Thread by @benskuhn

Trying to articulate a difficulty I have with the effective altruism community:

One thing that I, and I think many others, value about EA is that it's an antidote to using motivated reasoning to justify whatever you're currently doing for non-altruistic reasons as high-impact.

For this reason, people care a lot about what activities "count" as effective.

Periodically, people try to convince EAs that their "pet cause" is effective. In order to justify excluding those people, EA wants very objective standards for what "counts."

The original EA arguments satisfied this. They were simple, highly legible cases where it was easy to be objective—eg "guide dogs cost 100x more than trachoma surgery," "bednets save a life for $3500."

For this reason, many people in effective altruism got really excited about a set of simple, highly legible cognitive tools you can use to make arguments like this: back of the envelope calculations, expected-value reasoning, thinking about counterfactuals...

In some sense, this was an important part of the vision of the effective altruism community / the case for why it made sense: even if people disagreed about the object level, they had a shared commitment to a set of rigorous epistemic standards.

But: it turns out most of the highest-impact things one can do (probably) aren't amenable to these tools because they're too complicated. The uncertainty intervals on the relevant numbers span orders of magnitude and even figuring out the sign of a lot of things is hard.

Instead, the best arguments for what many EAs think are the most effective causes involve hundreds of pages of reasoning spanning many different lines of evidence, some of which is quantitative but a lot of which isn't. They require a huge amount of context to understand + judge.

Because of this, it's hard to define what "counts" as effective altruism in any way other than "stuff this set of cool kids thinks is important." There's no set of epistemic principles that I can point to and be like, "follow these to make a legitimate EA argument."

This also makes it a lot harder to justify the community's implicit claim to specialness or unusual effectiveness. If someone were to ask me why I pay special attention to the EA community, there's nothing in particular I can point to that seems like a unique insight or process.

And, it makes it harder to do the "antidote to motivated reasoning" thing I mentioned earlier. If someone tries to persuade me that their pet cause is the best possible thing to do, I can't just be like "come back with a back of the envelope calculation and we'll talk."

I don't know why this is so unsatisfying though. Maybe I just miss the days where it felt like you could figure out how to save the world by arguing about math on the internet.

(Lots of great replies here that Twitter's UI is making it very hard to navigate since they're spread across the entire thread. Sorry if I miss yours because of that!)