Mainly because I didn’t think we yet had product-market fit, so it was better to maintain a small flexible, able team and focus on learning, rather than focus on scaling. Today I think we’re close, but still not quite there.

Overall the trajectory is not dissimilar to GiveWell. It is a bit slower than for-profits (though even Airbnb went for like 2 years without hiring anyone if I remember correctly), but that’s to be expected because we’ve had to invest a lot of time in impact evaluation, which for-profits don’t. We’ve also had to do novel research into career choice before making our products, adding an extra stage to the whole process.

There’s other factors too e.g. with written content a small number of people can have a very large impact, so better to focus on that before hiring a load of people.

Also, we should be able to do a different model of scaling with EA orgs. Because we don’t have ownership and profit-sharing like companies do, you can imagine loose collaborations of charities (a bit like CEA, but looser) that group together to gain some economies of scale but less of a risk of bureaucracy.

I think it’s particularly important to understand that economies of scale tend to operate at an organisational level, while diminishing returns tend to operate at an industry/area level (and are typically only relevant when we could affect a good fraction of the entire area). So thinking about diminishing returns may be a helpful way to balance between different research areas, for instance, but doesn’t look good as a way of balancing between different research groups. Similarly economies of scale mean it makes sense for a lot of the effective altruism ‘movement building’ activities to be concentrated in a relatively small number of groups.