"Effective Altruism": A Contradiction in Terms

Thursday, June 14, 2018

The Economist recently carried an article on "effective altruism" [sic], an explicitly utilitarian fad. (If you wonder why I describe it so dismissively, keep reading.) I was thinking about commenting on it -- until I realized that Ayn Rand had covered the topic quite thoroughly in 1946:

The phrase "human sacrifice" is redundant. (Image via Pixabay)
"The greatest good for the greatest number" is one of the most vicious slogans ever foisted on humanity.

This slogan has no concrete, specific meaning. There is no way to interpret it benevolently, but a great many ways in which it can be used to justify the most vicious actions.

What is the definition of "the good" in this slogan? None, except: whatever is good for the greatest number. Who, in any particular issue, decides what is good for the greatest number? Why, the greatest number.

If you consider this moral, you would have to approve of the following examples, which are exact applications of this slogan in practice: fifty-one percent of humanity enslaving the other forty-nine; nine hungry cannibals eating the tenth one; a lynching mob murdering a man whom they consider dangerous to the community.

There were seventy million Germans in Germany and six hundred thousand Jews. The greatest number (the Germans) supported the Nazi government which told them that their greatest good would be served by exterminating the smaller number (the Jews) and grabbing their property. This was the horror achieved in practice by a vicious slogan accepted in theory.

But, you might say, the majority in all these examples did not achieve any real good for itself either? No. It didn't. Because "the good" is not determined by counting numbers and is not achieved by the sacrifice of anyone to anyone. [bold added]
So much for "utility-maximising automatons" -- and no wonder "Effective altruism can be a hard sell, even [!] for the rationally minded."

There is one aspect of this movement that merits further comment: If one wishes to give money to a cause -- and there are many valid, selfish reasons to do so; altruism does not own charity -- one obviously wants more bang for the buck. Counting lives saved by one donation is a poor metric (even if one grants improving large numbers of lives as an imperfect metric, albeit better than the one proposed by utilitarianism). Anyone with an ounce of sense can see this by considering whether it would be better (on such grounds) to save a thousand indigents from malaria vs., say, educating a Jonas Salk (whose research could save magnitudes more) or an Aristotle (who would make countless great men and even civilizations possible, by improving their minds). Even then, quantifying the impact of a donation might be difficult, to say the least.

"Effective altruists" should spend less time quantifying their results and more time considering what those results should be. It's ridiculous to ask, "How well am I doing?" when one doesn't really know what one is supposed to do, or why.

-- CAV


Dinwar said...

Effective Altruism seems, as far as I can tell, to be an outgrowth of the Bayesianist view that one should be able to place numerical values on one's certainty of one's conclusions. (Note that Bayesianists, such as the author of Slate Star Codex and those on Less Wrong, are distinct from those who use Bayesian statistics in their proper context.) It makes a certain amount of sense: If I think X will do the most good (by whatever standards I use to define it), I should logically contribute to X in some way. The EA folks go further, by discussing whether it's better to pay for a dedicated group to work on the problem or to do it yourself. Surprise surprise, paying someone is always the superior option...

As you say, the problem is that quantifying "good" is not something that can be easily done, if it can be done at all. I think that the root of the issue--a distorted view of statistical analysis substituting for thinking--is also worth examining.

Gus Van Horn said...


I see the problem as another example of "scientism." Too many people in the humanities should spend less time trying to make what they say be all math-y and "scientific" looking, and more time asking why they think they need to do that.

As you know (but I'll note for passers-by), Ayn Rand discussed a non-quantitative, but thoroughly objective scale for measuring values.