidence of effectiveness. Since existential risks by definition are one-off events, you can't measure their frequency. If an asteroid was going to strike earth and we built a rocket that moved the asteroid out of the earth's path, you could say building the rocket reduced existential risk. But how could you know whether research on the safety of biotech, nanotech, or AI makes the difference between human civilization perishing and persisting?

It makes sense to demand evidence of effectiveness from some charities, but there's something strange about demanding it from charities that couldn't even in principle provide it. "Name the times when you saved the world" is (in my opinion) an unreasonable prerequisite for donating to an existential-risk charity. But I might be attacking a straw man.

I don't think demanding evidence from an existential-risk research organization is that tremendously different from demanding evidence from any other research organization, such as a medical one. They just have to show what kind of research they can do at what amount, what they expect that research to do, etc. I'm not saying it will be easy to evaluate, but it's at least not outright impossible.

Below is the current draft of an essay I plan on publishing on my site on December 4th detailing my decisions and thoughts behind my chosen allocation of donations -- $397 to AMF, $267.15 to Vegan Outreach, $113.43 to GiveDirectly, and $56.72 to SCI. The donations haven't been made yet, so I'm giving you experts a chance to sway me.

This also features what I think are adequate cost-effectiveness estimates of Vegan Outreach and meta-charity donations to CEA, which could be worth looking over. I come up with a lower bound for Vegan Outreach based on the parts of the survey data we can be nearly certain about.

---

<Introduction about how much I want to donate omitted, because it duplicates what was stated above.>

<Introduction about giving effectively omitted, because you guys already know what I'd say.>

<Discussion about GiveWell top charities and splitting your donation omitted because it duplicates ideas already talked about over here. It's important to note that earlier I think I quoted a figure for AMF of $1600 per "life saved" (60 years of good life added), where the real calculation should be $2500 per "life saved" (30 years of good life added). Details below in Appendix B.>

~

Vegan Outreach and Saving Nonhuman Animals

While GiveWell is a great and established evaluator, it unfortunately has one drawback -- they place significantly little value on the welfare of nonhuman animals, whereas I personally think that nonhuman animals should be valued equally insofar as they suffer equally. This isn't to say that a human's life is identical to that of a chicken, but that if we have a chance to prevent human suffering or a chance to prevent nonhuman animal suffering of equivalent amount, we should consider these opportunities equally.

The actual interspecies comparison of suffering is obviously incredibly tricky and is going to be fast, loose, and uncertain; though it's better than not considering nonhuman welfare at all. There is another evaluator however called Effective Animal Activism that finds cost-effective organizations working to help nonhuman animals. After looking over hundreds of possibilities, they identified the most cost-effective opportunity is to buy Facebook ads at $0.19 a click that link to [url=whosagainstanimalcruelty.org]shocking videos of factory farming <YouTube>[/url] and persuade people to go vegetarian, and organizations like Vegan Outreach and The Humane League are prepared to do this.

The Survey
Using a back of the envelope calculation <PDF> followed up here and here by Brian Tomasik, this appears incredibly cost-effective. A $5000 campaign, evaluated by survey results <PDF>, resulted in 32 reported vegetarians.

I think this is all we can be relatively certain about. But it understates the case significantly. We can extrapolate the survey results to the non-surveyed population of people who still "liked" the video on Facebook or requested a vegetarian starter kit, we get 1155 vegetarians. We can also add the reported results of friends and family who went vegetarian (which seems surprisingly high), getting to 4828 total vegetarians. And this ignores those who said they reduced their meat consumption, while not going completely vegetarian. Additionally, some people went completely vegan.

The Calculation
Furthermore, Brian Tomasik mentions a survey that showed vegetarians refrain from eating meat for an average of 4.7 years or more. Another study I found says 93% of vegetarians stay vegetarian for at least three years.

Going off of this, we can do a little bit of math: $5000 produced 32 vegetarians (low bound) to 4828 vegetarians (upper bound). Assuming an average of 4.7 vegetarian-years per person, that's 150 years (low bound) to 22692 years (upper bound), or $33/year to $0.22/year. At 7.8 land animals saved per year, that's somewhere between $0.03 to $4 to save an animal from suffering in a factory farm.

~

GivingWhatWeCan and Global Poverty

While the calculations for how much it costs to save animals from suffering are compelling, they're also significantly more uncertain than that of GiveWell's calculations, so I'm uncertain about where I'd rather donate. But there's another "problem" -- as a member of GivingWhatWeCan, I have pledged to "give at least ten percent of what I earn to whichever organizations can most effectively use it to fight poverty in developing countries", and I can't sincerely say that saving animals from factory farming will fight poverty.

Thus, I'm left with a steeper dilemma -- should I give up my GWWC membership in order to allocate all my money to nonhuman animals? Right now, I'm thinking not. As a college student, I don't feel like I make enough money to outweigh the benefits of my membership in helping to convince others to donate more. So I'm tempted to still donate 10% of my income to helping global poverty.

~

My Final Decision

So my final decision is to keep my pledge at this time and donate 10% of my income (currently $567.15) to GiveWell, according to their split ($397 to AMF, $113.43 to GiveDirectly, and $56.72 to SCI). I'll then take all the money I plan on giving in addition to my 10% pledge and donate it to Vegan Outreach ($167.15). I also have money left over from last year that I planned to donate but haven't yet -- since this was before the time I took the pledge, I'll donate it all to Vegan Outreach (+$100 = $267.15).

Given how much more certain GiveWell's calculations are compared to the case for Vegan Outreach, I'm actually not all that currently concerned that my donation is being split. I hope Vegan Outreach can eventually do more research and make their case more clear. I naïvely assume Vegan Outreach to be 83% to 333333% better than donating to AMF (see appendix B for the calculation).

So there you go. If you disagree with my allocation and share my goals, feel free to comment so I can improve my planning next time around, when I'll have more income. If you disagree with my allocation and don't share my goals, feel free to comment anyway, because I'd love to hear what other people are doing.

Regardless of what you plan, I hope you consider some tips for giving and look to do better with cost-effective donation. I'll definitely be revisiting my thoughts on these areas in the future with the goal in mind of doing better than I already have. I'm slightly concerned I'm writing off the huger upsides of some other alternatives (specifically meta-charity; see Appendix A).

~

Appendix A: Other Great Ideas I'm Not Supporting

If you think I should have donated to existential risk, meta-charity, or to prevent wild animal suffering, here is where I quickly explain why I didn't. This is an appendix, so there's no real reason you need to read it.

Existential Risk
Existential risks are risks to humanity's very survival -- there are catastrophic events that could wipe out all of humanity (like supervolcanoes, superasteroids, robots gone wrong, that kind of thing) and there are people working on reducing these risks, like The Future of Humanity Institute (FHI) or The Singularity Institute (SIAI).

I agree that given how large the possible future is, any reduction in the chance of human extinction is likely to have very beneficial consequences for humanity. If we expect billions more humans to be born, reducing existential risk by even 0.1% is the equivalent of saving an expected millions of lives.

However, I have two problems with existential risk:

We don't know how beneficial the future really will be. Is the future "worth saving"? Well only if we expect the future to be a happier place. However, arguments have been made that the total amount of utility in the world right now could be negative (given the amount of suffering especially in factory farms and among wild nonhuman animals). Thus, it's not worth saving the future unless we can ensure that major changes are made to the future as well.

We don't know if FHI or SIAI are doing effective work. Essentially, these organizations seem to be doing good work, but right now there is no information I'm aware of on how much existential risk they're reducing, especially on a per dollar basis -- and while I suspect a very rough quantification of their impact (along the lines of VO's argument) is possible, I don't currently know what it would look like. Just because an issue itself is pressing does not mean that the organization working on it is effective. Indeed, Holden Karnofsky has raised some significant problems with the case for SIAI (SIAI response here).

Existential risk remains on the horizon for me as a possibility should it be demonstrated that an organization is actually making a difference in this area. I know Luke Muelhauser has done some work on how to purchase existential risk reduction, but it's not certain enough yet.

~

Meta Charity
I could also donate to organizations like 80,000 Hours and GivingWhatWeCan directly, since they're non-profits that spend money to convince people to donate more money, effectively. Will Crouch estimates that have "raised $8 in realised donations and $130 in future donations for every $1 worth of volunteer time invested in Giving What We Can". More specifics on how donations would be used are here.

All donations to GWWC are funneled to GiveWell's top charities and among donations to 80K, they are (according to survey data) split 34% to existential risk, 63% to global poverty reduction, and while the remaining 3% is unsaid, I assume it goes to nonhuman animal welfare. This means that a $1 to meta-charity will raise $65 for global poverty, $22.1 for existential risk, and $1.95 for animal welfare -- assuming money is split evenly between GWWC and 80K.

This would be great, I think, but it's still too uncertain for my taste, and I jump away at that:
1.) I worry about assuming member growth will grow linearly, since as more people have found out about it, less of the people who could be convinced remain in the population.
2.) I worry about not knowing how much money really goes to animal welfare, and just having to guess.
3.) I worry that while $65 is being raised for global poverty, only about 57% of it is actually going to a GiveWell top charity, where the rest are going other places.
4.) I worry that by donating through GiveWell, I already am helping to encourage more people to give more effectively. I think more benefit comes from increasing the effectiveness of one's donation rather than the size of one's donation.

I think it would be best for me to sit back and watch the growth of GWWC a bit longer. If it fits their predictions, I'd be happy to consider giving them my future donations. I also watch with keen interest the growth of the associated organization, Effective Animal Activism, because I think it has the potential to make a large impact in helping nonhuman animals. However, it's not currently accepting donations.

I'd appreciate people's thoughts on this. For more evidence on GWWC's case for meta-charity, you can also email William Crouch. I don't want to list his email here publicly without his permission, but you can find it via Google or the essays of his I linked to.

~

Appendix B: Comparison of AMF and Vegan Outreach

GiveWell can add an average of 30 years a life with $2500 in bed nets. Thus, AMF can add one year of healthy human life for $83.33.

Nonhuman animals in factory farms suffer an average of 1 year and can be saved for $0.03 to $4. Thus, VO can prevent one year of animal suffering for $0.03 to $4.

Counting a year of healthy human life and animal suffering equally (which is perhaps not appropriate and could use fine-tuning), VO can save 20.83x to 83333x more lives per dollar.

If I feel like GiveWell's evidence is 25x better (a complete guess), that makes VO a total of 83% to 333333% better.

~

Why do I feel like GiveWell's evidence feels 25x better? Well, the case for AMF is based on a significantly large amount of solid academic trials on malaria nets and in-depth studies on AMF in particular, none of which has been done for Vegan Outreach. The data for VO, while existent, is very thin. I would want to see a few more intentional replications, perhaps making use of more robust though complex methods. My other concerns with Vegan Outreach are thus:

1.) I'm worried about to what extent the video truly causes vegetarianism, or if people would have gone vegetarian for other reasons absent the video eventually, or if people who saw the video were already vegetarian.

2.) I'm worried that while the "how long do people stick with it" data is there and looks reliable, it isn't specifically related to the video. I have a weak suspicion that those converted by a video alone are less likely to stick with it. And I'd want this data to be replicated more.

I'd appreciate people's thoughts here too.

Some more thoughts in favour of marginal support to metacharity:

It increases the expectation of everyone's donations by making them more able to find and evaluate efficient causes faster. I suspect there's still huge gains to be made given that so far, almost all effectiveness research has gone into one small sector.

Conditional on there being exponential gain (even for a relatively short time) from increasing the EA population, donating to them now has a *huge* compounding benefit. So 'I worry about them growing linearly' seems like a weak reason not to support them, since it's asymmetric - even if there were only a 1% chance of exponential growth, it might (IMO probably would) be worth supporting them.

The fact that only 57% of member donations go to global poverty doesn't really tell you much in itself - you need to assess your expectation of the other causes. Assuming you think they're worse than global poverty, it matters a lot how much worse. If it's only fractional, then you can probably ignore the loss, since other factors will dominate it.

One thought about your proposed giving structure - splitting your donations does not seem justified by the reasoning you've provided above. Ok, there's some logic to Givewell recommending people split between their charities since it avoids a surplus going to AMF (though I still strongly disagree with it), but your estimation seems to be that VO has better expectation. There's almost 0 risk of VO getting saturated for the time being, so none of the factors that would suggest a split for Givewell-directed donors seem to apply to you.

I don't think demanding evidence from an existential-risk research organization is that tremendously different from demanding evidence from any other research organization, such as a medical one. They just have to show what kind of research they can do at what amount, what they expect that research to do, etc. I'm not saying it will be easy to evaluate, but it's at least not outright impossible.

We might agree on this. It's important to avoid a catch-22 in demanding evidence from existential-risk organizations (Charity: "Give us some money so we can save the world!" Donor: "I demand rigorous evidence of effectiveness, including a strong track record. So first save the world, then we can talk.") But it's also important that such charities have clear goals (some of them quantifiable) and that they be held accountable. There's also the issue of whether the future will be good or bad. So I don't think there's a clear case for donating to existential-risk charities.

Essentially, these organizations seem to be doing good work, but right now there is no information I'm aware of on how much existential risk they're reducing, especially on a per dollar basis -- and while I suspect a very rough quantification of their impact (along the lines of VO's argument) is possible, I don't currently know what it would look like.

This video contains a "very rough quantification." I think somebody on this forum (Arepo?) criticized the assumptions, but I can't find that thread. Also, I watched it a long time ago, I don't remember it well, and I've run into my ISP's bandwidth cap, so I can't verify that it's relevant.

peterhurford wrote:

Ruairi wrote:Re THL/VO I think you can ask for your donation to be used on fb ads, which are really effective

If FB ads are the most effective, why don't THL / VO already automatically do that with my donation if I donate to them in an unrestricted fashion?

I dunno but maybe Brian Tomasik can tell you

EAA are accepting donations! !

peterhurford wrote:Brian Tomasik mentions a survey that showed vegetarians refrain from eating meat for an average of 4.7 years or more. Another study I found says 93% of vegetarians stay vegetarian for at least three years.

Great findings! The first study doesn't explicitly give a figure of 4.7 years. I believe Brian reached that number by computing it from Table 1, which lists the number of people refraining from eating animal products for various lengths of time. I created a spreadsheet with this data. On my calculations, former limiters stayed vegan or vegetarian for an average of five and a half years, before reverting to an omnivorous diet. As I note in one of the comments on that spreadsheet, this calculation is sensitive to the mean assumed duration for the class interval defined as '10 years or more'; I assume a mean of 15 years. (To see the spreadsheet comments, you need to download the file and open it on MS Excel, as opposed to browsing it from Google Drive.)

Incidentally, I think answering the question 'How long do people stay vegetarian, on average?' raises some interesting methodological problems. For example, you cannot just poll current vegetarians and ask them how long they've been vegetarian for, since long-term vegetarians will be overrepresented in the sample. (A vegetarian who has been vegetarian for one year is twelve times more likely to appear in your sample than a vegetarian who has been vegetarian for one month.) Moreover, the time current vegetarians have been vegetarian for obviously does not equal the total time they will stay vegetarian. The Copernican principle would suggest that, to reach the later figure, one would need to multiply their answers by two. (Although this number would still need to be updated in light of actuarial knowledge: clearly a 70-year-old who has been vegetarian for several decades cannot be expected to live, and hence stay vegetarian, for more than 10 extra years or so.) I suppose that this is a standard sampling problem in statistics, and so I image there is an established methodology for dealing with it. I would be very interested to hear comments from someone familiar with these issues.

peterhurford wrote:I don't think the "poor meat-eating problem" is devastating. While meat eating does seem to go up, it doesn't seem to go up that much, and economic development also has an effect of reducing the population, which would tend to cancel out the effect. I'm less concerned.

The argument in favor of VO/THL, however, is there increased effectiveness, which I still peg to be 10x-1000x that of AMF.

It doesn't look like the reduced population growth cancels out the effect of increased meat consumption. At present, developed nations with lower birth rates have much higher meat consumption (especially from factory farms) than developing nations with higher birth rates.

Do you have any figures on how much meat-eating is increased by the work of GiveWell's top charities? I would guess that AMF has a fairly small effect; SCI probably has a bigger effect because it allows people to stay in school for longer, leading to future economic development; and GiveDirectly probably has a significant effect as well.

After chatting with Brian Tomasik and reconsidering the evidence for both veg ads and meta-charity, I've decided to change my mind and allocate everything ($839.40) to Effective Animal Activism, unless something else comes up to change my mind within the span of two weeks or so.

Your decision is commendable, Peter, both because EAA is very likely a more cost-effective organization than the ones you were going to support before you changed your mind, and because very few people do actually change their minds in light of new evidence.

You legend!:D!

peterhurford wrote:After chatting with Brian Tomasik and reconsidering the evidence for both veg ads and meta-charity, I've decided to change my mind and allocate everything ($839.40) to Effective Animal Activism, unless something else comes up to change my mind within the span of two weeks or so.

What was it that changed your mind?

Also, are you donating directly to EAA, or to EAA's top charities? If the former, does EAA have significant room for more funding?

Peter's praiseworthy post is now published.

MTGandP wrote:Also, are you donating directly to EAA, or to EAA's top charities? If the former, does EAA have significant room for more funding?

He's donating to EAA directly

I'd imagine they do and i encourage you to contact them if you're interested !

Also would you be interested in supporting a charity which will have a reasonably large focus on reducing wild animal suffering? If so please contact me or Brian Tomasik !

Michael Dickens wrote:What was it that changed your mind?

I decided I was more certain in my $4/animal low-bound estimation than I was initially giving it credit for. Also, Brian informed me that it is possible to donate to EAA and keep one's GWWC pledge.

~

Michael Dickens wrote:Also, are you donating directly to EAA, or to EAA's top charities? If the former, does EAA have significant room for more funding?

I'm donating directly to EAA. EAA has significant possibilities for using additional funding in the fact that they currently are doing very little and desperately need actual staff. However, I was careful not to say that they have "room for more funding" because Brian has told me that he and Pablo intend to fund all of EAA's needs themselves. Thus, in donating to EAA, I am really demonstrating support for the organization and displacing money from Brian and Pablo that they then will use to allocate to other utilitarian activities.

So really, I'm trusting that the value of showing my support via donation, plus shifting money to EAA sooner rather than later, and the value of what Brian and Pablo will do with the displaced cost will in total outweigh the value of donating to Vegan Outreach. Overall, I'm confident that I will be generating more than $1 in VO-equivalent goodness than $1 directly to VO.

~

Pablo Stafforini wrote:Incidentally, I think answering the question 'How long do people stay vegetarian, on average?' raises some interesting methodological problems. For example, you cannot just poll current vegetarians and ask them how long they've been vegetarian for, since long-term vegetarians will be overrepresented in the sample.

Generally, the methodological solution to this is to poll a bunch of people, say a representative sample of 2000. See how many of them are vegetarian, and ask how long they've been vegetarian. Then, a few months or years later, poll that exact same group of 2000. Because of attrition effects, you may not be able to re-contact all of them, but say you have 1800 left over. See what percentage of them are still vegetarian or have become vegetarian (or became vegetarian and then dropped it), and ask how long. That should get you fairly reliable information.

At minimum, you can see how people's vegetarian behaviors change over the time *between* studies and that will be immune to the sample effect. You can also see how likely it is for someone to become vegetarian during a timeframe. Furthermore, by asking people if they've ever been vegetarian and how long they were for, you can defeat even more of the sampling effect.

Do you think that works?

Great reply, Peter!

peterhurford wrote:because Brian has told me that he and Pablo intend to fund all of EAA's needs themselves.

It's worth noting that I'm only guaranteeing to fund them for this year; it's unclear for next year and beyond. The hope is that the new Executive Director will be able to do some fundraising of his/her own so that by next year, EAA will need less handholding from the angel investors. We'll see how long it takes to get funding from outsiders.

It's also worth noting that while EAA is unlikely to hire a second employee this year, other costs are still variable, like how much is spent on booklets, payments for outside help, online ads (if any), etc.

As far as the sampling methodology, I'm just kind of curious: Would it literally work to ask veg*ans "For how long have you been veg*an?" and then multiply by 2, ignoring the fact that people die? It works mathematically, at least when we assume that the average length for staying veg*an isn't changing over time, etc.

I haven't exactly thought this through, but could you have a control group of omnivores, who you ask 'how long have you been omnivorous'?

Brian Tomasik wrote:It's also worth noting that while EAA is unlikely to hire a second employee this year, other costs are still variable, like how much is spent on booklets, payments for outside help, online ads (if any), etc.

EAA is spending on booklets and online ads? What's that about?

~

Brian Tomasik wrote:As far as the sampling methodology, I'm just kind of curious: Would it literally work to ask veg*ans "For how long have you been veg*an?" and then multiply by 2, ignoring the fact that people die? It works mathematically, at least when we assume that the average length for staying veg*an isn't changing over time, etc.

This method strikes me intuitively as suspicious -- I don't think you can make the assumption that the vegetarians you encounter will all average out to be in the middle of their vegetarianism period.

~

Arepo wrote:I haven't exactly thought this through, but could you have a control group of omnivores, who you ask 'how long have you been omnivorous'?

By asking all people if they have ever been a vegetarian, and if so, how long they were a vegetarian for, this piece of data is included. I must also note that it's best to ask if people have self-identified as a vegetarian and also ask if they've ever stopped eating meat, because not everyone has a consistent definition of vegetarian, and some people don't eat meat but don't choose to self-identify as vegetarian.

peterhurford wrote:EAA is spending on booklets and online ads? What's that about?

By "booklets" I just meant a printed version of EAA's findings for handing out at conferences, etc. It probably wouldn't be that expensive, but it might cost a few hundred dollars or whatever to find a good designer (unless we could enlist a volunteer) and to do the printing.

I don't know if EAA would venture into ads or not, but it's possible we could try them out to increase traffic to our site, Likes on our page, or donations to our recommended charities. I don't know if this is (or should be) on the agenda to pursue, but if we did do it, it would just be experimental at the beginning.

peterhurford wrote:This method strikes me intuitively as suspicious -- I don't think you can make the assumption that the vegetarians you encounter will all average out to be in the middle of their vegetarianism period.

If you actively sample them without bias, they will be. Whether you can actually do that is another issue, but in theory, if the sample were random and you didn't have response-bias problems, it should work.

Brian Tomasik wrote:By "booklets" I just meant a printed version of EAA's findings for handing out at conferences, etc. It probably wouldn't be that expensive, but it might cost a few hundred dollars or whatever to find a good designer (unless we could enlist a volunteer) and to do the printing.

There seem to be untapped markets among those who are current vegetarians but not yet effective philanthropists, or those who are current smart philanthropists but unaware of EAA's findings. I think trying to get a person to become both a vegetarian and an effective philanthropist in one go may be mutually interfering and difficult.

Brian Tomasik wrote:I don't know if EAA would venture into ads or not, but it's possible we could try them out to increase traffic to our site, Likes on our page, or donations to our recommended charities. I don't know if this is (or should be) on the agenda to pursue, but if we did do it, it would just be experimental at the beginning.

Could be interesting to do a trial run if you can track increases in donations as a result of the ad, but that seems like it would be really hard. There are other conditions where a trial run would be important too.

Brian Tomasik wrote:

peterhurford wrote:This method strikes me intuitively as suspicious -- I don't think you can make the assumption that the vegetarians you encounter will all average out to be in the middle of their vegetarianism period.

If you actively sample them without bias, they will be. Whether you can actually do that is another issue, but in theory, if the sample were random and you didn't have response-bias problems, it should work.

Wouldn't vegetarian years have to be normally distributed for this to work? I can't say I really understand what is going on with this trick, one way or the other. Obviously there are some confoundings -- like massive distribution of in-vitro meat next year, for example.

peterhurford wrote:There seem to be untapped markets among those who are current vegetarians but not yet effective philanthropists, or those who are current smart philanthropists but unaware of EAA's findings.

The first target is current animal philanthropists (both small and large) who donate to suboptimal animal charities (e.g., helping pets, opposing zoos, etc.).

peterhurford wrote:Could be interesting to do a trial run if you can track increases in donations as a result of the ad, but that seems like it would be really hard.

Yeah, there might not be enough data points for it to be meaningful. These things are pretty discrete. A better way might be to ask members/donors how they joined and see what fraction joined due to the ads.

peterhurford wrote:Wouldn't vegetarian years have to be normally distributed for this to work?

Don't think so. Maybe I'm crazy, but I'm just picturing a bunch of horizontal lines of some length, stacked on top of each other. Each line has some random length (not necessarily normal), but as long as the probability distribution from which the lengths are drawn is constant with respect to the temporal axis, then for any vertical slice, the point where the slice intersects the horizontal lines will be uniformly distributed between 0 years and the max years being veg.

Say the actual length is L years, so the probability distribution function of seeing any particular year is 1/L. The observed duration being veg is v, and so you guess that they'll be veg a total of 2v years. Integral from 0 to L of (2v) * (1/L) * dv = (L^2 - 0^2) * (1/L) = L, so the estimator 2v is unbiased, i.e., over enough repetitions, if you guess 2v, your average will converge to the true average.

Haha, this explanation is way more complicated than it needs to be, and I didn't do any integrals when I first proposed this -- I just visualized it in my head -- but it's nice to see that it works formally.