Proposal: Impact List—like the Forbes List except for impact via donations

Summary

I’m seeking feedback on and collaborators for the creation, popularization, and maintenance of a list which ranks the top N living people by their positive impact via donations. The goal is to make the list popular enough to increase the status awarded to those who rank highly, bring more awareness to the importance of donation effectiveness, and ultimately cause people to donate more effectively and/​or donate more money.

Details of the list

I still have a lot of uncertainty about how the list should work, but this is my current view:

  • People are ranked by the total lifetime amount of money donated to altruistic causes, scaled by effectiveness as determined by Impact List researchers (via a mix of independent research and repurposing research from GiveWell and other organizations).

  • The first version might contain somewhere between 100 and 1,000 people.

  • Viewers can specify their values which are taken into account by the rankings: how much weight to give animals, people who will exist in the far future, etc. The default settings would attempt to capture the values of the median viewer.

  • The following columns exist, which the viewer can rank by:

    • The raw (unscaled) total amount donated (which can be changed to the amount in a specific year)

    • Amount pledged

    • Net worth

  • The list can be filtered by cause area.

  • There’s a drop-down to view alternate rankings provided by other organizations[1].

    • Any organization submitting a new ranking metric must also submit the reasoning /​ calculations behind the scaling factors they assign to donation recipients.

  • There’s a page for each person showing every donation they’ve made, when it was made, and the scaled effectiveness of the donation according to Impact List and other organizations.

  • There’s a page for each donation recipient showing the research and calculations for why it has the effectiveness rating that it does, according to Impact List and any other organization that has provided their own research.

Causal mechanisms

If Impact List becomes popular it could influence people’s donations in the following ways:

  1. The list focuses attention on large (and potentially surprising to many viewers) differences in donation effectiveness, and on how much positive impact others are having. This may lead people to want to have more impact for its own sake.

  2. Attention on the list increases the status reward for giving effectively as well as the status penalty for giving ineffectively.

  3. An explicit ranking might make some donors want to rank higher than others for competitive reasons.

  4. The website could serve as an on-ramp for future EAs, with links to further reading. Billionaires may occasionally visit the website to check their ranking and the reasoning behind it, which could help us acquire more Sam Bankman-Frieds or Dustin Moskovitzs.

Note that reasons #1 and (the first part of) #4 could apply roughly equally to anyone who is aware of the list, not just those wealthy enough to potentially appear on it.

A large continuous auction

Because so many existing donations by the world’s wealthiest people are much less effective than they could be, spots on the list will initially be available for relatively low donation amounts to those who donate effectively. A list of size 1,000 might initially present an opportunity to hundreds of thousands of people to potentially appear on it (and push an ineffective donor off). This means that all four reasons above could be relevant for many more people than the size of the list.

To encourage the donations of the list members to continuously become more effective the UI could have a tool which shows people who aren’t on the list how much money they’d need to donate to the current most effective recipient to claim a spot.

Popular lists like the Forbes Billionaire List, the US News College Rankings list, and the New York Times Best Seller list influence the status and behavior of those who appear (or want to appear) on them, and the behavior of those who view them.

This is well documented in the case of US News’ list. Colleges regularly try to boost their US News ranking via strategies that have nothing to do with improving the services they offer. US News has a near-monopoly on attention paid to college rankings, which has a huge effect on US higher education. In this case the effect seems net negative, but it’s a great example of the potential power of lists.

One takeaway is that it may be worthwhile to put an enormous amount of resources into making Impact List so popular that almost everyone is vaguely aware of the rankings and it dominates the attention paid to all philanthropy-related lists.

Expected value of solving this problem

Scale

Total donations in 2020 by US entities amounted to 471 billion dollars -- 324 billion by individuals, 88 billion by foundations, 41 billion by estates, and 17 billion by corporations. The top 50 individual donors in the US gave 33 billion dollars in 2021. The top ~1.2 million households in the US donate 13 of all individual donations, or about 108 billion dollars.

It’s difficult to get good global donation data, but let’s estimate global donations are twice the US amount, or 650 billion dollars. Assume the list has 1,000 slots. Given the above data we’ll estimate that the top 1,000 global individuals donate 50 billion dollars per year.

It’s unclear how much the list could realistically influence these donations, but the numbers are so large that even a small change would have a big impact. Shifting 1% of total individual donations into effective causes amounts to 6.5 billion dollars per year. Causing the top 1,000 most impactful donors to direct 10% of the currently donated amount to effective causes comes out to 5 billion dollars per year.

Neglectedness

A common response I get when telling EAs about Impact List is surprise that no one has done it yet. The general idea was suggested on Twitter by Kelsey Piper here and here and Nathan Young has since mentioned it a couple times in forum comments, but I don’t think anyone has started significant work on it.

If this is as neglected as it appears, it raises the question of why. My best guess is that non-EAs find the prospect of rating the effectiveness of a large number of donation recipients in a defensible way to be too daunting. Another possibility is that there are flaws in the idea that I haven’t discovered or am underestimating the magnitude of. It’s also possible that people have tried and failed but not left much evidence of their attempts. (Please contact me if you’re aware of any!)

Although I don’t know of attempts to rank people by donation impact, ranking people based on their donation amounts is common. See the Million Dollar List’s top US donors between 2000-2016, Business Insider’s list of the top 20 lifetime donors, Wikipedia’s list of the top 21 lifetime donors, The Forbes 400′s list of the top 400 richest Americans in 2021 where they give each a 1-5 philanthropy score based on the percent of their net worth that they’ve donated, Philanthropy.com’s list of the top 50 American donors of 2021, and Donation List Website’s ranked list of 55 EA donors, among many others. Note that Forbes’ philanthropy score is a measure of the virtue of the giver, not of donation effectiveness.

Aside from not attempting to rate donation impact, there are many features of these lists that prevent them from becoming influential:

  • The global lists rank only ~20 people.

  • The US lists show donations only over a limited time period.

  • Philanthropy.com requires registering for an account to see their list (which disqualifies it from ever becoming popular).

  • Aside from Wikipedia’s list, they’re snapshots rather than continuously updated.

  • Their UIs are either below average or not built for mass appeal, aside from Forbes’.

I haven’t found any examples where a lot of effort has gone into creating a popular Philanthropy-focused ranking of people. The regular Forbes List does reflect a lot of effort, but the Forbes 400 list is just a version of the main list restricted to US people and with a new low-effort philanthropy score.

TL;DR: This seems extremely neglected. I’m not aware of any effort going into producing a ranking that takes into account donation effectiveness. Efforts that ignore effectiveness don’t appear to be trying hard to become influential, so they don’t give us much information about Impact List’s chances of success.

Tractability

In general getting society to care about a new thing to the degree that people care about the Forbes List or the US News College Rankings seems extremely difficult. There are only so many things that can be that popular because attention is finite and the competition for attention is fierce.

If we can create a high quality list there are a couple factors which might help with making it popular:

  • The EA movement as a whole is increasing in influence. If prominent EA organizations and individuals think this project is worthwhile they may be willing to use their influence to help the list get traction.

    • Impact List might also partner with a popular EA-friendly organization like Our World In Data or The Economist.

  • There may be high inherent demand for this data.

    • Its high neglectedness means that even if we didn’t think people would be that interested, we shouldn’t be too confident in this belief. The value of information from trying this project seems high.

    • People seem pretty interested in what billionaires are up to in general.

    • Keeping track of who has helped others the most was an important concern throughout our evolutionary history.

    • A few people have told me they like this idea because billionaires often donate to their own ineffective foundations for tax deductions, to provide cushy jobs for their families and friends, and to give the appearance of being impactful. There may be a populist appetite for a list which highlights that many such donations are much less effective than they could be.

There’s also the question of whether the list is too hard to build well. I speculated in the previous section that this was the main reason why Impact List doesn’t already exist. I expect this to be very difficult, but the knowledge and expertise that EAs have developed around effectiveness evaluations should make this more feasible for us than for non-EA groups. However, Impact List has a few disadvantages relative to GiveWell:

  • GiveWell is trying to find the best recipients, so it can cut short its research when it determines a recipient won’t be among the best. Impact List will have to do deeper research into ineffective recipients because wealthy donors often donate large amounts of money to them.

    • Ineffective recipients are likely harder to research than effective ones—effective recipients probably care about being effective, and so care about establishing to themselves and others that they’re effective.

  • Impact List will have to evaluate a lot of recipients—in theory anyone who those on the list or within reach of the list have donated to.

    • In practice we’ll prioritize deep research into recipients by how much money has been donated to them and by their expected effectiveness, since those evaluations will have the most effect on the rankings.

    • We’ll also need to develop good techniques for approximating effectiveness after relatively shallow evaluations.

      • For example we might group recipients into similarity clusters, evaluate a small number of them, and tentatively extrapolate those evaluations to everything else in the cluster.

      • Recipients could also be put into a limited number of buckets corresponding to different orders of magnitude of effectiveness.

  • The effectiveness of a donation to a recipient is a function of time. GiveWell can focus on current effectiveness, while Impact List will have to assess the effectiveness of donations made long ago.

TL;DR: Both construction of the list and making it popular seem really hard. I’d guess that the combined difficulty of the outcome in the ‘Scale’ section (affecting 1% of total individual donations and 10% of top-1000 donations) is on the order of creating a billion dollar startup.

A very simple expected value calculation

I find calculating scores using the ITN framework for this problem less helpful than performing a direct expected value calculation[2].

If we consider the case where Impact List causes 1% of total individual donations to become effective, and 10% of top-1000 donations to become effective, then this results in ~11 billion extra dollars per year going to effective causes. Popular lists tend to remain popular for a while so if Impact List becomes influential we should expect benefits lasting longer than one year. If we estimate the duration of the list’s influence at ten years, we get $110 billion moved to effective causes[3].

What is the probability of at least this level of impact if the project were funded with around a million dollars per year? This is where almost all of the uncertainty comes in. I would guess it’s somewhere between 0.1% and 1%, which would result in a ten-year expected value between $110 million and $1.1 billion[4].

Impact List can still be very worthwhile if you think the amount of influence described above is unrealistic. If it affects 0.1% of total individual donations and 1% of top-1000 donations then the ten-year benefit is $11 billion. If the probability of this reduced level of influence is 1% then the resulting expected value is $110 million.

I’m very interested in seeing how others would estimate the expected value of this project.

Other risks and challenges

Rankings of people can feel arrogant and antagonistic

Publishing rankings of people has some inherent potential to rub people the wrong way. Done badly, we could come across as overly critical, arrogant know-it-alls who are claiming an authority to rank others which we don’t deserve. Introducing a sense of competition into altruistic donations may also cause a negative reaction.

The marketing of the list and the messaging on its website should be thoughtfully crafted to avoid causing negative reactions. We might stress that:

  • The list is primarily meant to celebrate people who are doing more good than almost everyone else in the world, and to inspire those who see it to increase their own impact.

  • We don’t claim our rankings are the final word and we welcome others to submit alternate metrics, which we’ll make available to viewers of the list.

  • Doing our best to figure out what’s effective is better than not trying, despite the difficulty, because the stakes are high and differences in effectiveness can be huge.

  • Wanting recognition for doing good is a common human desire, and if harnessing this desire helps create more good in the world then doing so seems better than the alternative.

The rankings may not seem credible enough to the public

Of course the rankings will seem less credible to the public than to EAs, but it’s unclear whether they’d still seem credible enough to the public to be worth paying attention to and engaging with. There are several reasons the public might dismiss the list’s rankings:

  • They may be too weird—for instance AI x-risk being judged as highly important.

  • Comparing effectiveness across domains (climate change vs. global health vs. animal suffering vs. x-risk) may not seem legitimate to people.

  • EAs are more willing than the general public to think it’s legitimate to tentatively accept highly uncertain best-effort estimates.

  • EAs will likely dominate early versions of the list because they’re already donating to the causes that we think are the most effective and because effectiveness matters so much. Impact List might then strike people as partly motivated by a desire to congratulate ourselves for being better than other philanthropists.

Allowing other organizations to submit their own rankings is one attempt to mitigate this. A few other things that we could do to make Impact List feel more credible to the public:

  • By default rank people by raw amount donated, and make our opinion of scaled impact just one of several columns that viewers could rank by if they chose to.

  • Assign effectiveness ratings more conservatively than our actual best estimates, especially for longtermist causes.

  • Initially separate out controversial cause areas into their own sections.

My guess is that we should not initially do any of these bulleted items, and should instead establish our credibility by publishing high quality explanations of our effectiveness research, but they’re options that we could consider based on viewer feedback.

Customization options may prevent a canonical version of the list from being recognized

There’s a tension between letting the viewer customize the list to match their preferences (about values and effectiveness-ranking providers) and having a single canonical list that becomes popular enough to influence donor behavior.

My low-confidence guess is that providing reasonable default values (possibly arrived at by collecting preference data from viewers) for these parameters would result in the default view being treated as canonical by most people. If there are multiple high quality effectiveness rankings we could use some blend of them as the default view.

Even if the default view of the list never becomes canonical, if the data is interesting enough then it might still put enough attention on donation effectiveness to have a significant impact.

Not all donations are public

The raw donation data for the list won’t be fully accurate because some people donate anonymously. The popularity of the Forbes List (which suffers from a similar issue) suggests to me that this is probably not a big deal.

Future extensions

Rank people by net externalities

In the ideal version of this list people would be ranked by the value of their net externalities (positive minus negative) from all activities (donations, running companies, writing, etc.). Calculating this would be much more work and would raise more issues about the credibility of the list.

Make the list much larger

If the list had 100,000 slots it may be possible for a typical software developer to appear on it by donating 30% of their income very effectively. Evaluating the effectiveness of all donations made by 100,000 donors would be infeasible, but it would be easy to give people credit for donating to any already-evaluated recipient.

This could be done automatically if Impact List integrated with effective donation recipients. Imaging donating to the Long Term Future Fund through Impact List’s UI. Your ranking could be updated immediately without any human effort. Organizations could also have options on their own donation pages to “share this data with Impact List”.

Feedback and collaboration

I’m very interested in feedback, especially on these questions:

  • How much of the EA community’s funds should go toward this project, assuming a high quality team can be assembled? Why?

  • How would you calculate the expected value of this project?

  • How would you make this proposal better? What did I miss?

  • Can you think of low-cost ways that we could test whether this is a good idea[5]?

Assuming this project is worthwhile, I’d like it to be led by the best person for the role, which may not be me. If you’re interested in working on this in any capacity, including leading the project, let me know.

I’ll be at EAG San Francisco from July 29th-31st if anyone wants to discuss this in person. I’ve also set up an Impact List discord server.

Thanks to Branimir Dolicki, Eric Jorgenson, Spencer Pearson, William Ehlhardt, Jack Stennett, Claire Barwise, Pasha Kamyshev, Vaidehi Agarwalla, and Issa Rice for providing feedback on drafts of this post.

As of 2022-06-08, the certificate of this article is owned by Elliot Olds (100%).

  1. ^

    Given the amount of work involved this would not happen before Impact List got very popular.

  2. ^

    My calculations using these scales for the ITN framework give the project a value of 26 -- 12 for importance, 12 for neglectedness, 2 for tractability. An uncertain assumption was that $5,000 donated effectively produces 50 QALYs. This table shows scores for other problems.

  3. ^

    Assume inflation and expected growth of donation amounts will roughly cancel.

  4. ^

    For a better approximation we’d subtract the impact of any counterfactual ineffective donations, but I’m assuming not doing so gives a good approximation since the most effective donations are much more effective than almost all others.

  5. ^

    An idea suggested by Vaidehi Agarwalla is to publish a much smaller version of the list (maybe top-10) as an article in something like Vox Future Perfect to see how much attention it gets.