Skip to main content

Longtermism attracted some of the sharpest criticism the movement has faced, much of it from philosophers who share its concern for the future but doubt its reasoning. The objections fall into four families: worries about tiny probabilities, about what we can know, about how to count people who do not yet exist, and about the people who exist now.

Pascal's wager and its mugger

The problem of tiny probabilities is old. In his Pensées, published after his death in 1670, Blaise Pascal argued that a reasonable person should bet on God's existence, since the stakes are infinite: "If you gain, you gain all; if you lose, you lose nothing."

Modern philosophers turned the point into a puzzle. In the story called Pascal's mugging, a stranger without a weapon asks for your wallet and promises to repay you with an astronomically large reward tomorrow. You think the promise almost certainly false, but not impossible. If the promised reward is large enough, expected value tells you to hand over the wallet. Pascal's mugging shows how expected value reasoning can be hijacked by tiny chances of enormous payoffs.

Philosophers call the willingness to follow such reasoning fanaticism. Fanaticism is the view that a tiny chance of a vast enough payoff should outweigh any sure, modest good. A few philosophers have defended it, arguing that every alternative leads to worse paradoxes. Many others think that if longtermism depends on fanaticism, so much the worse for longtermism.

Cluelessness

A second worry concerns knowledge. Every act ripples outward. A small kindness in a medieval village might change who met whom, and so which people were born in every generation after. Cluelessness is the worry that we cannot know whether our actions make the far future better or worse. The philosopher James Lenman pressed this point against consequentialism in 2000, and Hilary Greaves later distinguished cases where our ignorance cancels out from harder cases where it does not.

A related objection says that predictable effects tend to fade. The washing-out objection says the effects of most actions dissolve into noise over long periods, so we have no reliable way to steer the distant future. Longtermists reply that preventing extinction is special because its effects are permanent rather than fading.

Counting people who do not exist

The third family of worries comes from population ethics, the study of how to compare outcomes containing different numbers of people. The total view ranks outcomes by adding up all the well-being across everyone who ever exists. It supports the longtermist claim that extinction is a catastrophe partly because of all the good lives that would never be lived.

Parfit showed that the total view has a disturbing implication. For any population of very happy people, there is some much larger population of people whose lives are barely worth living that would contain more total well-being. The repugnant conclusion is that a vast population of barely worthwhile lives could be better than a smaller flourishing one. Some longtermists, MacAskill among them, have argued that the conclusion may be less damning than it sounds.

Rival person-affecting views hold that an outcome is worse only if it is worse for someone. On such views, a future person who is never born has not been harmed. Extinction would still be terrible for those alive when it happened, but the vast loss of future lives would carry far less weight, and the case for longtermism would shrink.

The people here now

The fourth worry is moral and political. A frequent political criticism is that longtermism can justify neglecting urgent present suffering. If the far future dominates every calculation, then children dying of malaria today seem like a rounding error. Critics such as the philosopher Émile P. Torres have also argued that longtermism appeals to wealthy technologists because it directs attention to speculative futures rather than to present injustices they might be asked to fix. Defenders answer that most longtermists still give to present causes, that risk reduction protects the living.

Worked example: bed nets or biosecurity

A foundation has 10 million dollars to spend. Given to an excellent malaria charity, on GiveWell's recent estimates of a few thousand dollars per life saved, it might save around two thousand children. Given to biosecurity research, it might lower the risk of a catastrophic engineered pandemic by some tiny amount. If that reduction were one in ten million, and the future held a trillion lives, the expected value would be a hundred thousand lives.

Apply the objections. Fanaticism: the one in ten million is a guess, and the calculation is only as good as that guess. Cluelessness: research might accelerate dangerous knowledge as easily as defenses. Population ethics: on a person-affecting view, most of those hundred thousand lives are not people who would have been harmed. Present people: the two thousand children are real and their deaths are certain.

Many funders respond by refusing to put everything on one bet. They split their money across several worldviews, each given a share in proportion to how plausible they find it, so that no single calculation decides everything.

Quiz

5 questions · Practice: scored, not saved

1. What does Pascal's mugging illustrate?
2. What is fanaticism, in the philosophers' sense?
3. What is cluelessness?
4. What does the washing-out objection say?
5. How do longtermists reply to the washing-out objection?
6. How does the total view rank outcomes?
7. What is the repugnant conclusion?
8. What is a frequent political criticism of longtermism?

CivilSimian.com created by AxiomaticPanic, CivilSimian, Kalokagathia