“Voting is effective altruism,” many EAs say.

Is it, though?

It’s weird that so many effective altruists celebrate voting. Here’s why:

1. The question of the best way to calculate the probability of a vote being decisive is still open. Most models say it’s very low. EAs love to latch on the Edlin, Gelman, and Kaplan view, which says it can be pretty high if you live a swing state.

2. They often just assign subjective values to the differences in the expected value of two different candidates or parties. This seems rather anti-EA. Aren’t we supposed to carefully measure the differences using the best available research, just like we do when assessing where to donate our cash? Also, lots of research says that parties and candidates have little to no real impact on policy.

3. They ignore how votes cancel each other. If voting for a Democrat is like donating $5000 to charity, isn’t voting Republican like stealing $5000?

4. They ignore how difficult it is for a typical person–and even the typical academic–to figure out which way to vote is better. Voters are in a very bad epistemic state. Most have few stable political beliefs, and the few they have are not justified. Their reasons for being attached to one party instead of another are almost always non-doxastic and non-epistemic, but are instead based on arbitrary historical connections that do not track their interests or anyone else’s. Most voters should be highly self-skeptical and should think they have as much chance of doing something harmful as they do of voting in helpful ways.

And so on.

Here’s the rub, though. Utilitarianism, like most moral theories, is often self-effacing. That is, telling people the truth about consequences might have bad consequences.

A good consequentialist will almost never vote. In addition, the good consequentialist will realize they probably are not in a position to judge which candidate or party is better in major election. However, for the rare cosequentialist who does justifiably believe that party D > R and who also has lots of public influence, they should try to convince as many people as possible to vote the right way.

The EA pro-voting message is probably false but also useful. Given who hears the EA message, it’s strategically useful to convince lots of people in the audience to vote, even though it’s still the case that for any single one of them, their vote is not a form of effective altruism.