Book Summary: Doing Good Better by William MacAskill

Doing Good Better by William MacAskill

This summary of Doing Good Better explains the reasoning behind many Effective Altruism (EA) ideas. Its author, William MacAskill, is one of the founders of the EA movement.

I note that the EA movement has changed a bit since 2015, when MacAskill wrote the book, with a stronger focus now on longtermism. That shift is reflected in MacAskill’s second book, What We Owe the Future. However, I think the core ideas explored in Doing Good Better are still very solid and worth understanding.

You can get a free physical copy of Doing Good Better, as well as some other effective altruism books, here. If you want the audiobook or e-book, you can buy it at: Amazon | Kobo (affiliate links)

Key Takeaways from Doing Good Better

  • Effective altruism is about asking how we can use evidence and reasoning to try and find make the biggest difference we can.
  • If you live in a developed country, you can probably do a lot of good at relatively little cost to yourself.
  • We should take a reasoned, scientific approach to doing good:
    • Resources are scarce so we inevitably have to make trade-offs, even if all causes are “good”.
    • Some causes can be much, much more effective than others. So even if we can’t measure the effectiveness of different causes accurately (as is often the case), ballpark estimates can still be useful.
  • How to compare causes:
    • MacAskill suggests using a 3 dimension framework of Scale, Neglectedness and Tractability. (Personal fit is a 4th dimension if considering career decisions rather than donations.) Not many causes will be high on all 3 dimensions and reasonable people can differ on how much weight to give each dimension.
    • Look at expected value. A cause that is unlikely to succeed may still be worth supporting if its potential impact is very high.
    • Consider Quality Adjusted Life Years (QALYs) saved by different causes.
    • Apply counterfactual thinking – what would happen if you didn’t take that course of action? Would that life get saved anyway?
  • How to evaluate charities:
    • Ask 5 questions including: what does this charity do, how cost effective are their programmes, how robust is the evidence behind their programmes, how well do they implement their programmes, and how much room do they have for additional funds?
    • Don’t take into account a charity’s overhead or CEO pay as these are irrelevant to the effectiveness of a cause.
  • Ethical consumerism is usually not a very effective way of doing good, mostly because it’s not very targeted. You’d usually be better off saving extra the money and donating it to effective charities instead. (An exception might be not eating meat for animal welfare reasons).
  • When thinking about doing good in your career:
    • Consider personal fit, impact, and career capital.
    • If you’re not sure what to do, especially early on, build generalisable skills and try different things to get more information.
    • Some career options, like direct work and earning to give, are solid bets in that you’ll likely have a positive impact. Other options, like entrepreneurship and research, are speculative long shots. The chance of success is low, but if you do succeed, your impact could be enormous.

Detailed Summary of Doing Good Better

You can probably do a lot of good at relatively little cost to yourself

Most people living in developed countries could potentially make a huge impact through donations.

If you earn above US$52,000 per year, you’re in the top 1% of income in the world. If you earn above US$28,000 per year, that’s still in the top 5% of income globally. [These thresholds have likely increased slightly due to inflation as Doing Good Better was written in 2015, but the point remains.] Those figures are after taking into account the fact that money usually goes further overseas than in the US, and the fact that people in poor countries may produce a lot of their own goods. However, MacAskill points out that in poor countries, there are markets for extremely cheap goods that simply do not exist in richer countries. For example, the lowest quality rice available in the US is much better than the lowest quality rice in Ethiopia.

There are diminishing returns to income, so you can have a huge positive impact on others’ lives at relatively little cost to yourself.

  • Evidence shows that when income doubles, subjective well-being also doubles. So doubling someone’s $1,000 salary will generate the same amount of happiness as doubling another person’s $50,000 salary.
  • The typical US wage is US$28,000 per year, which is around 100 times as much as the poorest people in the world. So one additional unit of income can do 100 times as much to benefit the extreme poor as it can to benefit the average wage-earner in the US.

Current estimates of the cost to save a life in the developing world is about $3,400. Most people in rich countries could donate that amount every year and maintain about the same quality of life. If you think of being able to save a life each year, that would be pretty impressive.

Sometimes we might look at the size of the problems in the world and get discouraged. If anything we do would just be a drop in the bucket, why bother? But MacAskill points out this reasoning is flawed:

“It’s the size of the drop that matters, not the size of the bucket, and, if we choose, we can create an enormous splash.”

William MacAskill in Doing Good Better (2015)

Aids sceptics are often wrong

Aid sceptics suggest that lots of money has been spent on foreign aid, with little to show for it. This is wrong on both fronts.

Not that much has been spent on foreign aid, relative to the size of the problems:

  • For example, in William Easterly’s book The White Man’s Burden, he emphasises that the West spent $2.3 trillion on foreign aid over five decades. In contrast, the annual economic output of the world is $87 trillion, and the US alone spends about $800 billion in social security every year.
  • Over 60 years of aid spending, $1 trillion is just under $17 billion per year. If you further divide by the population of sub-Saharan Africa at the time (412 million), that’s merely $40 per person per year.

And there has been plenty to show for it:

  • In 1950, life expectancy in sub-Saharan Africa was 36.7 years. At the time of MacAskill’s writing, it was 56 years. Obviously correlation does not equal causation but there are other reasons to think that aid spending has helped.
  • The best aid programme ever might be the eradication of smallpox in 1977. Smallpox was a horrific disease that badly disfigured people and killed around 30% of those infected. Even if foreign aid did nothing other than eradicate smallpox, it would have been a bargain. Eradicating smallpox saved around 60 million lives (a conservative estimate), which equates to $40,000 per life saved. For comparison, US government departments use figures of around $7 million per life to determine if safety improvements are worth it.
  • Immunisations have reduced annual deaths from preventable illnesses from 5 million in 1960 to 1.4 million in 2001, even though the world population doubled over that period.
  • Annual malaria deaths have also declined from 3.8 million to about 0.7 million.

In the footnotes, MacAskill also refers to some studies published since William Easterly and Dambisa Moyo’s books showing a positive relationship between aid and economic growth.

Moreover, even aid sceptics tend to agree that the best development programmes, especially in global health, are very effective. MacAskill argues that when evaluating if aid works, you should look at the best cases instead of the average cases. We can choose to fund only the best programmes.

Why should we take a scientific approach?

The point isn’t to lay blame or claim that some causes are unworthy. It’s just about trying to work out which ways of doing good are best, and do those first. Resources are limited. We inevitably have to “triage” when we make decisions about where to donate our money.

Example: The importance of triaging

When James Orbinski worked at the Red Cross hospital during the Rwandan genocide, patients came non-stop. Staff had to tape numbers on patients’ foreheads: 1 (treat now), 2 (treat within 24 hours) and 3 (irretrievable). They had to triage and prioritise to try and save the most lives possible.

If staff had simply refused to triage, treating people on a first-come-first-served basis, many more would have died. Our world is the same. There are more problems than there are resources to solve them, so we are forced to triage and prioritise.

How do you work out which approaches are “best”?

There are 5 key questions in effective altruism:

  1. How many people benefit, and by how much? [Impact]
  2. Is this the most effective thing you can do?
  3. Is this area neglected? [Neglectedness]
  4. What would have happened otherwise? [Counterfactual reasoning]
  5. What are the chances of success, and how good would success be? [Tractability and Expected Value]

Impact: How many benefit and by how much?

It’s often not clear what a charity will do with your donation. For example, the Salvation Army runs many programmes, but they don’t tell you how much any of their programmes cost and therefore what your donation can achieve. This is so common we don’t think much of it. But it would be bizarre if this also applied to our regular spending. It would be like paying some amount of money at a shop and getting a selection of goods chosen by the shop assistant.

MacAskill suggests that the first step in deciding how to allocate our resources is to consider:

  • The cost of the activity (in time or money),
  • The number of people it benefits, [or “beings”, if we also care about animals] and
  • How much it improves people’s lives. One way of measuring this is with QALYs (see box).

Quality-Adjusted Life Years (QALYs)

Economists have developed a metric called the QALY to compare different health programmes. All else equal, we should spend resources in ways that will produce the most QALYs.

A QALY is measured by comparing how many years of life people are willing to give up to preserve a higher quality of life. For example, surveys show that people generally rate a life with untreated AIDS as 50% as good as a life without AIDS. A QALY therefore suggests you can improve a person’s life in two ways: by lengthening it or by improving their quality of life.

QALYs do have some flaws. One is that patients tend to rate their medical conditions as less bad than people without those conditions. For example, people who have never been on dialysis estimate that their quality of life on dialysis would be 39% of what it is at full health. But people actually on dialysis rate their quality of life at 56% of full health.

However, we don’t always need precise estimates to compare programmes. Even rough numbers are often adequate, as many programmes dramatically differ in effectiveness.

Wellbeing comparisons

In principle, the idea behind QALYs can be extended to other things that affect wellbeing. This helps us to compare very different outcomes that otherwise seem incomparable. For example, doubling someone’s income increases reported subjective wellbeing by 5%. So doubling someone’s income for 20 years would provide one wellbeing-adjusted life year (WALY).

MacAskill points out that not everyone shares this view. Ken Berger, a former CEO of Charity Navigator, thought it was impossible to weigh ones person’s interests against another’s. MacAskill argues that this can’t be correct. We should be able to say that giving someone an extra dessert provides less benefit than saving someone’s life. We should also be able to say that saving a million lives is better than saving one. While it may be difficult emotionally and practically to make such comparisons, it isn’t impossible.

Is this the most effective thing you can do?

The most effective causes can be much, much more effective than other good-sounding (but less effective) causes. The distribution is “fat-tailed”.

Some things, like height, have normal distributions. Most people are familiar with these. But other things, like income, have fat-tailed distributions. Fat-tailed distributions are unintuitive. They are marked by extreme events. The 80/20 “rule” is an example of a fat-tailed distribution. The ratio won’t always be 80/20, but the idea that most of the value is generated by the very best activities is very common.

Compare several measures aimed at increasing education in developing countries. For every $1,000 spent:

  • Providing cash rewards to children in school yields 0.2 years of schooling;
  • Providing free primary school uniforms yields 7.1 years; and
  • Deworming schoolchildren yields a whopping 139 years.

Since doing good is “fat-tailed”, we should make our decisions thoughtfully if we want to maximise our impact.

Is this area neglected?

Think at the margin

When deciding what to do, we should think about marginal value, rather than average value.

The Water-Diamonds Paradox

There is a famous water and diamonds paradox in economics that illustrates the importance of thinking at the margin.

Water is arguably more valuable than diamonds, because water is necessary for survival. But the marginal value of water is very low, because there’s already a lot of water in the world.

In contrast, diamonds are relatively scarce. So the marginal value of a diamond is high, even though diamonds aren’t necessary like water is.

The law of diminishing returns

Most things have diminishing returns. The law of diminishing returns provides a useful rule of thumb for comparing causes. If an area has already gotten lots of attention and funding (e.g. disaster relief, see further below), it’s less likely that we can do a lot of good in that area.

The law of diminishing returns also suggests you should generally focus your efforts in poorer countries. For example, it costs about $50,000 to train a guide dog for one blind person. In contrast, $50,000 is enough to cure 500 people of blindness in the developing world, if spent on surgery to prevent blindness caused by trachoma. Highly cost-effective health programmes will usually be funded already in rich countries (i.e. no low-hanging fruit left). The same is not true in poor countries.

Example: How many lives could you save as a doctor?

Greg Lewis, a medical student, tried to find out how many lives a doctor is likely to save over their career. An epidemiologist, John Bunker, estimated that the total benefits from medicine in the US were about 2.2 billion QALYs. Divided by the number of US doctors (878,194), this worked out to about 2,500 QALYs per doctor. This amounts to saving about 70 lives over one doctor’s career. Adjusted for the impact of nurses, hospital administrators etc, the real figure would likely be lower.

Greg realised that this figure was an estimate of the average value of a doctor, not the marginal value of one. If there were already 878,194 doctors in the US, what was the marginal value of being the next one?

Using statistics, Greg compared the quality of health in different countries with the number of doctors in each of those countries (controlling for other factors like wealth and education). He concluded that adding one doctor to the US saves about 4 additional lives over a 40-year career. In contrast, a doctor in, say, Ethiopia, would likely save about 300 additional lives over their career.

Disaster relief

Disaster relief donations are generally less effective than donations to the most effective charities because of the law of diminishing returns.

The amount of funding disaster relief gets depends on how evocative it is and how much publicity it gets, not on how severe the problem is. Disasters are new and dramatic, and cause urgent emotions. But every day, over 18,000 children die of preventable diseases like AIDS, malaria or tuberculosis. We grow accustomed to everyday emergencies like disease and poverty, and forget about how serious they are.

Example: Earthquakes in Japan, Haiti and China

Compare the 2010 earthquake in Haiti and the 2011 earthquake in Tohoku, Japan.

Both earthquakes received a lot of international media attention and aid (around $5 billion in both).

But the Japanese earthquake killed around 15,000 people while the Haiti earthquake killed around 150,000. [This seems to be a wider issue of Scope Insensitivity – our brains simply aren’t wired to differentiate between big numbers beyond a certain point.]

Moreover, Japan was the fourth richest country in the world and already had resources to deal with the disaster. Haiti did not. As a country, Japan was about 1,000 times richer than Haiti. In fact, four days after the Japanese earthquake, the Japanese Red Cross issued a statement saying that they did not require external assistance and were not seeking funding from donors.

The 2008 earthquake in Sichuan, China provide another example. It killed 87,000 people but wasn’t as publicised as the Japanese or Haiti earthquakes. The amount raised in international aid was therefore only $500 million.

What would have happened otherwise?

MacAskill stresses the importance of considering the counterfactual. The measure of how much good you do is the difference between what happens as a result of your actions and what would have happened anyway.

For example, Viktor Zhdanov, a former deputy minister of health for the Soviet Union, might have done the most good in the world. Zhdanov played a key role in getting the World Health Organization (WHO) to start a (successful) smallpox eradication campaign. MacAskill acknowledges that even if Zhdanov had not lobbied the WHO, smallpox would probably have been eradicated eventually. But even if Zhdanov only brought forward the campaign by 10 years, he probably prevented between 10 to 20 million deaths.

Counterfactual thinking is also important when you think about how much good you can do in your career. For example, if you become a doctor, you’re probably not adding one doctor to the supply of doctor. There are usually a fixed number of places at medical school so the difference you make is just the difference between how many lives you save vs the lives saved by the next person that could’ve become a doctor if you hadn’t.

Example: Lives saved as a doctor (continued)

Greg from the earlier example estimated that once the counterfactual is taken into account, the number of additional lives he could save over the course of his career as a doctor is not 4, but just 1 or 2.

He still ended up becoming a doctor, but decided to focus on earning to give instead. While he accepted he wouldn’t save a lot of lives directly, he could save hundreds through his donations. If he didn’t become a doctor, the next person who would take his place would probably donate very little – the average being about 2%.

Example: Scared Straight and the importance of counterfactual thinking

Scared Straight was a programme intended to prevent juvenile reoffending, by showing kids how bad prison was. The producers claimed the program worked, as reoffending declined for kids who’d been through the program.

But higher quality studies showed that reoffending declined for kids who hadn’t been through the program, too. In fact, reoffending declined more without the program. They estimated that $1 spent on Scared Straight created costs of $203 to society! So when the counterfactual is considered, it suggests Scared Straight was an outright harmful program.

What are the chances of success, and how good would success be?

Some causes may be worth supporting even if their effectiveness is not guaranteed – or not even likely. This is because their expected value could be very high. That is, if it does succeed, its positive impact could be large enough to make the risk worth it. We should generally try to maximise expected value, rather than maximise the chances of success.

Philanthropy is different from ordinary, personal decisions. When making a personal decision, it may be rational to be risk averse because of diminishing returns. For example, it may be rational to prefer getting $1m for certain than a 10% chance of $11m, even though the expected value of the latter is higher. With philanthropy however, returns diminish very slowly as the problems in the world are so big.

Thinking in expected value terms is important because most of us are terrible at assessing low-probability high-value events. People tend to overweight them (thinking they are much more likely than they are) or ignore them entirely (treating them as zero).

Example: Expected value of going into politics

Laura was a student of philosophy, politics and economics (PPE) at Oxford. She wanted to find out the expected value of pursuing party politics.

Laura estimated that a PPE graduate from Oxford had a historical chance of about 1 in 30 of becoming an MP and 1 in 3,000 of becoming prime minister.

Calculating the impact she would have as an MP was more difficult, so they generally adopted conservative assumptions:

  • Total UK government spending was £723 billion in 2014-15.
  • They estimated that MPs (combined) could influence only 1/16th of that spending, as MPs are constrained by other actors such as politics, international bodies, and civil servants. This gives us £45 billion.
  • They then estimated that non-Cabinet MPs would influence about half of that £45 billion, with Cabinet MPs (including the Prime Minister) influencing the other half. Oxford PPE graduates made up 5% of non-Cabinet MPs and 32% of Cabinet ministers, so the calculation is 5% x £22.5 billion + 32% x £22.5 billion = £8 billion annually.
  • Each year, Oxford produces 200 PPE graduates. Around 50 of them pursue a career in party politics. The expected influence of each graduate is therefore £8 billion / 50 = £160 million.
  • Assuming (conservatively) that the money each graduate can influence will only do 2% as much good as money donated directly to the most effective causes, you get £160 million x 2% = £8 million.

Even with these conservative assumptions, £8 million seemed to be considerably more than Laura could expect to donate if she pursued earning to give. And even though Laura was still more likely to fail than to succeed, the expected value of going into politics was high enough for her to pursue.

Because of expected value, MacAskill argues that gains from systemic change may be even greater than donating to charities that show reliable and measurable results. Even if we can’t get exact figures to show the expected value of attempting systemic change, rough estimates will often be enough to give you an indication of how great the expected value is. [One of the most frequent criticisms of EA is that it neglects systemic change. But EAs have responded by saying they actually love systemic change, they just had to start somewhere and focusing on areas where empirical evidence was the strongest was a natural place to start.]

Climate Change and Expected Value

Expected value is useful when looking at climate change in three ways.

  • First, even if it’s unclear whether climate change is happening, the consequences of not preparing for it if it happens are so much worse than the consequences of preparing for it if it doesn’t happen. [This is like Pascal’s wager. A problem arises if a sceptic doesn’t accept the probability of the disastrous event as being high enough to outweigh the disastrous consequences if the event did end up happening. So it can only persuade a climate change denier who accepts there is at least a risk of climate change.]
  • The second reason expected value is useful is in considering what individual actions you might take. Your individual greenhouse gas contributions will only increase the Earth’s temperature by about half a billionth of a degree Celsius. But on occasion, that half a billionth of a degree will cause a flood or heatwave that wouldn’t have happened otherwise, so the expected harm of your contributions could still be fairly high. [Not too persuaded by this, at least without further information. If the probability is low enough, the expected value of harm could still be low even if the consequences from a flood or heatwave are great.]
  • Lastly, expected value is useful when assessing just how bad climate change is, and how much we should be willing to pay to address it.

    Economists tend to assess climate change as costing only around 2% of global GDP, which doesn’t sound very bad. Similarly, the social cost of one metric ton of carbon dioxide equivalent is estimated at about $32. The average American emits about 21 metric tons of CDE, about $670 per year. But this is only considering the effects from the most likely scenario: 2-4°C rise in temperature.

    There’s a small but significant risk of a greater increase. The Intergovernmental Panel on Climate Change (IPCC) estimates a 5% probability of temperature rises over 6°C, and acknowledges a small risk of temperature rises over 10°C. If these costs are also taken into account (and risk-weighted), the true social costs of carbon dioxide equivalent could be much higher than $32 per metric ton.
Small things that could make a big difference

MacAskill gives examples of small actions that could make a big difference, even if they generally won’t. The examples given are voting in elections and not buying meat in a supermarket. (Participating in a political rally is a third one, which I haven’t covered here.)

Voting in an election won’t make a difference in the vast majority of cases. But on the slim chance that it does, it could have an enormous impact and its expected value could still therefore be very high.

  • Nate Silver has calculated the odds of an individual vote swaying the outcome of the 2008 presidential election is about 1 in 60 million.
  • However, if Party A being in power would benefit each American by $1,000 more than Party B would, that equates to a total benefit of around $314 billion. Even dividing by 60 million, that amounts of $5,200 of value per vote.

MacAskill does caveat that the impact of voting varies depending on where you live and how much the different candidates vary in impact. If you have no idea which party will be better then your expected value of voting drops to zero.

Not buying meat in the supermarket usually won’t make any difference to the amount of meat the supermarket buys from suppliers (and therefore the number of animals killed). On very rare occasions, however, you could make a difference if your decision not to buy meat causes the supermarket to decrease its order – e.g. they may follow a rule where if monthly sales fall below 5,000, they decrease stock by 20%. Economists have found that giving up one egg causes total egg production to fall by 0.91 eggs; giving up one pound of beef causes beef production to drop by 0.68 pounds.

How to evaluate charities

Overheads and CEO pay are irrelevant

  • Things like CEO pay, or amounts spent on overhead, are irrelevant when comparing different causes. A charity may be much more effective than another, even if it pays its CEO more and spends more on overhead.
    • If you’re deciding whether to buy a Mac or PC, you don’t care about how much Apple and Microsoft spend on overhead, or how much their CEOs are paid. You only care about how good (effective) their product or service is. That same principle should apply even when we buying products or services for other people.
  • In a footnote, MacAskill suggests that one aspect of the “overheads” metric that may be important to keep low is fundraising costs. Some fundraising may “grow the pie” by creating new philanthropic money, but some fundraising just cannibalises another charity’s funding.

Example: Books for Africa vs Development Media International

Books for Africa (BFA) tries to improve education by shipping donated books from the US to Africa. BFA’s overhead costs are only 0.8% of their total costs, and their CEO is only paid around $110,000. They look like a very good charity if you judge it based on overhead and CEO pay.

In contrast, Development Media International (DMI) produces and broadcasts radio and TV programmes to encourage breastfeeding, handwashing, and use of antimalarial bed nets. DMI’s overheads are around 44% of its total budget. (Apparently the definition of “overhead” can be quite arbitrary, and DMI’s overhead can vary from 16% to 44%. The 44% figure includes all headquarters expenses, even if some of those staff work directly on project delivery from headquarters.)

However, economists have found that providing textbooks has virtually no effect on children’s school performance. On the other hand, DMI seems to be very impactful. Diarrhoea kills around 760,000 children every year, and many of those deaths could be prevented by simple hand-washing. Studies estimate that mass media education can provide one QALY for about $10, suggesting you could “save a life” for $360.

Five questions to ask

MacAskill suggests asking 5 questions (based on GiveWell’s criteria) when deciding where to donate:

  1. What does this charity do? Often this is not what people expect. For example, people often think medical charities spend most of their funds on research but the American Cancer Society spends only 22% of its funds on research. The Society spends the rest on patient support (43%), prevention (21%), and detection/treatment (14%).
  2. How cost-effective is each programme area? Find out how much the charity spends per person to run its programme.
  3. How robust is the evidence behind each programme? MacAskill suggests we should generally prefer strong evidence of being fairly cost-effective to weak evidence of being very cost-effective. Many programmes don’t work, and it can be hard to distinguish them from programmes that do.
  4. How well is each programme implemented? For example, distributing antimalarial bed nets is extremely cost-effective, but only if implemented correctly with some education. Otherwise, people might use the bed nets for other purposes (e.g. fishing), not understanding their antimalarial benefits.
  5. Does the charity need additional funds? Many very effective programmes are fully funded. It can also be difficult for a specific charity to scale up rapidly. When GiveWell named the Against Malaria Foundation as its top charity in 2012, it received a surge in donations. It struggled to spend that money effectively, so GiveWell didn’t list them in 2013. (They later increased their capacity, so GiveWell listed them again in 2014.)

Assessing GiveDirectly vs Development Media International

The evidence in favour of direct cash transfers (GiveDirectly) is very robust, and the programme is so simple, so we should consider it as a benchmark. We should only donate to charities that outperform this benchmark. In other words, we should only assume we’re in a better position to help the poor than they are to help themselves if we have some compelling reason to do so.

In the case of DMI, there is a compelling reason. Mass media health education cannot be “bought” by an individual. Even if they could, they wouldn’t know how valuable it is. [In other words, the market will not supply it efficiently because it’s a public good and there is information asymmetry.]

However, while there is evidence to suggest that mass-media education could be much more effective than direct cash transfers, the evidence for mass-media education is a lot weaker than for direct cash transfers. MacAskill explains that he deliberately chose DMI and GiveDirectly as examples because it’s unclear which one will do more good with your donation. Both appear to be excellent charities.

Examples of charities

Some top charities

In addition to DMI and GiveDirectly, MacAskill quickly runs some other top charities (as of January 2015, anyway) through his 5 questions:

  • Deworm the World Initiative, which provides technical assistance to governments to help run deworming programmes. Deworming has proven to be one of the most effective ways to improve educational outcomes for children by reducing absenteeism. It also provides other economic and health benefits.
  • Schistosomiasis Control Initiative, which provides funding for governments to run deworming programmes. It used to focus on schistosomiasis but now treats other parasitic worm infections also.
  • Against Malaria Foundation, which provides funding to buy and distribute insecticide treated bed nets to poor households in sub-Saharan Africa.
  • Living Goods, which runs a network of people in Uganda who go door-to-door selling affordable health products such as soap, menstrual pads, contraception, and malaria, diarrhoea and pneumonia treatments. They also provide healthcare advice.
  • Iodine Global Network, which advocates for governments to fortify salt with iodine and monitors progress of programmes.

MacAskill notes that there are no “megacharities” like Oxfam, Unicef or World Vision on the list for several reasons. First, they often run a variety of programmes, so are hard to evaluate. Moreover, many of these charities spend a lot on disaster relief but for reasons explained above, that’s not generally the most effective use of funds.

He also notes that most of the charities that he lists focus on health programmes, rather than education, water provision or economic empowerment. While these are also promising areas, global health stands out because:

  • It has a proven track record (e.g. smallpox eradication, and also developments in reducing polio, measles, diarrheal and guinea worm disease)
  • The evidence is more robust. Human bodies are pretty similar throughout the world, so a drug that kills a parasite in one country will probably do the same in another. Whereas an educational programme that works in one country may not work as well in another.
Some not so great charities

Example: The PlayPump

The PlayPump is an example of a good sounding idea that didn’t work out in practice. The idea was to use a children’s roundabout also as a water pump. People in poor countries could get fresh water and children could play on it at the same time.

Unfortunately, the physics of it didn’t work. Playground roundabouts are meant to spin freely once they’ve gained enough momentum – that’s the fun part. But to pump water, a PlayPump needed constant force, which isn’t fun at all.

In one village, children were paid to “play” on the pump. Most of the time, women in the village just ended up pushing the PlayPump themselves to pump water – which was about 1/5th as efficient as the handpumps they had previously used.

Example: Microcredit

Microcredit involves lending small amounts to the very poor.

High quality studies have found that microcredit programmes had little or no effect on income, consumption, health or education. People generally used these loans for extra consumption (e.g. food or healthcare) rather than investment like starting a new business. The latest evidence suggests a small positive effect but not nearly as large as initial anecdotes suggested.

Note, however that there are other forms of microfinance may be more effective. For example, “microsavings” involves providing secure places for poor people to save money and has shown promise.

How to evaluate causes

Scale-Neglectedness-Tractability framework

MacAskill suggests a framework to compare causes by assessing them against 3 (or 4) dimensions:

  • Scale. This is the size of the problem, measured in terms of total impact on others’ wellbeing.
  • Neglectedness. This dimension is important because of diminishing returns. Causes that get less attention tend to be those where you can make a larger impact.
  • Tractability. We should try to estimate the long-run tractability of the cause, as estimated cost-effectiveness of specific programmes within a cause area will change over time.
  • (Personal fit. This is relevant if thinking of contributions of time rather than money and is discussed further below under “Using your career to make a difference”.)

Causes generally won’t score highly on all three dimensions, so choosing a cause to focus on will involve judgement calls. Reasonable people may disagree on how to weigh the criteria against each other.

Examples of high priority causes

MacAskill also lists some promising organisations that are working in each of these areas, if people want to investigate further.

US criminal justice reform

Normally opportunities to make a massive difference are in developing countries, but MacAskill thinks US criminal justice reform is unusually tractable and neglected, and is still fairly large in scale.

  • Scale – fairly large. US has one of the highest incarceration rates in the world, at 0.7% of its population (2.2 million people). Criminologists think that we could reduce incarceration rates by 10% or more (200,000 people) while keeping criminality levels the same, or even lower.
  • Neglectedness – fairly neglected. Non-governmental organisations only spend around $20 million per year on prison reform to substantially reduce incarceration rates.
  • Tractability – extremely tractable. There appears to be an unusual level of bipartisan support for prison reform.
International labour mobility

Increasing migration from poor to rich countries would provide substantial benefits for the poorest people in the world.

  • Scale – very large. Estimates say around 600 million people worldwide would migrate if they could. There would also be indirect benefits as migrants send remittances home to those who don’t migrate.
  • Neglectedness – very neglected. Very few organisations focus on relaxing migration policy. More focus is given to migrants already in the US.
  • Tractability – not very tractable. Increased levels of immigration are very unpopular in developed countries.
Factory farming

Relatively small changes to farming practices could substantially improve the animal welfare in factory farms.

  • Scale – up to very large, depending on how much weight you put on the interests of animals. Around 50 billion animals are raised and killed in factory farms each year.
  • Neglectedness – extremely neglected. Non-profits spend less than $20 million each year on factory farming practices.
  • Tractability – fairly tractable. Rates of meat consumption are decreasing and in the EU there are moves to improve conditions in factory farms.
2-4°C climate change
  • Scale – fairly large. Economic assessments tend to focus on human costs. If you also value the natural environment, the cost will be higher.
  • Neglectedness – not very neglected.
  • Tractability – fairly tractable. There are reliable ways for individuals to reduce carbon emissions, but the opportunity to effect political change is unclear and progress has been slow. [I’m surprised he thinks this is fairly tractable. The political side of things doesn’t seem very tractable but I don’t follow it closely. I was also surprised MacAskill said there are reliable ways for individuals to cut their carbon emissions. Earlier in the book he says that most ways don’t do much, and it would usually be more effective to just donate to Cool Earth. So there appears to be an inconsistency here, unless I’m missing something.]
Catastrophic climate change
  • Scale – fairly or extremely large, depending on how much you value preserving humanity into the far future. [Even if you don’t place much value on preserving humanity into the far future, I think this should be “extremely” large. Particularly as he put “very large” for International Labour Mobility and “fairly large” for US criminal justice reform.]
  • Neglectedness – fairly neglected. Most focus on climate change is on reducing emissions. There is not much research on catastrophic climate change.
  • Tractability – fairly tractable. More research could be done on geoengineering, which could be used as a last resort. Geoengineering is an attempt to cool the planet by pumping sulphates into the stratosphere. There are significant risks to this, but it may be worth it if the temperature increases are bad enough. A single country could also unilaterally undertake geoengineering so it would be good to have a better understanding of its risks beforehand.
Other global catastrophic risks

For example, nuclear war, pandemics and bioterrorism are low-probability risks that could have catastrophic consequences.

  • Scale – fairly large to extremely large, again depending on value judgments about preserving the future of humanity.
  • Neglectedness – fairly neglected. The amount of funding on these issues is relatively small ($30m per year on nuclear security, a few million per year on biosecurity, and $1-2 million on global catastrophic risks). However, there is some funding from government at least on nuclear security and biosecurity.
  • Tractability – fairly tractable. There are opportunities for funding more academic research and increasing policy influence. [When MacAskill says something is “fairly tractable”, I think he’s just saying there are things we could do to make headway into the problem, rather than “solve” it entirely.]

Ethical consumerism

Generally, changing consumption habits is not a very effective way of doing good as it’s not very targeted.

There’s even some reason to believe that the rise in ethical consumerism can be doing more harm than good, due to moral licensing. Studies have found people who have done one good thing tend to do fewer good things in the future. Even just saying you’d do something good can create a moral licensing effect.

Moral licensing may not be such a problem if people do good effectively and compensate by doing some less-good stuff. But it can be a big problem if people do some fairly ineffective acts of altruism, and compensate by doing less effective ones (like donating).

Sweatshops

Sweatshops are basically factories in poor countries that produce goods for rich countries under pretty terrible working conditions.

However, while those working conditions are terrible by developed country standards, the alternatives for the workers can be even worse (e.g. farm labour, scavenging, unemployment). In the developing world, there is great demand for sweatshop jobs. [There’s a more recent study which has cast doubt on this. The study found that most people who got a job at a sweatshop ended up quitting within a few months. Only about a third remained after a year. Those who quit to work in agriculture and market selling generally earned about the same as in a factory, but factory work tended to be more hazardous.]

The Child Labor Deterrence Act was proposed in the US in 1993, which would have made it illegal to import goods from countries using child labour. Bangladesh factories responded by laying off 50,000 child workers. Instead of going to school or finding better jobs, UNICEF found that many children resorted to even more desperate measures such as prostitution or street hustling.

MacAskill accepts that we should feel outrage and horror at sweatshop working conditions. But boycotting sweatshops would just make poor people worse off. Instead, we should try to end the extreme poverty which makes workers want sweatshop jobs in the first place.

In a footnote, MacAskill notes that his arguments don’t apply to forced labour as there is no argument that non-voluntary work benefits the workers.

Fairtrade

Fairtrade certification is an attempt to give higher pay to workers in poor countries. Producers are guaranteed a minimum price for their goods, and are paid a “social premium” on top of the market rate. The social premium is then used to fund community programmes.

The Fairtrade licence is only given to producers who meet certain criteria, such as paying workers a minimum wage and meeting safety requirements. But the standards are difficult to meet, so those in the poorest countries usually can’t get it. Most Fairtrade coffee comes from Mexico and Costa Rica, which are relatively rich compared to, say, Ethiopia.

  • Of the additional money spent on Fairtrade, most is taken by middlemen and doesn’t make it to the producers.
  • Dr Peter Griffiths, an economic consultant for the World Bank, found that for one British coffee chain, less than 1% of the Fairtrade premium reached coffee exporters in poor countries.
  • In Finland, Valkila, Haaparanta and Niemi found that only 11% of the Fairtrade premium reached the coffee-producing countries.
  • Bernard Kilian and colleagues found that in the US, coffee producers only received about 8% of the Fairtrade premium.

Even the small percentage of additional money that reaches producers doesn’t necessarily translate into higher wages. The Fairtrade-certified organisations get the higher prices, not necessarily their workers. The evidence is limited, but the consistent finding is that a Fairtrade certification doesn’t improve workers’ lives.

  • Christopher Cramer and others conducted a four-year study looking at Fairtrade workers in Ethiopia and Uganda. They found those workers were systematically paid less and had worse working conditions than comparable non-Fairtrade workers.
  • Even a review commissioned by the Fairtrade Foundation concluded “there is limited evidence of impact on workers of participation in Fairtrade”.

So to summarise, you’d do more good buying cheaper goods and donating the money saved to a cost-effective charity.

Carbon footprints

Many popular ways of reducing greenhouse gas emissions are ineffective:

  • For example, shutting off electronic devices or turning off lights when you’re not using them does eff-all. One hot bath adds more to your carbon footprint than leaving your phone charger plugged in for a whole year.
  • Buying locally produced goods also doesn’t do that much. 10% of the carbon footprint of food comes from transportation while 80% comes from production. Sometimes locally produced food can even have a higher carbon footprint. Tomatoes grown in northern Europe had a 5x carbon footprint than tomatoes grown in Spain because of the extra heating and lighting required.

Some people have expressed concerns about the effectiveness of offsetting. Giving What We Can (GWWC) looked into more than 100 organisations that claimed to offset to try and figure out which ones were most cost effective. They thought Cool Earth was best (see the below box).

Example: Cool Earth carbon offsets

Cool Earth claims it costs them about $100 to prevent an acre of rainforest being cut down, and each acre locks in 260 metric tons of carbon dioxide. That equates to just 38 cents per metric ton of carbon dioxide. GWWC applied some more conservative estimates, reaching $1.34 to offset one metric ton.

Being extra safe and assuming a 300% margin of error (so $5 per metric ton), they worked out the average American would only need to spend $105 per year to offset all carbon emissions. MacAskill suggests this may be a more effective and easy way to reduce your carbon footprint than changing your lifestyle. For example, the carbon footprint reduction from completely cutting meat from your diet is about the same (or less) than donating $5 a week to Cool Earth. [Sounds good – perhaps too good – but I’d be concerned about how scalable Cool Earth’s measures are.]

Going vegetarian for animal welfare reasons

The animal welfare argument is much stronger for some animals than others. You can remove most of the animal suffering from your diet by giving up chicken and eggs. This is because their living conditions are particularly bad, and you need more animals to get the same number of calories (compared to pigs, cows, etc).

Bailey Norwood, an economist and agricultural expert, rated the welfare of different animals on a scale of -10 to 10, where a negative number suggests the animal would prefer to be dead than alive.

  • Beef cattle: 6
  • Dairy cows: 4
  • Broiler chickens: -1
  • Pigs and caged hens: both -5.

The most effective way to cut animal suffering out of your diet is to stop eating chickens, then eggs, then pork. This conflicts with what most animal welfare and environmental activists advocate, which is to cut red meat from your diet.

One crucial difference between greenhouse gas emissions and meat consumption is the fact that you can prevent harm if you successful offset greenhouse gas emissions. You cannot do the same with meat consumption. Some animal charity evaluators suggest it costs about $100 to convince one person to stop eating meat for one year. But MacAskill thinks convincing someone else to give up meat doesn’t stop animals being harmed – it just changes which animals are harmed. (MacAskill also points out in a footnote that this $100 figure is also less reliable than evidence on other programmes discussed in his book.)

Using your career to make a difference

Your choice of career can determine how you spend over 80,000 hours in the course of your life (hence the organisation 80,000 Hours) . Spending just 1% of that time thinking about it (800 hours) may be time well spent. The following 3 questions may be useful in deciding your next career steps:

  • How do I personally fit with this job?
  • What’s my impact working at this job?
  • How does this job contribute to my impact later in life?

For people thinking about making moves later on in their career, the same framework applies but future career capital becomes a lot less important and your existing career capital becomes more important. If your skills don’t transfer well to making a difference, earning to give can still be a good option. If your skills do transfer, then it may be better to contribute directly to an effective area. For example, Rob Mather of the Against Malaria Foundation had a background in business and sales, so he already understood how to run an organisation, pitch ideas and make things happen. He also didn’t need to take a salary, which helped impress donors in early stages.

Personal fit

If you’re not happy, you’ll be less productive and more likely to burn out, so you’ll have less impact in the longer term. But this doesn’t mean you should “follow your passion” or your “gut”, for several reasons:

  • Most people don’t have passions that fit the world of work, fields that people tend to be passionate about (e.g. sports, entertainment) tend to be very competitive.
  • Secondly, interests change over time, usually more than we expect. For example, professors tend to be much less happy after getting tenure than they expected beforehand. One reason may be that they focused on the positive features of getting tenure and neglected the downsides. This is a bit like Mark Manson’s point about choosing the pain in your life, not the benefits.
  • Lastly, the best predictors of job satisfaction are features of the job itself rather than personal passion – independence, sense of completion, variety, feedback from the job, and contribution. These 5 factors, known in psychology as “job characteristics theory” correlate not just with job satisfaction, but also motivation, productivity and commitment to your employer. They are also similar to the factors required to develop “Flow”.

Since it’s hard to predict where you’ll be most satisfied and perform best, it’s best to take an empirical approach and try out different things at the start of your career. You can also try to learn as much as you can about career path by talking to people in that career, and asking about the most important traits for success, the main reasons people leave, etc.

Impact

Maximising your impact doesn’t necessarily mean taking a job in the social sector for several reasons.

  • First, many social sector organisations are not very effective.
  • Second, if you consider the counterfactual, you need to provide more value over than the next person the charity would have hired.
  • Third, earning to give may be a more effective way of making a difference. There are also other ways if making a difference (e.g. journalism, research, entrepreneurship, politics) that are not in the social sector that may be more effective.
  • Fourth, if you’re just starting out, it may be much more important to build skills and career capital than it is to have a direct impact on the job. Senior people have a disproportionate amount of influence and impact, so maximising your chance of getting into those positions is a key part of maximising your impact (and that may not mean taking a job in the social sector). Building career capital may also be good if you’re not sure which causes to support yet.

Career capital

Career capital refers to the skills, network and credentials that will help you take a higher-impact job later. In deciding on your initial jobs, think about:

  • How much does it keep your options open? For example, it’s easier to move from the for-profit sector to the non-profit sector.
  • How much do you learn from it? Do you build transferable or highly specific skills? Will you learn a lot about other possible careers that will help you with later decisions?

People often think of choosing a career as a once-off decision that you can’t change. But as life progresses, you’ll get new information and unexpected opportunities and problems will arise. You should think of your career as a work in progress, with a model that you constantly revise as you get new information instead of a fixed “plan”. You can also try to find out where you’re uncertain, and take steps to reduce that uncertainty. [Sounds a bit like How to Decide by Annie Duke.]

Examples of career options

Solid bets

MacAskill suggests the following career options are “solid bets”, where one is very likely to make a positive impact:

Direct work for a highly effective organisation

These may or may not be non-profits. Companies could also be effective if they are benefiting lots of people.

MacAskill says that they don’t often recommend people go into non-profit work straight out of university, as for-profit companies usually have more resources to invest in staff training, etc.

But this can still be a good option for some people. If you worry your values might wane with a different role, or if you find it important to be around like-minded people to be inspired, this could be the best option for you.

Earning to give

Earning to give exploits the fact that even average workers in developed countries are among the top income earners in the world, and some charities can do a lot of good for the world’s poorest with relatively little money. Another advantage of earning to give is that it’s a career available to everyone. It can also allow you to build skills and networks that will be valuable later on.

It’s important to think of the long term. The highest paying careers (e.g. front-office finance, consulting) are extremely competitive and also come with a high chance of dropping out. Tech entrepreneurship and quantitative trading offer even higher earnings but entrepreneurship comes with higher risks. Other options discussed briefly are medicine, law, software engineering, sales and marketing, accounting, skilled trades, public services, air traffic control.

MacAskill also suggests giving some thought to whether a job will be around in the future. Jobs that require social skills, precise perception and manipulation, physical proximity, or high levels of training are less likely to become automated.

One risk with earning to give is that you might lose your values if the people surrounding you aren’t altruistically inclined. But this risk may not be too bad since you can always leave and find another job if you find that starting to happen. You can also guard against this by involving yourself in the effective altruism community and publicly stating your intentions to donate. MacAskill points out that there are many examples of people who have successfully pursued earning to give without losing their values.

Skill-building

In the short term, focusing on building your skills may be a good option if you aren’t sure about what you want to do in the longer term.

Jobs which are likely to be good for general skill-building include consultancy, sales and marketing, and getting a PhD in a useful area. Pursuing a PhD often offers more flexibility than a full-time job, so you can use that time to explore other things on the side.

High-potential long shots

These career options offer a small chance at making a very large impact.

Entrepreneurship

Entrepreneurship is extremely promising in terms of ability to effect massive change, build career capital and potentially making large profits to donate. It also has relatively low barriers to entry.

The downsides are that most start-ups fail, and entrepreneurship normally involves very long hours and lots of stress. Not everyone is cut out for it.

If you’re thinking of starting a non-profit, ask yourself why the markets, governments or philanthropists haven’t solved the problem yet. Usually the answers will suggest that the problem is very difficult to solve.

Research

The market tends to undersupply research because of externalities. Also, because many of the benefits of research occur only in the long-term.

The distribution of research achievements is heavily fat-tailed – a large proportion of scientific achievement comes from a very small number of scientists. So research may only be the best option if you might be able to excel in it.

In some fields, it’s really hard to get an academic job after your PhD. For example, it’s much harder in philosophy than in economics. [Probably because there are more non-academic jobs for economists than for philosophers.] 80,000 Hours therefore suggests that the greatest potential to do high-impact research while gaining career capital to keep options open are: economics, statistics, computer science, and some areas of psychology.

A good way to have impact within research is to combine fields. Research at the intersection of two disciplines is often neglected. Combining fields can be particularly useful when moving from a theoretical area to practical areas with real-world applications. For example, Daniel Kahneman and Amos Tversky invented the field of behavioural economics by combining psychology with economics. Effective altruism similarly made progress by combining moral philosophy with economics.

Example: Norman Borlaug’s research

Norman Borlaug worked in agricultural research. He invented a short-stem disease-resistant wheat which was able to radically increase crop yield in poor countries. Doesn’t sound particularly sexy.

But Borlaug was awarded the Nobel Peace Prize in 1970. The committee suggested he may have saved a billion lives with his invention.

Politics and advocacy

The potential impact from entering UK party politics was discussed above in the Laura example.

Advocacy also has potentially high payoffs by influencing potentially thousands of people. This could involve journalism or becoming a “public intellectual” after some time in academia. The distribution of impact is again highly fat-tailed.

Volunteering

Volunteering typically involves similar considerations to choosing a career. It is mainly useful for gaining skills and experience which will be more valuable later on.

But volunteering can sometimes be harmful to a charity. Volunteers often won’t be trained in the relevant area and will use up valuable management capacity. You should therefore only volunteer in ways that cost an organisation relatively little. Anecdotally, MacAskill has heard that some charities use volunteers mainly because the volunteers will donate back to the charity.

Instead of volunteering, you may be able to work extra hours at your normal job and donate the money instead.

Four ways to take action now

MacAskill suggests that if you feel inspired after reading the book, you should take some actions now instead of waiting around for that feeling to dissipate. He suggests four ideas:

  • Establish a habit of regular giving. This is the easiest and most tangible way of having a positive impact. Even if this isn’t the main way you intend to help others, it can be a good way to start by solidifying your intentions. You could even consider taking the Giving What We Can pledge to donate 10% of your income.
  • Write down how you’re going to incorporate effective altruism into your life. Make a note of specific steps you might take – e.g. if you’re going to start giving, note down how much and when you’ll start. If you’re going to change your career, set aside some dates to find out more information for your next steps.
  • Join the Effective Altruism community. You can join the mailing list on the Effective Altruism website.
  • Tell others about effective altruism. MacAskill acknowledges it can be awkward to raise effective altruism with others as it can come across holier-than-thou or critical of other causes. He suggests setting up a page on CauseVox and asking people for donations there instead of gifts for your birthday.

Other Interesting Points

  • In the 20th century alone, smallpox killed over 300 million people. For comparison, if we’d achieved world peace by 1973 (thereby avoiding the Khmer Rouge, Rwandan genocide, two Congo wars, 9/11, the Afghanistan and Iraq wars), we would have avoided merely 12 million deaths. During that time, the eradication of smallpox probably saved between 60 and 120 million lives.
  • If height distribution was similar to income distribution, we would regularly see people 270 feet tall (82 meters), towering over skyscrapers.
  • We can compare risks of death using a unit called a micromort. One micromort equates to a 1 in a million chance of death.
    • One ecstasy session is about 1 micromort.
    • Going skydiving is 9 micromorts.
    • Flying in a space shuttle is 17,000 micromorts.
    • Smoking a cigarette is 0.7 micromorts (a single cigarette won’t kill you but increases your chance of dying later).
    • Driving a car for an hour is about 0.1 micromorts, which equates to about 3 minutes of expected life lost for a 20-year-old. In contrast, an hour on a motorbike costs 3.75 hours of expected life lost.
  • Steve Jobs was passionate about Zen Buddhism when he was young, and only got into electronics reluctantly to earn cash on the side.

My Thoughts

I was already somewhat familiar with Effective Altruism (EA) before I read this book. I’d first heard about it years ago when EA focused more on global health measures (e.g. GiveWell’s recommended charities) and approaches like earning to give. More recently, I’d come across 80,000 hours and read a couple of their articles and listened to some podcasts. I hadn’t picked up MacAskilll’s book because I figured I knew the basic ideas and, honestly, there already seemed to be so much information in the EA community that I just didn’t see his book as a priority.

I finally read Doing Good Better because I signed up for an 8-week Intro to EA online course and several chapters were required reading for the course. (They also send you the book for free when you sign up for the course, though I bought the e-book because of the shipping time.) And I enjoyed reading the book, more than I thought I would. Even though most of the ideas in the book weren’t novel to me, it was still helpful to see them presented in such a coherent, ordered way. MacAskill is a very clear and compelling writer. He effectively uses examples to illustrate his points and the examples were entertaining.

Despite containing a lot of information, Doing Good Better is a reasonably short and easy read. I thought MacAskill did a good job explaining the ideas behind effective altruism in a balanced, non-judgemental way (although, as someone already sympathetic to that style of reasoning, I may have been pretty easy to win over). The core ideas are relatively simple, but the examples really bring them to life and make the arguments more compelling.

The book seems to be well-researched and referenced, and MacAskill caveats his claims as appropriate. One minor quibble is that I didn’t like the way MacAskill presented references (at the end of each chapter, without any footnotes in the text – at least in the e-book version). I prefer to read footnotes as I go along, so I found that style off-putting. But it’s a minor quibble in the grand scheme of things. Doing Good Better is definitely worth reading for anyone who wants to, well, do good better.

As mentioned above, you can get a free physical copy of Doing Good Better, as well as some other effective altruism books, here. If you want the audiobook or e-book, you can buy it at: Amazon | Kobo <– These are affiliate links, which means I may earn a small commission if you make a purchase.

What did you think of this summary for Doing Good Better? Disagree with my thoughts? Let me know what you think in the comments below.

If you liked this summary of Doing Good Better, you may also enjoy the following:

2 thoughts on “Book Summary: Doing Good Better by William MacAskill

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.