The movement data is action-relevant for me, as I’m spending several hours a week going to meetup groups purely to recruit GiveWell donors. I’ve found skeptic/atheist groups particularly fertile, and lefty political groups (and ‘A’ rather than ‘E’ groups generally) the opposite. I haven’t tried any conservative or libertarian groups yet.
I haven’t tried any conservative or libertarian groups yet.
Given that conservative (I believe especially evangelical groups) donate the most to charity, it’s probably worthwhile checking them out.
My understanding is that their current approach to the inefficient charity problem involves organizing trips to the countries in question and having members personally help the charity. While this is clearly not the most efficient approach, it does help with the “most of the money winding up in the hands of middlemen” problem while also generating warm fuzzes.
and lefty political groups (and ‘A’ rather than ‘E’ groups generally) the opposite.
That’s because lefty and ‘A’ groups are mostly about signalling one’s virtue, thus someone who shows up and starts telling them how none of the ‘virtuous’ things they’ve been doing are actually helping people is most certainly not welcome.
Uhm, upvoted the comment, but don’t completely agree with the linked article.
It suggests that when fans of something are worried when it becomes too popular, they object against losing their positional good. That’s just one possible explanation. Sometimes the fact that X becomes widely popular changes X, and there are people who genuinely preferred the original version. -- As a simple example, imagine that tomorrow million new readers will come to LW; would that be a good thing or a bad thing? Depends on what happens to LW. If the quality of debate remains the same, that it’s obviously a huge win, and anyone who resents that is guilty of caring about their positional good too much. On the other hand, the new people could easily shift LW towards the popular (in sense: frequent in population) stuff, so we would get a lot of nonsense sprinkled by LW buzzwords.
I can imagine leftist groups believing they are working “more meta than thou”; solving a problem which taken in isolation doesn’t seem so important (compared with the causes effective altruists care about), but would start a huge cascade of improvement afterwards (their model of the world says so, your model doesn’t). Making mosquito nets instead is not an improvement according to their model.
Edit: Stupid comment, too much reddit today. Infantile regression. I apologize. I disagree with the parent comment (“small trend toward the mean in the census = smartest move on to greener pastures”) and meant to poke fun at it by showing the absurd fringe case; only dumb cows remaining (which I’m not, hence my disagreement would be conveyed). Convoluted. Sorry.
It suggests that when fans of something are worried when it becomes too popular, they object against losing their positional good. That’s just one possible explanation. Sometimes the fact that X becomes widely popular changes X, and there are people who genuinely preferred the original version.
That doesn’t explain why the new X looks much more like an extreme version of the popular version of X rather than the original X.
I see Larks’ point.
The movement data is action-relevant for me, as I’m spending several hours a week going to meetup groups purely to recruit GiveWell donors. I’ve found skeptic/atheist groups particularly fertile, and lefty political groups (and ‘A’ rather than ‘E’ groups generally) the opposite. I haven’t tried any conservative or libertarian groups yet.
Given that conservative (I believe especially evangelical groups) donate the most to charity, it’s probably worthwhile checking them out.
My understanding is that their current approach to the inefficient charity problem involves organizing trips to the countries in question and having members personally help the charity. While this is clearly not the most efficient approach, it does help with the “most of the money winding up in the hands of middlemen” problem while also generating warm fuzzes.
That’s because lefty and ‘A’ groups are mostly about signalling one’s virtue, thus someone who shows up and starts telling them how none of the ‘virtuous’ things they’ve been doing are actually helping people is most certainly not welcome.
Uhm, upvoted the comment, but don’t completely agree with the linked article.
It suggests that when fans of something are worried when it becomes too popular, they object against losing their positional good. That’s just one possible explanation. Sometimes the fact that X becomes widely popular changes X, and there are people who genuinely preferred the original version. -- As a simple example, imagine that tomorrow million new readers will come to LW; would that be a good thing or a bad thing? Depends on what happens to LW. If the quality of debate remains the same, that it’s obviously a huge win, and anyone who resents that is guilty of caring about their positional good too much. On the other hand, the new people could easily shift LW towards the popular (in sense: frequent in population) stuff, so we would get a lot of nonsense sprinkled by LW buzzwords.
I can imagine leftist groups believing they are working “more meta than thou”; solving a problem which taken in isolation doesn’t seem so important (compared with the causes effective altruists care about), but would start a huge cascade of improvement afterwards (their model of the world says so, your model doesn’t). Making mosquito nets instead is not an improvement according to their model.
The results can already been seen in the Census Survey: There is a small trend toward the mean. The smartest move on to greener pastures.
Moo?
Edit: Stupid comment, too much reddit today. Infantile regression. I apologize. I disagree with the parent comment (“small trend toward the mean in the census = smartest move on to greener pastures”) and meant to poke fun at it by showing the absurd fringe case; only dumb cows remaining (which I’m not, hence my disagreement would be conveyed). Convoluted. Sorry.
Saw this in recent comments, thought how curious is that there is a context in which this comment is not silly. I was wrong. What did you mean, again?
I see those two points to be independently supported by the survey and not to imply each other in any obvious way.
That doesn’t explain why the new X looks much more like an extreme version of the popular version of X rather than the original X.