Larks
Karma: 1,807
NewTop
Thanks for writing this, I thought it was quite a good summary. However, I would like to push back on two things.
Effective altruism is egalitarian. Effective altruism values all people equally
I often think of age as being one dimension that egalitarians think should not influence how important someone is. However, despite GiveWell being one of the archetypal EA organisations (along with GWWC/CEA), they do not do this. Rather, they value middle-aged years of life more highly than baby years or life or old people years of life. See for example this page here. Perhaps EA should be egalitarian, but de facto it does not seem to be.
Effective altruism is secular. It does not recommend charities that most effectively get people into Heaven …
This item seem rather different from the other items on the list. Most of the others seem like rational positions for virtually anyone to hold. However, if you were religious, this tennant seems very irrational—helping people get into heaven would be the most effective thing you could do! Putting this here seems akin to saying that AMF is an EA value; rather, these are conclusions, not premises.
Additionally, there is some evidence that promoting religion might be beneficial even on strictly material grounds. Have you seen the recent pre-registered RCT on protestant evangelism?
To test the causal impact of religiosity, we conducted a randomized evaluation of an evangelical Protestant Christian values and theology education program that consisted of 15 weekly half-hour sessions. We analyze outcomes for 6,276 ultra-poor Filipino households six months after the program ended. We find significant increases in religiosity and income, no significant changes in total labor supply, assets, consumption, food security, or life satisfaction, and a significant decrease in perceived relative economic status. Exploratory analysis suggests the program may have improved hygienic practices and increased household discord, and that the income treatment effect may operate through increasing grit.
https://www.nber.org/papers/w24278.pdf
I don’t have a strong view on whether or not this is actually a good thing to do, let alone the best thing. RCTs provide high quality causal evidence, but even then most interventions do not work very well, and I’m not an expert on the impact of evangelism. But it seems strange to assume from very beginning that it is not something EAs would ever be interested in.
Congratulations guys, this is really impressive. Thanks for all the work you put into this.
My general model is that charities get funding in two waves:
1) December
2) The rest of the year
As such, if I ask groups for their runway at the beginning of 1), and they say they have 12 months, that basically means that even if they failed to raise any money at all in the following 1) and 2) they would still survive until next December, at which point they could be bailed out.
However, I now think this is rather unfair, as in some sense I’m playing donor-of-last-resort with other December donors. So yes, I think 18 months may be a more reasonable threshold.
No principled reason, other than that this is not really my field, and I ran out of time, especially for work produced outside donate-able organizations. Sorry!
It’s also worth noting that I believe the new managers do not have access to large pots of discretionary funding (easier to deploy than EA Funds) that they can use to fund opportunities that they find.
Good point!
I’m glad you found it helpful!
I don’t have a great system. I combined a few things:
1) Organisations’ websites
2) Backtracking from citations in papers, especially those published very recently
3) Author’s own websites for some key authors
4) ‘cited by’ in Google scholar for key papers, like Concrete Problems
5) Asking organisations what else I should read—many do not have up to date websites.
6) Randomly coming accross things on facebook, twitter, etc.
7) Rohin’s excelent newsletter.
Great post, thanks for collecting all these in one place.
According to this article on the pledge:
While the Pledge was originally focused on global poverty, since 2014 it has been cause-neutral. Members commit to donate to the organizations they believe are most effective at improving the lives of others.
Specifically, originally the pledge did not include animal welfare groups, but was later ‘amended’ to include them. Is there a principled reason to include animal welfare, but not religious outreach? They seem quite similar:
1) Both ingroups have (by their lights) strong reasons to think what they are doing is literally the most important thing in the world.
2) Many/most people agree with premises that logically imply the importance of both causes (i.e. many people are religious and believe in heaven, and many people believe animal cruelty is bad)
3) Both causes are seen as somewhat wierd by most people, despite 2)
4) Both causes are quite far from the original stated and de facto goals of GWWC, namely helping people in the third world.
Overall I think this is a great article. It seems like it could be one of the best pieces for introducing new people to the subject.
People sometimes try to gauge the overall views of an author by the relative amounts of page-space they dedicate to different topics, which is bad if you generally agree with something, but want to make a detailed objection to a minor point. I think Kelsey’s article is good, and don’t want the below to detract from this.
To try to help with this effect, I have deliberately made the top three paragraphs, where I explain that this article is very good before coming to the main point of the comment.
However, I do object to this section:
When you train a computer system to predict which convicted felons will reoffend, you’re using inputs from a criminal justice system biased against black people and low-income people — and so its outputs will likely be biased against black and low-income people too.
The text links to another Vox article, which ultimately linked to this ProPublica article, which argues that a specific reoffending-prediction system was bad because:
The formula was particularly likely to falsely flag black defendants as future criminals, wrongly labeling them this way at almost twice the rate as white defendants.
Separately it notes
When a full range of crimes were taken into account — including misdemeanors such as driving with an expired license — the algorithm was somewhat more accurate than a coin flip. Of those deemed likely to re-offend, 61 percent were arrested for any subsequent crimes within two years.
At this point, alarm bells should be ringing in your head. “More accurate than a coin flip” is not the correct way to analyze the accuracy of a binary test for an outcome unless the actual distribution is also 50:50! If fewer than 50% of people re-offend, a coin flip will get less than 50% right on those is classifies as high risk. Using the coin flip analogy is a rhetorical sleight of hand to make readers adopt the wrong analytical framework, and make the test look significantly worse than it actually is.
Now we’ve seen that the ProPublica authors perhaps cannot be entirely trusted to represent the data accurately, lets go back to the headline statement: that the false positive rate is higher for blacks than whites.
This is true, but in a trivial sense.
Blacks commit more crime than whites. This is true regardless of whether you look at arrest data, conviction data, or victimization surveys. (Even if you only asked Black people who committed crimes against them, this result still holds. Also holds true just looking at recidivism.) As a result of this base rate, any unbiased algorithm will have more false positives for blacks, even if it is equally accurate for both races at any given level of risk.
Here are some simple numbers, lifted from Chris’s excellent presentation on the subject, to illustrate this point:
Simplified numbers: High risk == 60% chance of recidivism, low risk = 20%.
Black people: 60% labelled high risk * 40% chance of no recidivism = 24% chance of “labelled high risk, didn’t recidivate”.
White people: 30% labelled high risk * 40% chance of no recidivism = 12% chance of “labelled high risk, didn’t recidivate”.
It is a trivial statistical fact that any decent statistical test will have a higher false positive rate for subgroups with higher incidence. To avoid this, you’d have to adopt a test which included a specific “if white, increase risk” factor, and you would end up releasing more people who would reoffend, and keeping in jail people who would not. None of these seem like acceptable consequences.
Strangely however, neither the Vox article that this one linked to, nor the original ProPublica piece, mentioned this fact—I suspect due to the same political bias kbog discussed recently. There are good reasons to be concerned about the application of algorithms in areas like these. But damning the algorithms as racist for statistically misleading reasons, without explaining to readers the underlying reasons for these statistics, suggests that the authors have either failed to understand the data, or are actively trying to mislead their readers. I would recommend against linking to either article in future as evidence for the claims.
EDIT: Washington Post had a very good article explaining this also.
“at least 35% of women worldwide have experienced some form of physical or sexual violence.”
The article uses this statistic to try to motivate why we might be interested in charities that focus specifically on women. However, we cannot evaluate this statistic in isolation: to draw this conclusion we need to compare against assault rates for men.
I wasn’t able to immediately find a comparable stat for men—the source for the stat appears to be a women-specific WHO report—but I was able to find homicide data. This data is often regarded as especially reliable, because there are fewer issues about underreporting when there is a dead body. (I apologize in advance if the authors did in fact compare assault rates between sexes and just omitted this from the report).
So what does the data say? According to the UN Office on Drugs and Crime, men are dramatically more likely to be victims of homicide in virtually every country. Almost 80% of global homicide victims are male. And the small number of countries where this is not the case tend to be in the developed world, which is not where the charities in this post focus, or very small countries where I suspect there was only one homicide that year.
So a neutral observer would conclude this was a reason to support charities that reduced violence against men, not women, if one were inclined to choose one or the other.
The fact that this article does not seem to even investigate this makes me sceptical of the quality of the rest of the work. If EAs are going to write non-cause-neutral reports, we should at least be clear at the very beginning of the report that other causes are likely to be be better—rather than than presenting misleading evidence to the contrary. Otherwise we are in danger of sacrificing a very important part of what makes EA distinctive.
Sure, that’s why I criticized Vox, not the individual author. I suspect the author did not complain about the title though.
When Vox launched I was very excited, as I thought it would be a good source of high-quality journalism, even if they did censor authors for having the wrong conclusions. However, it seems like virtually every article, even when otherwise high quality, contains some unrelated and unnecessary jibe at conservatives—an unusually direct example of Politics is the Mindkiller. Perhaps this lead to their being in something of an echo chamber, where conservatives stopped reading?
Here’s a recent example, to help make the above more concrete:
1) Trump signed a good law this week. Yes, really. - why does this need the snark in the title? The meaning would have been clearer, and less insulting, if they had just written “Trump signed a good law about HIV this week.”
I worry about this in general with Future Perfect. This behaviour is not something the EA movement wants, but if Future Perfect ends up producing a very large volume of ‘EA’ articles, we risk getting tarnished by association.
Thanks, I thought this article was very thoughtful.
I have one quick question about the examples you mention. While I agree that pro-life examples are a great idea, I’m not sure what you are getting at with the heaven-infinite-value example. Is the problem that people have been using this as a reductio?
2018 AI Alignment Literature Review and Charity Comparison
In response to the title question: yes.
It seems to me that TRIA is really stretching the definition of ‘equality’. Could I not equally suggest a Citizenship-Relative-Interest-Account? This would fit well with people’s nationalistic intuitions. Indeed, if we look at the list of things GWWC claimed EAs do not discriminate based on, we could circumvent all of them with cunningly crafted X-Relative-Interest-Accounts.
I agree a moral discontinuity would be very perverse. But it seems there are many better options. For example, a totalist view—that people matter even before they are conceived—avoids this issue, and doesn’t suffer from the various inconsistencies that person-affecting views do. Alternatively, if you thought that we should not value people who don’t exist in any way, conception provides a clear discontinuity in many ways, such that it does not seem like it would be weird if there was a moral value discontinuity there also.
But I think the biggest problem is that, even if you accept TRIA, I suspect that most people’s moral intuitions would produce a very different weighting distribution. Specifically, they would be more averse to causing pain to 5 year olds than adults—especially adult men. If I have time I might look into whether there has been any empirical research on the subject; it could be a useful project.
Thanks for writing this very detailed analysis. I especially enjoyed the arguments for why we can compare LS scores between people, like the Canadian immigrant study.
The section I found most suprising was the part on Givewell using the Time-Relative Interest Account. I’ve always thought of some kind of egalitarianism as being relatively important to EA—the idea that all people are in some sense equally deserving of happiness/welfair/good outcomes. We might save a young person over an old person, but this is only because by doing this we’re counterfactually saving more life-years.
For example, here is Giving What We Can:
People[2] are equal — everyone has an equal claim to being happy, healthy, fulfilled and free, whatever their circumstances. All people matter, wherever they live, however rich they are, and whatever their ethnicity, age, gender, ability, religious views, etc. [emphasis added]
But the TRIA explicitly goes against this. It directly weighs a year of health for a 25 year old as being inherently more valuable than a year of health for a 5 year old—or a 50 year old. This seems very perverse. Is it really acceptable to cause a large amount of pain to a child, in order to prevent a smaller amount of pain for an adult? I think the majority of people would not agree with this—if anything people prefer to prioritize the suffering of children over that of adults.
Thanks for writing this. This sort of evaluation, which has the potential to radically change the consensus view on a charity, seems significantly under-supplied in our community, even though individual instances are tractable for a lone individual to produce. It’s also obviously good timing at the start of the giving season.
I think the post would be improved without the section on contraception, however. There are many simple environmental interventions we could benchmark against instead, that don’t involve population ethics. Preventing a future human from being born has many impacts—they will probably have a job, some probability of inventing a new discovery, and most importantly they will probably be grateful to be alive—of which emitting some CO2 is likely to be one of the smaller impacts. Any evaluation of contraception that only looks at direct environmental impact is going to be so lacking that I suspect you’d be better off choosing a different intervention to compare to.
Thanks, this is a cool idea.
Inger from Norway wants to support the Good Food Institute (GFI) with a donation of 5000 USD. Robert from the USA wants to support the Against Malaria Foundation (AMF) with a donation of 5000 USD. AMF is tax deductible in both countries, GFI is only tax deductible in the USA. The EA donation swap system introduces Robert and Inger together and they agree to swap donations.
Inger donates 5000 USD to AMF, Robert donates 5000 USD to GFI. They both get their tax deductions at the end of the financial year.
In this example Inger gains tax deductability, but Robert gains nothing in return for taking on the counterparty risk of the swap. Wouldn’t it make sense for Robert to donate slightly less than $5000, or Inger slightly more, such that both parties benefit?
This reminds me a little bit of Critch’s Rent Division Calculator, which aims to produce a room-and-rent-allocation for shared houses that everyone likes as much as possible; not merely one that no-one actively dislikes.
You might enjoy this post Claire wrote: Ethical Offsetting is Antithetical to EA.