This gets the heart of my problem with GiveWell, which is that often I hear about something that sounds potentially really effective, I look for it on GiveWell’s site, they’ve analyzed it to death, and their conclusion is “yes, this potentially sounds really effective, but we decided not to make it one of our top charities”, without great explanation of why.

See eg:
https://www.givewell.org/international/technical/programs/salt-iodization
https://www.givewell.org/charities/IGN#Cost-effectiveness_of_salt_iodization_implementation
https://www.givingwhatwecan.org/post/2016/01/are-we-underestimating-benefits-salt-iodization/

The best I can do is this quote from the last source:  

“The major unresolved issue with these charities is that GiveWell has “thus far been unable to document a demonstrable track record of impact; we believe it may have had significant impacts but we are unable to be confident in this with what we know now.” Essentially, since these charities assist governments and salt producers in iodising salt, the counterfactual impact (what would have happened if not for the charity?) is very difficult to assess.”

Whenever I reach this point, I get confused about how to deal with this lack of perfect information (see http://www.overcomingbias.com/2008/08/doctor-there-ar.html ). Surely if it sounds really promising, but we don’t have perfect evidence that every step works, we should revert to our prior that it’s really promising, right?

I *think* the argument here is the same one I made about the Reverse Streetlight Fallacy a few days ago - that charities have such a bad track record of working that GiveWell’s prior is strongly weighted towards “doesn’t work” until all of the evidence comes in, even if it looks very promising.

Another argument might be that if you’re trying to raise IQ and you don’t care that much about whether the evidence is absolutely perfect, deworming might be higher expected value.

Like you, I’m a little disappointed all of this isn’t more explicit so that I can have an informed opinion on it just from reading websites and stuff.

I’ve heard some complaints about the subreddit, but none of them have focused on the mod team; can you be more specific about what your concerns are?

(don’t do this unless you really want; right now I have a high prior on trusting the mods and it’s unlikely you’ll change that with standard “they banned me when I didn’t deserve it” level complaints, although if you have something worse than this by all means let me know.)

I worry this (and the post on fish-oil/bioperine earlier) are a kind of reverse streetlight fallacy. IE when we investigate things we figure out they don’t work very well, so everyone gets bored of them and interested in subjects that can’t be or aren’t investigated rigorously.

I can see a level on which this makes sense - if you have two options, 1 and 2, and 1 is definitely bad, and 2 is unknown, go with 2.

But I think there’s another level where if you have one hundred options, and you’ve investigated 1 through 99, and they all looked really promising at first, but in the end 98 of them were useless and 50 was merely okay, and now you also have a new uninvestigated option 100 that looks really promising at first, you should still have a prior that option 50 is better than option 100.

That was a terrible explanation but hopefully helped communicate what I mean? I just see this in medicine all the time, where we keep getting promised miracle cures, most of them fall apart, a few of them work sort of okay, and people ignore those and get really excited about the next potential miracle cure here. I’m not talking about homeopathy, I’m talking about the thing which has good theory behind it, is very promising, but realistically will fall apart just like everything else or work a little like everything else.

I’m mostly against doing this.

First, I might be wrong about Trump. But even if I’m right, there were many other people who were aggressive and unpopular during their own time but who as far as I can tell ended up being a net positive to their movements. FDR and Reagan seem like good examples here. I don’t think anyone is very good at predicting who will be in this category and who will be in the more Trump-ish category.

Second, even if Trump has caused a 5 pp shift away from populist attitudes and this was predictable beforehand, I think you would have to weight this against all the other potential effects of him being in power. Before his election, it wasn’t really clear if he was restrained enough not to start a great power war. Now that he’s demonstrated a tendency to concentrate on tweeting and not personally make major foreign policy decisions I’m a little less concerned; maybe there are people who could have predicted this beforehand, but I couldn’t. Progressives should also be concerned about his two Supreme Court appointments, his climate policy, et cetera. Even if I’m 100% right that Trump caused populism to lose support in a way other people wouldn’t, and even if this change is lasting, it has to be balanced against both his expected and his real downsides.

Third, making decisions like this creates weird incentives; if some candidates know you’re supporting the worst person with an ideology, in order to discredit that ideology, they will try to get worse.

I think these are important enough points that I would be very reluctant to try to make something like this happen strategically.

For a counterpoint, Nick Land believes the winning conservative strategy is to try to never win the presidency, then blame the (liberal) president for everything - see http://www.xenosystems.net/popcorn-activism/ . I think this could work, but only because Nick doesn’t really care about policy in a normal way and is just trying to maximize conservative popularity and/or lulz.

I am sort of tempted to apply something like this to Bernie Sanders, but only because I think it would be good for the country to be shifted towards a welfare state in the short-term, practical way that a president can do, and also good for socialism to be discredited in a longer-term more philosophical way. This is sufficiently weird that I don’t think it generalizes to very much else.

There’s an issue where it looks like lots of stuff should work for health, and then we test it and it doesn’t.

And some people use this to conclude “I guess we should be much more rigorous before trying anything, because the stuff we try based on mere conjecture never ends up working.”

And other people use it to conclude “Well, then let’s try this next thing!”

I think it’s *possible* to justify the second one, with some argument like “The risk is very low and I guess there’s a chance this is the one that might work”, except the risk being low is another one of those assumptions which frequently doesn’t work out.