Why Aren't Effective Altruists Pro-Life?

Millions of QALYs are being left on the table.

Last month I asked a genuine question: why aren’t utilitarians pro-life? You should read the post for the full reasoning, but the argument boils down to this: in America newborns have an expected 78 years of life ahead of them. Given that, an abortion destroys an estimated 78 quality adjusted life years (QALYs) of utility. Since there are over 2 million families waiting to adopt newborns in America, how can the choice to have an abortion result in greater net utility? Do utilitarians expect the utility gained by the mother from having an abortion to outweigh 78 QALYs worth of utility?

I got a few responses both here and on ACX where I posed the same question. They mostly fell into three categories¹:

1. Repugnant conclusion! Many utilitarians do not believe we have a duty to maximize the number of humans that exist, even if it results in greater net utility.

2. The fetus is not a being of moral concern, so there is no utility lost from aborting it.

3. People who have abortions often have kids later, and probably would have ended up with the same number of kids either way so there is no lost utility.

I don’t find these arguments very compelling.

First, even if you reject the repugnant conclusion and believe that utilitarians have no duty to create as many potential lives as possible, the fetus is not a potential life. It is a currently existing life, and abortion is a choice to end it. I can respect an argument that a utilitarian has no duty to make choices that will maximize future utility, such as choosing to have a kid when he wouldn’t have otherwise; however, surely all forms of utilitarianism hold that you should not take actions that result in significantly less utility than would have occurred if you did nothing? Imagine the trolley problem but this time if you do nothing the trolley continues on safely, and if you hit the switch you kill a fetus and lose 78 years of future utility. What form of utilitarianism says it’s okay to hit the switch in that case?

In the case of the third argument, I would need to see evidence that each, or even most, abortions are replaced by kids later down the line, and that the new kids wouldn’t have existed if not for the abortion. I don’t think that’s plausible on the face of it, and even if it was true for some amount of abortions it certainly isn’t true for all of them (if a woman has four abortions, do we really expect her to raise four kids later?). Yet even if that was the case surely the utilitarian thing to do is to put the kids up for adoption instead of aborting them? There are millions of happy families waiting in line to adopt a kid and give them a good life, so if you have the first kid you keep his 78 QALYs, provide extra utility for the family adopting them, and you can still have the number of kids you would have anyway. The abortion option results in hugely less utility than the adoption option in that case!

However, today I want to focus on the second argument (that a fetus does not have enough moral status) because I think it’s deeply wrong in a way that seems to not be intuitive, and has a particular impact on the Effective Altruism movement.

Effective Altruism

Effective Altruism (EA) is a movement aimed around donating your money as efficiently as possible to “do the most good”. There are many EA organizations that work on identifying the most bang for your buck, though the most prominent is probably GiveWell. I mention EA’s because most EAs are utilitarians, and among all utilitarians they are particularly used to thinking practically in terms of actions you can take to increase utility from a moral perspective. According to the “EA Handbook” a single QALY may be worth around $39,000 on the low end. They use that estimate to judge whether interventions are worthwhile: if you spend $50,000 to save 1 QALY, you’re spending very inefficiently (especially as GiveWell believes there are current opportunities to save a life in Nigeria by only spending $3,000). So by their math the choice to abort destroys an expected $3,042,000 in value (78 QALYs x $39,000). Even if the abortion is in the first trimester, where there is about a 25% chance of miscarriage, that’s still $2,281,500 of lost utility ((78 x .75) x $39,000). In other words, you’d have to donate over 2 million dollars to break even on the choice to abort².

Some EAs have noticed this problem with utilitarianism and abortion before, though pro-life concerns do not seem to have caught on in EA spaces. One article that caught my eye was from Lesswrong.com, a “Rationalist” website where a lot of effective altruists post as well. The author of the post does some good old fashioned EA math on the question of abortion and comes up with the following chart:

Her argument is that if we are unsure whether the fetus has “moral significance” then there are four possible outcomes when deciding whether to abort or put the kid up for adoption. If the fetus has no moral significance then aborting it loses no QALYS, while if it does we lose 77.126 QALYS. If we choose adoption then regardless of whether it has moral significance we lose 0.276 QALYS worth of utility (which she calculates is the utility lost to carrying a pregnancy to term). Given this, it seems like adoption is the right choice for utilitarians if you think there is any real chance the fetus could have moral significance: even if you think there is only a 30% chance that the fetus is morally significant the math still says adoption is the right choice, as the author explains:

If she aborts the fetus, our expected QALYs are 70%x0 + 30%(-78.126) = -23.138

If she carries the baby to term and puts it up for adoption, our expected QALYs are 70%(-0.247) + 30%(-0.247) = -0.247

Which again suggests that the moral thing to do is to not abort the baby.

This is a pretty good argument but it has one glaring mistake in it: the same mistake many commenters have made when trying to answer my question. The mistake is thinking that it matters to the math at all whether the fetus has moral significance, when it simply doesn’t.

That chart up there should read -77.126 in both the “moral significance” and “no moral significance” columns because regardless of whether a fetus has moral significance we still expect it to develop into a being that does have moral significance, and produce 77.126 QALYs worth of utility. The utility we are arguing over is not utility that the fetus is currently experiencing so it doesn’t matter to the argument whether it has moral significance or not. The utility we will lose is utility that will happen to a being with moral significance, regardless of the fetus’s moral status now. With that in mind, the calculation is not an expected loss of -23.138 QALYs if you think there’s a 30% chance of the fetus having moral worth: instead it’s -77.126 QALYs lost in all scenarios, even if you think there is 0% chance that a fetus has moral worth.

Here’s an example of what I mean. Many utilitarians are vegans because they value animal utility to some extent: the animals are, to them, beings of moral significance. However, there are also many utilitarians who are not vegans and many of them don’t believe the animals we eat are beings of moral significance. Take fish for example: fish might not be able to feel pain, their brains are very primitive compared to our own, so you could make an argument that they have no moral significance and their utility should not be considered when trying to make correct moral choices. Yes, killing the fish results in it losing all it’s future utility and also suffering in the present, but it’s utility doesn’t matter to the calculation. Many utilitarians say that the fetus is the same as the fish, and that killing it results in no loss of utility from suffering or losing it’s future utility.

Yet almost all utilitarians think that adult human lives have moral value. If you kill a 30 year old you not only cause negative utility from any suffering experienced during death, but also negative utility from the lost years of life they would have experienced.³ An adult human has moral value, so future QALYs lost matter. Which is why EAs value saving lives, as it increases present and future utility.

So even if you think that the fetus is like a fish, and that the utility it experiences has no value, we still expect it to become a being with moral value and to experience utility. Even if you think that only adult human utility has value you’re still destroying 60 QALYs by aborting a fetus: life years that are expected to be experienced by a being with moral value. We can discount the fish’s future utility because it will be utility experienced by a fish, but whether or not you think a fetus is morally equivalent to a fish we all know it doesn’t stay that way. Even if you discount all it’s utility as a fetus, you’re still losing a lot of future utility by killing it! Imagine if there was a type of salmon that is like any other salmon in it’s juvenile phase but develops into a sentient mermaid as intelligent as any human after a year: would it make sense for utilitarians to treat killing that kind of salmon exactly the same as all the other types of fish we eat?

To sum up: I don’t see how an EA can be pro-choice and be consistent with their own moral values. You could make an argument that making abortion illegal wouldn’t lower the number of abortions, I suppose (though don’t think the data backs that up⁴). Yet even if you believe that you should still be pro-life in terms of seeking to minimize the number of abortions.

There are over half a million abortions every year in the United States. Even if you take a very cautious approach and only calculate losing 30 QALYs per abortion that’s more than 15,000,000 QALYs lost every year ($585 billion in equivalent utility). And that’s just the United States! This is a huge potential area for effective altruists to make a difference. The effective altruist Calumn Miller laid this out well in an excellent post on abortion as a potential field of EA intervention:

…if the pro-life arguments are correct, then it is arguably harder to find an issue that is more serious – at least for human beings in the near term – than abortion. But it is partly for this reason that it is so tractable: even tiny reductions in the abortion rate globally and domestically can save large numbers of lives. Much bigger reductions can be made by interventions which are popular even in strongly pro-choice countries, a fortiori in countries with significantly more pro-life sentiment. A prima facie assessment of the cost-effectiveness of pro-life advocacy therefore appears to highly commend it.

Now, I’m not a utilitarian. But I am sympathetic to utilitarians and I think the EA movement is a positive force for good in the world. Yet the EA community seems to be blind to this issue. What argument justifies EA ignoring abortion as a field of intervention, especially when they’ve taken seriously arguments to intervene in the case of chickens, fish, and even insects?

The EAs claim that their movement is underpinned by “impartial altruism” and “open truthseeking”. Yet in the case of abortion I would say that open truthseeking takes a backseat to cultural and political concerns. Most EAs are Democrats and live in blue cities in blue states. If they became pro-life it would cost them socially. Doing the math on farming mealworms, on the other hand, has no social costs. This kind of willful blindness would be more excusable in a movement that didn’t champion truthseeking and impartiality.

1. There was also one person who argued that every child who is aborted would have had a net negative utility existence so it’s better that they were killed in the womb. That argument is pretty crazy on the merits, but does at least have the virtue that if it were true it would be a good reason for a utilitarian to be pro-choice.

2. On average, anyway: if there really is a chance to save a life for $3,000 right now then you’d only have to donate around $7,600. That’s because EA’s consider a “life saved” by charity to be usually equivalent to around 30 QALY, averaged out.

3. If you don’t count future utility, then killing people painlessly and instantly would not be wrong. Yet I would guess that most utilitarians think that murder is wrong even if you do it instantly and painlessly.

4. Prior to legalization there were between 20,000-30,000 illegal abortions per year, compared to over 170,000 abortions seven years after legalization, in eastern Europe countries that made abortion legal and easier to access experienced more significant falls in birth rates than countries that didn’t (suggesting that there were more legal abortions after the laws changed than illegal abortions before the change), and research indicates that only 10-50% of women denied a legal abortion obtain an illegal abortion.