While I find much of this post to be plausible, I’m not sure Ollie’s post supports your conclusions.

Ollie’s post is evaluating a set of retreats which averaged a cost of $1,500 per person. As commenters on the post noted, this seems very high. (I recall reading that low end EAG costs are around the same spot.) For the one retreat I’m aware of, costs were 6-7x less. (This doesn’t include CEA staff costs, but those shouldn’t be able to make up the gap.) 

Additionally, you write about how retreats might have lower outcomes due a lack of scale. While I’m sympathetic to the idea that larger scale events can provide better sorting and outcomes per person, this doesn’t seem to be the case for the sample Ollie looked at. He notes that “Outcomes per person are approximately similar in value”. 

On the whole, I don’t think the post shows that EAGx’s outperform retreats on cost effectiveness in a useful sense, mainly because of the cost issue. Ultimately there have been many more retreats and conferences since Ollie’s post, and I would love to hear from someone at CEA about their present feelings on the relative cost effectiveness of different types of “small” events.

I’m also a co-president at EA Purdue.

We had very brief intros for each charity and then a non-binding vote at the start, a short discussion on the vote, followed by an introduction to effective giving with a discussion in the middle. We then gave more info on each charity. Voting was done as: discussion, first round vote to eliminate one charity, discussion, final round of voting to choose winner.

Charities were AMF, GiveDirectly, and the Humane League. The slides were a modified version of GWWC’s donation election presentation so the initial vote also included the Play Pumps fake out.

What about counties that exit low income in the next fifty years? Under your assumptions and framework, we can be sure that we won’t accidentally exclude a future low income country, but we can’t be sure we won’t fail to select a future low income country.

I somewhat agree with 2, although I would phrase it slightly differently.

The local maximum for outreach is a function of the quality of the organizing, the quality of the students, and the size of the student body. (Quality of students as potential org members, not in any objective sense). The latter two are essentially fixed, but organizing quality can shift dramatically over time. The form of the organization could also matter, but I feel this is pretty similar across EA groups and within groups over time. (I would love to see evidence that I’m wrong on this). Organizing quality has diminishing returns, but many orgs aren’t necessarily anywhere near this. 

Without knowing Yale EA’s situation at all, I would be willing to guess that they had a substantial return on outreach due to an increase in organizing quality, which wasn’t previously close to diminishing returns, and which paid particularly large dividends thanks to very high student quality.