[Disclosures and disclaimers: I do work for the Future of Humanity Institute, and significant consulting for Open Phil. I have previously worked at MIRI, and consulted for 80,000 Hours/CEA. I have advised the EA Giving Group DAF regarding donations, and suggested to Paul Christiano that he facilitate the donor lottery discussed in the post. I am writing only for myself, and not for any of the above organizations.]  

Summary

Risk-neutral small donors should aim to make better charitable bets at the margin than giga-donors like the Open Philanthropy Project (Open Phil) and Good Ventures using donor lotteries, and can do at least as well as giga-donors by letting themselves be funged. To do so, I recommend entering Paul Christiano's donor lottery. If you win, do more research, including into the option of entering a larger lottery to reach optimal size, in between small and giga-donors. 

Introduction

It's giving season and numerous sets of recommendations for individual donors are out: GiveWell top charities, GiveWell staff donations, Center for Effective Altruism (CEA) staff donations, Open Philanthropy (Open Phil) staff recommendations for individual donors, Animal Charity Evaluators (ACE) recommendations, ACE staff personal donations, and an additional post on the EA Forum with several donors' decisions and reasoning (plus more in comments).

There are also opportunities that get pursued and funded by major funders but not pitched to small donors in these lists. The Open Philanthropy grant database is a rich source of examples (these are not recommendations for individual donors). These opportunities may be costly to explain to small donors, benefit from large minimum grant sizes (startup funds for a new organization, or stable funding to hire for a new position, etc), are funded as soon as they are identified, or just happen to lack any charity evaluator doing the work of evaluating them for small donors.

Some people in the effective altruist community have argued that small donors should accept that they will use marginal charitable dollars less efficiently than large actors such as Open Phil, for lack of time, skill, and scale to find and choose between charitable opportunities. Sometimes this is phrased as advice that small donors follow GiveWell's recommendations, while Open Phil pursues other causes and strategies such as scientific research and policy.

In contrast, my view is that risk-neutral effective altruist small donors should plan to make bets at least as good as the benefit of the marginal dollar for a giga-donor like Good Ventures with access to research resources like Open Phil. The 'at least' component is easy to achieve: to match the performance of a giga-donor's marginal dollar, one can substitute one of your dollars for it in a grant that it would otherwise make and email asking it to 'funge' you by accordingly reducing its grant. This is a great lower bound to have, as it takes advantage of the research capacity and economies of scale of a sophisticated donor working hard to find the best opportunities. It is also a very challenging benchmark to beat: how could a small donor with limited research time or scale economies plan to make even better bets?

The availability of donor lotteries (see here for details) means that small donors can convert their small donation into a probability of being a large donor. So whether the scale of best marginal donor (per dollar) would be small, medium, large, or enormous, the small donor can access that scale. There are both advantages and disadvantages of larger scale, and the optimum distribution could lie in various places. Would it be best to have a single ultra-scale donor making grants (including grants for re-granting), 10 donors of 1/10th the scale, or 2 donors with 25% each and 100 with 0.5%?

In this post, I argue that while economies of scale favor donors with substantial budgets to enable deeper research and sourcing of opportunities, diseconomies of scale favor an intermediate range well short of $10 billion at the current margin, in the $100,000-$100,000,000 range. Small donors can access these opportunities through staged ascent via donor lotteries to larger scales (e.g. a donor with a budget of $1,000 might first lottery up to $100k, then assess further lottery steps).

In outline:

1. Small donors can use donor lotteries to become larger donors, capturing scale economies and making it worthwhile to research key questions about donation conditional on winning.
2. If you believe that any particular large donor will make better donation decisions than you in expectation (by your values), then you can delegate your donation decision, e.g. by donating to an Open Phil grantee and asking Open Phil to 'funge' you and reduce its donation by that amount.
3. Selecting a delegate for donation decisions, or comparing delegation to an object-level opportunity, is itself a research problem one can invest in, investments that can be made cheaper using donor lotteries.
4. Assessing the value of a marginal dollar given to a giga-foundation depends on the value of the 'last dollar,' after further research and updates, but also after diminishing returns.
5. Organization-wide risks and distractions are more costly for larger donors, allowing smaller donors to better harvest opportunities with these downsides.
6. Funding charitable opportunities entirely from a single source carries some downsides. Multiple intermediate size donors can reduce these.
7. Intermediate scale donors may be able to spend more time per dollar allocated than smaller or larger donors.
8. Differences in values or worldviews may mean a small donor can't find a large donor with fully aligned aims to delegate to, and may want to create one, or use a more aligned delegate of intermediate size.
9. In my view some medium-size donors have been able to take advantage of some of these factors to outperform the 'last dollar' of giga-donors.
10. Advice for small donors transitioning to medium and larger scales using donor lotteries.

Donor lotteries make highly-informed donation affordable for small donors in expectation

As I discussed in a recent post, donor lotteries allow small donors to take advantage of economies of scale in donation and research by buying a small chance of allocating a large donation pool. E.g. instead of giving $1,000, one could buy a 1 in 100 chance of allocating $100,000, or a 1 in 1,000 chance of allocating $1,000,000. If you 'win' the lottery then you can invest in better-researched donation, but that investment only needs to made in a minority of cases. In the previous examples this would cut the expected cost of research per dollar allocated by over 99%.

There are many investments of time and resources in improving donation quality one could make after a 'win.'

• Evaluating the rest of the arguments in this post
• Read the work of charity evaluators and advisors like GiveWell or ACE in depth to evaluate your trust in their recommendations and spot-check or audit sampled claims
• Evaluate how idiosyncratic values and priorities interact with the recommendations (e.g. putting your own values and estimates into the GiveWell cost-effectiveness spreadsheet)
• Personal discussions (although sharing public notes can generate additional value) with staff from organizations and respected independent advisors; this can be done in depth that would not be worth their time for a small donation
• Deeply engaging with ethical questions relevant to prioritization, such as the treatment of nonhuman animals and future generations, or population ethics
• Evaluating whether to lottery up to larger amounts where diminishing returns might matter more (lottery for $100,000 initially, then evaluate whether to go for $1,000,000 if you win the first lottery)
• Considering how much to donate now vs later (since the funds are in a DAF, they could compound there until donation)
• Taking time off work for research (otherwise unattractive)
• Seeking out donation opportunities with large minimum size
• Putting out a call for proposals inviting organizations and individuals to pitch you
• Hiring your own research staff, along the lines of Open Phil
• Negotiating with other large donors regarding funging and coordination
• Spot-checking/auditing the work of charities or charity recommenders

The post also has information about a donor lottery being run by Paul Christiano at my suggestion, with personal participation by a number of effective altruists, including GiveWell and Open Phil staff), and the details of the public random draw.

In light of the availability of donor lotteries the rest of this post will be assuming that large donation sizes and time investments are accessible for small donors (probabilistically).

If you believe the expected impacts of donations by another donor are greater than your own, you can delegate your donation

Suppose that you thought the Bill and Melinda Gates Foundation was the 'smart money' and better at giving than you according to your values. In that case if you had no better alternative you could simply donate to the Foundation. That's what Warren Buffett is doing, with a donation worth over $30 billion. In principle one one could donate to the donor-advised fund (DAF) of Open Phil, directly increasing its ultimate donation capacity. At the moment, this doesn't seem to be set up, but one could instead donate to something that Open Phil is donating to (inframarginal), and request that it 'funge' you by reducing its own donation to that charity by the corresponding amount, increasing the reserves of Good Ventures and other Open Phil backers accordingly.  So the marginal Open Phil/Good Ventures dollar sets a minimum standard for risk-neutral donors: if you don't expect to do better than Open Phil, just arrange to get 'funged'.

Likewise, one can delegate to other trusted medium and small donors. Holden Karnofsky discusses this on the 2016 GiveWell staff personal donations page:

• I thought about reallocating my giving to another individual, someone who is quite value-aligned with me and quite knowledgeable, and thinks differently enough that they might see opportunities I don’t. As a general point, I think reallocating to others addresses a similar issue to the donor lottery—trying to consolidate donations so that a smaller number of people can put in a greater amount of effort – and it seems to me that it is a better way of doing so when one has a person in mind they’re comfortable reallocating to. (Of course, hybrid approaches are possible too —one could reallocate to a person who then plays the lottery, with the winner of the lottery considering reallocation as well.)

I haven’t finalized my decision yet, but I am leaning toward the last option. The “EA Giving Group” DAF mentioned by Nick is one possibility, and there are others as well.

This option means that risk-neutral effective altruist donors trying to maximize expected impact with their donation should take donation delegates as lower bounds for an expected value 'hurdle rate.' If one lacks much evidence for the quality of a donor, this may be a fairly low bar: just like selecting object-level charities, evaluating the knowledge, capacities, motives and constraints of a possible delegate is a significant research task. However, there is substantial evidence available about these things for a number of possible delegates, and donor lotteries can be used to reduce that research cost and identify some strong donors.

Thus, a risk-neutral effectiveness-oriented donor who donates to charity X implicitly communicates that they think it will use the marginal dollar better than the Gates Foundation, or Good Ventures and Open Phil, multibillion dollar organizations with large professional staffs (particularly the former) working full-time to pick out the best grant opportunities.

Having the marginal Open Phil dollar as a lower bound hurdle rate is great news: my expectation for the 'last dollar' of that portfolio is exceptionally high relative to the general world of charity. Among other things, I think even after diminishing returns some combination of scientific research (e.g. gene drives to eradicate vector-borne diseases), policy work (e.g. on foreign aid or science policy), nonhuman animals, global catastrophic risks (potential risks from AI, biosecurity, nuclear risk), and others put the expected value of the 'last dollar' for Good Ventures higher than for GiveWell's top charities.

So saying that charity X is a better bet than adding to Open Phil's reserves, or saving in a donor-advised fund to make use of future findings from Open Phil research, is a strong claim. Nonetheless, despite full knowledge of this argument, including blogging about it and related issues, I have made other recommendations to effective altruists seeking donation advice for a number of years (albeit frequently mentioning this consideration) which implied the claim about various opportunities. Why did I believe that?

Outperforming a giga-donor means outperforming the expectation of their 'last dollar,' after increases in knowledge and diminishing returns

In a recent Open Phil post Holden wrote "we expect to have over $100 million worth of grants for which the investigation is completed this year [2016])" for OpenPhil, not including the $50 MM donation to GiveWell top charities recommended to Good Ventures. While these are sizable amounts, in light of Good Ventures' resources (on the order of $10BB), it is saving almost all of its financial resources for better future opportunities, at the same time as it supports, through Open Phil, research into better prioritizing and identifying such opportunities.

In the debate on giving now vs later, so far this reflects a lean towards later. A 2015 GiveWell blog post discussing recommendations of grant timing to Good Ventures discusses plans for giving to rise as research into cause and charity selection continues and staff capacity at Open Phil increases. I would expect this process to tend to improve the quality of future recommendations and donations, as true beliefs will tend to be favored by careful investigation.

However, a massive offsetting contrary consideration is diminishing returns. Donations that make a large proportional difference to funding in a field, or seed a new field, can pluck 'low-hanging fruit.' [Also see Owen Cotton-Barratt on diminishing returns.] Open Phil is large relative to some of the philanthropic fields it works in, such as factory farming or potential risks from artificial intelligence, but in global health it is small relative to players such as the Gates Foundation. As it builds up capacity and grows small priority fields, there will be much less room for on major proportional changes to funding waterlines.

For illustration, imagine logarithmic returns, where a proportional expansion provides the same utility gain regardless of the previous size of the field. For some smaller fields expansion by a factor of 10-100x is possible, which would then correspond to 90-99% reduction in marginal impact therein. Even if further research will predictably substantially improve donation allocation and reduce uncertainty, allocating enough for some growth while fields are small can beat the last dollar in expectation (until the rival factors are in balance, which would still involve most spending lying in the future for donors in aggregate).

The chance to pluck time-sensitive low-hanging fruit seems to me to be the main reasons for donors, large and small, to be giving some now, rather than saving (or giving to a DAF to take advantage of tax benefits) to await improved research into opportunities and investment returns. For small donors, getting to low-hanging fruit when existing large donors have not seems key to outperformance, but such opportunities depend on some barrier to entry, an explanation of why the large donors haven't taken the opportunities.

Smaller donors have less to lose from organizational systemic risk

A giga-donor on track to spend billions of dollars plucking low-hanging fruit in many fields is a tremendously valuable asset. If some grants trade off a fixed altruistic benefit per dollar for some risk to the donor's organization, where the risk scales with the importance of the organization, then they may have negative value for large donors and positive value for small donors.

Such risks could include time-consuming complications that distract senior management making organization-wide decisions, controversies that affect the organization's reputation, and impacts on staff morale or organizational culture, among other things. Other risks could involve changes with contrary effects on work in different causes. For example, consider a foundation attempting to separately promote critical study of religious texts in ways that are seen to promote atheism and simultaneously to build relationships with offended religious leaders for cooperation on other policy issues.

A recent Open Phil blog post by Holden discusses non-monetary costs of grants, including communication costs, and the fact that grant decisions "reflect to some degree on all of the 20+ people who work for the Open Philanthropy Project." Another post mentions effects of diverse grants with "secondary benefits...specific to a public-facing organization with multiple staff" including ones on morale and recruiting.

This provides rational reasons for large donors to be more cautious and risk-averse in many of their activities, including their communication and grantmaking strategies, but also opportunities for new entrants to benefit from smaller size and having less to lose. It would still be crucial to consider systemic risk across movements and cause areas they participate in, and could come at the expense of some synergies of diverse portfolios, but would at least mitigate organization-specific risks.

As the number of entrants increased they could specialize. Multiple independent specialized entrants would further reduce these risks compared to a hypothetical 'Controversial Philanthropy Project.'

Similar considerations may also provide extra reason to be cautious in hiring and staffing, as impacts on organizational culture become increasingly important.

Problems with single-donor funding

There are a number of possible problems with being the sole funder to a nonprofit. If these apply, it might be that an additional dollar from a large donor has a different impact on a charity than a dollar from a small donor.

The most important involve the compromise of independence on the part of a charity, which affects both its actual decision-making, and the way it is perceived by others. For example, GiveWell has tried to limit donations for its operating budget from Good Ventures (particularly for non-Open Phil work) to preserve its independence, real and perceived, and similar considerations may apply elsewhere. Advocacy efforts may be seen as 'astroturf' or just the voice of one funder, analysts may feel less free to produce results that are unwelcome to the funder, scientists may optimize excessively for impressing the grant-maker, etc.

Further, an individual grantmaker may have particular biases, relationships, and other factors that can influence giving separately from maximum benefit. When a nonprofit receives funds from multiple sources there is less pressure to optimize for such biases (since they can cancel out across funders).

A grantee that becomes dependent on a single large funder makes it much more difficult for the large funder to withdraw: the organization could collapse, creating significant harm for its staff, unwelcome media attention, and negative feelings and feedback in the grantmaker (who may have developed relationships with the grantee). This can harm the grantmaker's reputation and make others less willing to deal with it or depend on its funding.

There are also some legal considerations, which are generally of lesser importance. In the United States nonprofits that qualify for 'public charity' status enjoy favorable regulatory treatment, e.g. avoiding a 2% tax on investment income, limits on self-dealing and business holdings, prohibitions on lobbying, somewhat worse tax-deductibility, ability to receive donations from DAFs, etc. One requirement is that a nonprofit pass a public support test, showing that it receives at least a third of its donations from other public charities (including DAFs), government, and the general public (a given donor can only account for a maximum of 2% of this support), or 10% plus additional supporting facts and circumstances.

If public charity status is important for a charity, and it is close to the cutoff, then this would be a reason why there could be additional valuable donation opportunities a large donor could not claim. However, it is not clear to me that this status is essential in many cases, particularly if the alternative was much greater expansion, and it often does not apply (e.g. for donations to support a program within a large organization like a university). If it is an issue, this would be a very logical occasion for a legitimate donation matching challenge where the large donor matches donors below the 2% threshold.

The problems of single-donor-dominance are again limitations that could give a per-dollar-donated advantage to new entrants, even if they had equal skills and identical views.

Spending more time investigating per dollar allocated

Major foundations commonly grant millions of dollars per year per employee and that is currently the case for Open Phil. The Wikipedia page for Open Phil shows ~$10 MM in each of  farm animal welfare and criminal justice reform since program officers Chloe Cockburn and Lewis Bollard were hired in mid and late 2015. These figures do not include any grants that have been made but not yet been published to the Open Phil grants database and copied to Wikipedia, but in any case amount to allocations of thousands of dollars of donations per program officer hour.

Considering the ~$60 MM in published grants and the entire FTE staff across all roles funds allocated still appear well over $1,000 per hour. Moreover, much staff and management time has gone into capacity-building, e.g. hiring much of the current team. Open Phil has written that it would like a short-run budget closer to 5% of available capital, i.e. over $400 MM annually, even before reaching 'peak capacity' to evaluate opportunities.

These figures suggest a new donor could invest more hours into investigating donation opportunities per dollar donated than Open Phil is, or add more donations with the same ratio. However, that raises questions about why any inputs by the new donor aren't being used by existing well-funded donors:

• If a new donor hires research staff and program officers, why won't existing foundations hire them? Are they less effective, or diverting talent from similar activities elsewhere?
• If the donor does investigation to guide their own donations and/or publish the results, why can't she simply do the same research and let existing players act on it?
• Would this time be better put towards executing direct projects, letting others fund them?

However, there are clearly cases where additional organizations can relieve the difficulty of hiring in ways that are positive sum, e.g.:

• Hiring staff, or granting to autonomous grant pools, carries organization-wide risks, as discussed above
• Hiring staff to live in different cities, organizational cultures, etc
• Relieving bottlenecks on the time of top management and funders in approving and supervising hires (see this GiveWell post explaining its management time constraints on hiring)
• A donor lottery winner may be more confident of alignment with their own aims

The findings of additional research can be published and contributed to the common stock of knowledge, including for use by giga-donors.

Scale economies for different values or worldviews

Most of the considerations above have been general structural ones about optimal scale and concentration for donors. But historically I think a lot of the outperformance by small donors that I have seen has reflected differences in what a recent Open Phil blog post calls 'worldviews.'

For example, in addition to GiveWell's explicit criteria, its published analyses of charities outside the Open Philanthropy Project do not discuss effects on nonhuman animals or future generations, and its evaluations depend on particular judgments about population ethics. Charities focused on nonhuman animals today offer the potential to affect orders of magnitude more life-years of creatures such as chickens than GiveWell recommendations affect human life-years. This disproportion is substantially larger than the disproportion in the scale of human and chicken nervous systems (see this post for some discussion), and expert opinion on average favors chicken consciousness and the moral significance of animal welfare. The disproportion between the size of the current population and future generations is even greater. The Open Phil post discusses some numbers:

For a relatively clear example, consider GiveWell’s top charities vs. our work so far on farm animal welfare:

• GiveWell estimates that its top charity (Against Malaria Foundation) can prevent the loss of one year of life for every $100 or so.
• We’ve estimated that corporate campaigns can spare over 200 hens from cage confinement for each dollar spent. If we roughly imagine that each hen gains two years of 25%-improved life, this is equivalent to one hen-life-year for every $0.01 spent.
• If you value chicken life-years equally to human life-years, this implies that corporate campaigns do about 10,000x as much good per dollar as top charities. If you believe that chickens do not suffer in a morally relevant way, this implies that corporate campaigns do no good.3
• One could, of course, value chickens while valuing humans more. If one values humans 10-100x as much, this still implies that corporate campaigns are a far better use of funds (100-1,000x). If one values humans astronomically more, this still implies that top charities are a far better use of funds. It seems unlikely that the ratio would be in the precise, narrow range needed for these two uses of funds to have similar cost-effectiveness

I think similar considerations broadly apply to other comparisons, such as reducing global catastrophic risks vs. improving policy, though quantifying such causes is much more fraught.

The weight one puts on different worldviews is a natural question one could research to guide donations and try to improve on Open Phil's recommendations, if only in better reflecting one's idiosyncratic preferences. If one puts relatively more weight on nonhuman animals or future generations, or treats uncertainty differently, then this could imply a somewhat lower threshold for donation than Open Phil's diversification would when, e.g. considering factory farming or global catastrophic risk interventions.

In the context of a donor lottery, deeper investigation into worldview-related questions is one way to improve decision quality, or to better reflect idiosyncratic values. For both, but especially the former case, I would recommend cautious reflection on the reliability of one's own intuitions, and the evidence and reflection informing various views.

Have medium-size donors in effective altruism been able to add value to large donors in the past?

The above conditions are fairly general and abstract. But, more concretely, in past years I have recommended donation targets other than contributing to the Open Phil grant pool to people asking my advice, generally in the areas of reducing potential existential risk from future developments in artificial intelligence, and developing institutions in effective altruism. These recommendations were for areas where several of the above factors applied: organizational risks, reputational/communication problems, staff bottlenecks, and interactions with broader worldviews. In subsequent years OpenPhil did enter the areas, but the early grants were able to fund time-sensitive opportunities such as seed and growth funding.

This focus area, and many of the recommendations and evaluations have had extensive overlap with those of the 'EA Giving Group' DAF mentioned by Nick Beckstead on the 2016 GiveWell personal donations page (and I have frequently discussed charity opportunities with Nick):

This year I am donating to the “EA Giving Group” DAF (donor-advised fund). Since 2012, one of my side projects has been working with a private individual (who has provided the vast majority of the funds and prefers to remain anonymous) to make donations to organizations working in the effective altruism space and organizations working on mitigating global catastrophic risks (especially potential risks from advanced AI). We meet every three weeks to discuss potential donation opportunities and make decisions, and we both keep up with activities in the space through relationships we’ve built up over time. The DAF is jointly controlled by me and this partner.

A list of donations we’ve made in the past (without dollar amounts) is available here (arranged by year and decreasing order of grant size). The organizations that received the most funding were the Centre for Effective Altruism (CEA), the Future of Life Institute, 80,000 Hours (part of CEA), and Founders Pledge. I think these grants have gone well overall, as has our support for Charity Entrepreneurship and the Cambridge Centre for the Study of Existential Risk. In most cases, we supported these organizations relatively early in their existence, and we’ve mainly supported them when they were new or relatively young.

Over the last year, Open Phil has also made grants in these areas based on my recommendations. I anticipate that there will be some cases where a grant would be a good fit for this DAF but not Open Phil. However, with Open Phil as a funder in this space it has been harder to find opportunities that are as promising and neglected as we were able to find previously.

I don’t yet know what this DAF will support in the coming year, but it will probably have a similar flavor to what was supported in the past.

I am making this donation instead of a donation to GiveWell’s top charities primarily because (i) I think this is more optimized for influencing long-term outcomes for the world (which is my primary altruistic objective—reasoning here) and secondarily because (ii) I think we have a good chance of getting a “multiplier effect” where support of the effective altruist community eventually results in more total donations to GiveWell’s top charities and other things I find comparably good.

If you want to make a contribution to this DAF, then fill out this form.

This might be a good fit for people who have some combination of the following properties: interest in effective altruism and/or global catastrophic risks, context needed to assess our (still early) track record, trust in my judgment and/or my partner’s judgment, limited time/context available to make donation decisions themselves. We update contributors on grants made a couple of times per year.

This DAF was able to enter these areas years before GiveWell or Open Phil in part because of considerations along the lines discussed in earlier sections. A 2015 Holden post at the Effective Altruism Forum explained reasons why Open Phil was not at that time funding organizations in the effective altruism community including:

• High staff time requirements for investigating and making grants outside of a focus area, competing against other uses of management and staff time (like hiring more program officers)
• Complications involving the independence of organizations (large single funder issues)
• Organizational risks from difficulty communicating Open Phil's areas of agreement and disagreement with effective altruism and members of that community
• Other donors in the EA community were highly familiar with the organizations in question and able to donate with fewer of the above costs
• The opportunities were not considered so outstanding as to outweigh the rest of the factors

Holden's post also noted these non-monetary costs might decline (or benefits rise), and their decision might change in the future, and since then Open Phil has started to investigate and make grants in the area of effective altruism.

In the area of potential risks from advanced artificial intelligence, non-monetary and investigation costs were high for an area that was relatively unusual, controversial (potentially with various organizational costs and risks) and hard to evaluate. This cause was more important in worldviews that put more weight on long-run outcomes (and strategies to affect them at a large scale rather than through local linear effects), which I placed more credence in (after long thought). It was also an area that I had spent a lot of time investigating from many angles, and that took a lot of time to fully communicate the evidence base behind.

Open Phil has taken on potential risks from advanced artificial intelligence as a major focus area, and some staff have updated some of the heuristics that affected this opportunity. So in both these areas future opportunities may less often look like funding what Open Phil can't, and more often amount to different allocations or amounts of funding. Nonetheless, focusing on differences between the situations of smaller donors and Open Phil may also help identify new opportunities (or ones that are more costly for large donors) here, and similar gains may be had in other domains.

Another case where small and medium size effective altruist donors got to a cause area to pluck some low-hanging fruit prior to Open Phil subsequently becoming involved was in the case of nonhuman animals, both farmed animals and wild animals.

In the area of global poverty Thomas Mather contributed to work on using gene drives to eliminate schistosomiasis, also commissioning a research report on the topic by the Philanthropy Advisory Fellowship (stemming from Harvard Effective Altruism). In this case the funds preceded Open Phil's grantmaking in gene drives by a shorter time, and were much less independent, but still to some extent may have expedited support for the area. This is probably less of a success than some of the other candidates above, but is the sort of thing that small and medium donors might try in order to outperform giga-donors.

The fact that Open Phil subsequently funded an area does not mean that earlier EA donors were outperforming the 'last dollar,' since the last dollar will be informed by much more information, but because of low-hanging fruit, it is plausible that this sort of early arrival to a cause area pays off for a portfolio of such cases.

Advice for donor lottery winners

Suppose that you participate and win Paul Christiano's donor lottery and are in a position to recommend an allocation of $100,000 in charitable donations after January 15th: what would I advise to take advantage of scale economies and diseconomies?

One issue to consider is that while one could then attempt a further round of donation lotteries, e.g. to get a 10% chance of allocating $1,000,000, or a 1% chance of allocating $10,000,000, if there are diminishing returns over moderate amounts, low-hanging fruit might be missed (in expectation).

For example, if the most attractive opportunity (taking into account the limitations on other donors,  issues with single donors providing too much funding to an organization, etc) has a budget of only a few hundred thousand, then a 10% chance of $1,000,000 might do significantly less good at the margin. A donor who was large relative to an organization's funding dropping out because of a lottery loss could cause a fluctuation that could complicate planning.

For donors scaling up to $100,000 such risks are quite manageable, and I think outweighed by the gains of decision quality and options versus small donations, but I flag this as a concern for that level. Following such a lottery win and assessment of options (including investigating worldview questions, possible donors to delegate to, diminishing returns in promising areas and organizations) one can assess this issue for the next stage. If the possibility of disruption of funding to an area by chance is a big concern, one can conduct a donor lottery with other (current) supporters of the same cause or organization, so that one can be confident the assessment with the concentrated donation pool will be made by someone within the starting group.

At a larger scale, e.g. $1,000,000, $10,000,000 or more, it may make sense to 'institutionalize' to a greater extent, forming a team to do research, donating in support of answering specific research questions for information value, and similar. One option would be to pool with other medium-size donors or groups like the EA Giving Group DAF, but I suspect that creating more than one such group (if one can find people with the necessary skills who want to participate) could add value by incorporating more perspectives and managing some of the diseconomies of scale discussed earlier.