Contents

Building Effective Altruism Japan, I often find myself explaining the following progression in Japanese:

  1. I want to make the world a better place as much as I possibly can.
  2. I should donate to x, that is working on problem y.

This blog post is intended to externalize my thought process when it comes to donating money1 - providing a Japanese version for budding Japanese Effective Altruists (EAs) that do not speak or read English.

While aligned with the principles of Effective Altruism, I ask readers to avoid conflating my thought process with EA as a whole.

For a more nuanced guide to Effective Altruism, I recommend the following resources:

What is Effective Altruism?

The world has many problems, many different ways of solving these problems, and only limited resources.

The Effective Altruism movement seeks to figure out, of all the possible uses of these resources, which uses produce the most good, impartially considered2.

What does it mean to “produce the most good”?

80,000 Hours suggests the following on this topic3:

“...‘having a positive impact’ is about promoting welfare over the long term. However, we’re highly uncertain about this definition, so in practice we aim to consider other perspectives.”

Considering moral frameworks

The focus on welfare as the only intrinsic value4 could be considered broadly utilitarian; an action can be determined as either good or bad by its ultimate effect on the welfare of people and other sentient beings.

When thinking about how to make the world a better place at higher levels of abstraction, I believe this is a useful starting point - but a complete focus on wellbeing is likely also non-optimal5.

Other theories of moral philosophy provide alternatives. In practice, people incorporate elements from many different frameworks into their everyday decisions and actions.

These include:

  • Deontology, which offers a complex set of rules to determine whether an action is good or bad.
  • Virtue ethics, which encourages a focus on building your character to be a “good” person.
  • Contractarianism, a view in which morality emerges naturally in a society as people cooperate for a better overall outcome for all participating agents.

With an emphasis on improving welfare in EA, it’s important to acknowledge a degree of uncertainty when we imply that it is a large part of what matters morally6.

What cause should I focus on?

Introducing the INT framework

It can be easy to equate social impact with a specific cause area (maybe due to its popularity or influence by those in your social circle) without properly considering alternatives. I think this is a mistake; if you want to have the biggest impact possible, it is worth spending time to examine multiple promising cause areas.

In the EA community, the INT (Importance, Neglectedness, Tractability) framework is often used to narrow down and prioritize cause areas:

Importance: How many people (and sentient beings) are affected by an issue and by how much?

Example: Climate change, in particular the non-negligible risk of extreme climate change, has the potential to affect many people, in a big way - thus it scores highly on the importance criteria.

Neglectedness: What degree of resources are already focused on resolving an issue?

Example: Climate change receives a lot of attention and a lot of resources; 80,000 Hours suggests that an effective $10-100 billion per year is spent trying to prevent climate change7.

There are other cause areas that may be just as important as climate change, but receive far less attention and resources - you should consider whether you may have a larger impact on the margin by focusing your resources on these.

Tractability: How easy is it to make progress resolving an issue?

Example: Aging affects everyone, in a big way, and is comparatively neglected as a cause area; but stopping or preventing the aging process is traditionally thought of as being extremely difficult8.

Expected value

Considering the “expected value” of an action or donation is a way of understanding its effectiveness.

Expected value = Outcome x Probability of Outcome

Let us consider two cause areas for which the intuition behind an “expected value” calculation may be entirely different.

1. Global health (smaller-scale outcome, higher probability)

There are a number of well-understood interventions in global health, including the distribution of anti-malaria bed nets, deworming, and giving cash to people directly. Many of these interventions have a large evidence base, often through Randomized Controlled Trials (RCTs)9.

GiveWell is a meta-charity that evaluates the effectiveness of charities in global health. They make charity recommendations based on four criteria:

  • Evidence of effectiveness (using aforementioned RCTs)
  • Cost-effectiveness (using their own model)
  • Room for more funding
  • Transparency

Using GiveWell’s recommendations, you can expect to a reasonable degree of certainty that your donation will have its intended effect - current (Nov 2020) estimates suggest that $5,000 given to the Against Malaria Foundation would save a life1011.

You can take a look at GiveWell’s recommended charities here.

Choosing a place to donate: Normal distribution vs. Fat-tail distribution (additional note)

When considering a place to donate in global health (where more information is available than a typical cause area), it is worth considering the difference in effectiveness between the most effective charities and the median.

  • Normal distribution; the tallest people are only about 50% taller than the average.
  • Fat-tail distribution; the highest-earning people earn thousands of times more than average.

80,000 Hours surveyed people on how much they think the most effective charities are relative to the median and found the typical response to be that the best charities are about 66% more effective (a normal distribution). Instead, the difference is more like 10x-100x (a fat-tail distribution); choosing a top charity over the median charity can have a huge effect on expected impact.

2. Reducing existential risk (larger-scale outcome, lower probability)

The future holds a lot of potential. The Earth could remain habitable for 600-800 million years, so there could be about 21 million future generations, and they could lead great lives, whatever you think “great” consists of12.

Existential risk refers to the likelihood of threats to this potential by either eliminating humanity entirely such that future generations cannot exist, or causing such great damage to humanity that it cannot flourish as it otherwise would. Risks of this nature include, but are by no means limited to, nuclear war, an engineered pandemic (that either kills or prevents the ability of its victims to reproduce), the extreme risks of climate change, or even the emergence of misaligned AI1314.

Efforts to reduce the risk of such threats constitute an outcome so vast that your contribution can be much larger in expectation, even if the probability of having such an impact is miniscule.

There are objections to focusing on the long term addressed in this paper, including, but not limited to:

  • Working on outcomes that occur in the far future is intractable.
  • People in the far future are likely to be far better off, so we should focus on improving welfare in the nearer term.

If you want to direct your donations towards reducing existential risk, the Long-Term Future Fund may be a good place to start.

Other cause areas

While the above two cause areas are popular within the Effective Altruism community, there are plenty more to consider. 80,000 Hours provides a body of research called “Problem Profiles” that I would strongly recommend reading.

The case for investing to give

A priori, it is unlikely that now, of all times, is the best moment to give. The size of the opportunity to make a difference at a given moment is sometimes referred to as its hinginess15.

In a paper “Patience and Philanthropy” (currently a work in progress) Philip Trammell discusses the pros and cons of waiting:

Pros:

  • Instead of donating $1,000 today, you could invest in the stock market - at a return of 5% per year, the amount of that donation would be $125,000 after 100 years, and $17 million after 200 years.
  • Much more information will be available about how best to use a donation.

Cons:

  • There is a chance the fund is destroyed through war, revolution or financial collapse.
  • The eventual use of the donation could drift from the original intended use.
  • The original intention of the donation could later be shown to be deeply morally mistaken.

EA-affiliated organization Founder’s Pledge has been conducting their own research into the promise of investing to give, with tentative plans to launch a Long-term Investment Fund in 202116.

So, where do I give?

In general, my donations have gradually moved from funding global health interventions (smaller-scale outcomes, higher probability of success) towards work that addresses global catastrophic risks (much larger-scale outcomes, much lower probability of success).

2016-2018

Back in 2016, I started by giving to charities recommended by GiveWell - GiveDirectly and Helen Keller International.

I chose GiveDirectly because the value proposition was clear - money is transferred directly to those who need it, and the evidence base for doing so suggests this money is used productively.

I chose to donate to Helen Keller International after coming across the model GiveWell uses to estimate cost-effectiveness. After adding my own moral weights to the model, it output the lowest cost per “outcome equivalent to a life saved” for Helen Keller.

2018-2019

Around this time I found some particularly insightful essays on the potential scale of wild animal suffering by Brian Tomasik, who also established the Center on Long-Term Risk, which has the unusual research focus of preventing s-risks (suffering risks) among future generations. This was one of a few projects funded by the Effective Altruism Foundation, where I directed one larger donation.

Late-2019 onwards

I have made several donations to EA Funds, typically at a rate of 75% to the Long-Term Future Fund and 25% to the Effective Altruism Infrastructure Fund.

I have also made smaller donations to The Humane League, which focuses on farm animal welfare, and the Malaria Consortium.

Future donations

I find the “investing to give” argument quite convincing, so I’m keeping track of the Founder’s Pledge investment vehicle in development. While by no means concrete, I expect short-term future donations to be a mix of investing to give and the aforementioned EA funds.

How much do I give?

While I have not taken it myself, the Giving What We Can pledge has been taken by 5,000+ people who give away 10%+ of their income.

I donated ~10% of my after tax income from 2016-2018, which I have since dropped to around ~7% (as of 2020). Since 2018, I have placed more emphasis on saving, at ~33% of my after tax income.

Over time, the following two points have led me to decrease the amount I donate and increase the amount I save:

  • Where I think may be the best place to donate has changed quite significantly in a short space of time; I now acknowledge a deep uncertainty.
  • I could potentially have a bigger impact by saving as much as possible to give myself more freedom to study and work on impactful projects17. If I have such freedom + excess funds, the option to donate still exists.

Footnotes

  1. Donating money is just one way of having an impact. Unless earning vast amounts of money is your comparative advantage by a large margin, it may well be that you can have a far larger impact by working on a specific cause area. ↩

  2. Condensed from a paper by Will MacAskill on “The Definition of Effective Altruism”. ↩

  3. https://80000hours.org/key-ideas/#two-minute-summary ↩

  4. In moral philosophy, there is a notable distinction between intrinsic and instrumental value. For some intuition of this distinction, consider the following - Think of a sublime natural landscape. Imagine now that this landscape exists, but no conscious being is able to experience it. Does the natural landscape have value without being experienced? Maybe you believe it does, in which case something about the landscape has intrinsic value - even without being experienced. Maybe you believe it does not, the value provided by nature is instrumental; a person is able to experience its beauty, feel an emotion, and ultimately a positive subjective experience - or increased wellbeing - which in itself has intrinsic value. ↩

  5. Most people are likely to have a set of intrinsic values, rather than just one. This article/talk by Spencer Greenberg does a great job of explaining how one might discover their own set of intrinsic values. ↩

  6. For further reading on dealing with moral uncertainty, see here (available for free from the “Open Access” button). ↩

  7. https://80000hours.org/problem-profiles/climate-change/ ↩

  8. Proponents of anti-aging research include Aubrey de Grey and Ethereum creator Vitalik Buterin. ↩

  9. There is some pushback in the EA community regarding the focus on interventions that can be tested by RCTs (randomista development) over actions that promote economic growth as a whole. ↩

  10. https://www.givewell.org/sources/blog-post/q1-q2-2020-discretionary-funding-allocation#2 ↩

  11. While I have suggested in the sub-header of this section that this is a “smaller-scale outcome”, remember this is still saving a human life! ↩

  12. https://80000hours.org/articles/future-generations/ ↩

  13. Further reading from the Global Priorities Project or The Precipice by Toby Ord. ↩

  14. In addition to known risks, there are unknown risks that will emerge over time that also need to be identified and solved. ↩

  15. https://forum.effectivealtruism.org/posts/XXLf6FmWujkxna3E6/are-we-living-at-the-most-influential-time-in-history-1 ↩

  16. https://founderspledge.com/stories/investing-to-give ↩

  17. My skill set is rare for an EA and there is potential to use this for more impact than donations alone - I’m based in Tokyo, speak/read/write Japanese to a high level, and have some (admittedly limited) research/data science experience. This has led to research work with Mercy For Animals (previously recommended by Animal Charity Evaluators), and other work not yet public - work that I may not have been able to take, or would have been less willing to take, if I didn’t have financial security. ↩