I think working this through on guesttimate rather than mulitplying point estimates is really important.

I tried doing it myself with similar figures, and I found the climate change came out ~80x better than global health (even though my point estimate that that global health is better) - which suggests the title of the article could maybe use editing!

When you're dealing with huge uncertainties like these, the tails of the distribution can drive the EV, so point estimates can be pretty misleading.

Here's a screenshot of the model: https://www.dropbox.com/s/adtwlz3k2myv8gc/Screenshot 2020-05-25 20.57.11.png?dl=0

I also tried doing the calculations in a different way that I found more intuitive - where I estimate the 'utils' of each intervention: https://www.dropbox.com/s/8uczqc1qhi71lte/Screenshot 2020-05-25 20.58.02.png?dl=0

Some other reasons in favour of this approach:

• Rob's point that by multiplying together extreme values, your confidence intervals are unreasonably wide.
• Some of the confidence intervals you give for the individual parameters also seem too wide (and seem to not be mathematically possible to fit to a lognormal distribution).

I’ve discussed this post with a couple of people, and realised it’s unclear about where I think the clearest gaps actually are, so I thought I’d add a list.

Each career service idea needs to pick (i) an audience (ii) a set of programmes (iii) a set of causes. You can make different ideas by combining these 3 factors.

Some of the gaps within each factor that we’ll likely leave unfilled include:

i) Audience – there are several audience groups that 80k won’t reach for some time. For instance, careers advice for people over 40 could be useful to increase age diversity in the movement, and find more experienced people, which is a key gap. You could also pick out an audience with a cause area e.g. ‘careers advice for EAs who want to work on global health.’ Another clear group is services for other countries we’re not going to cover (e.g. a German job board), or perhaps focused on certain career paths. Some of the most valuable audiences that we’re not ideally suited to are groups like academics and policy makers, though it’s hard to credibly work with these groups unless you’re a member of this audience. I would also like to see more people working on student groups at top universities. Each of these audiences is clearly differentiable from 80k’s focus,

ii) Programmes – It seems to me like the biggest bottlenecks are around one-on-one advice and headhunting, since we just don’t have enough staff to cover everyone worth talking to, and this means these people don’t get direct help. We’re also not going to get to specialist content outside our priority problem areas for a while, such as in-depth guides to global health careers, or a guide to how to switch career mid-career. On the other hand, we plan to continue to provide more general purpose written content (e.g. advice on high level principles like career capital).

iii) Causes – I think the highest priority is for people to fill programming gaps that we’re leaving open within our priority areas and other promising areas (e.g. better advice for people who want to work on nuclear security). However, there are also some issues we’re not going to cover for some time so you could also fill a gap by replicating one of our existing programs for one of those areas. Global health is perhaps the most obvious example, but you might also want to include some longtermist areas here, such as reducing great power conflict.

Both of the new groups I mentioned in the main post match all these factors pretty well.

Another factor is how easily new programmes can fit into and serve as a multiplier on the existing infrastructure. For instance, specialists in specific topics and causes are fairly easy to slot in, since we and others can simply link or refer people to them when someone needs help with those areas.

On the other hand, I think starting a new job board aimed at the effective altruism community is less of an obvious gap, since the 80k board does cover multiple cause areas (including listing 110 global health jobs and 50 factory farming jobs currently). This argument holds less for job boards aimed at a particular country or cause.

Of course, it may eventually be better to have a direct competitor to the 80k job board (or other core programmes), especially if most of the biggest gaps have already been filled.

Another general thought, my personal advice is to start by doing one thing well, and then broadening over time. I think 80k started by doing too many things at once, and we could have gone faster if we’d started more focused. It also makes it much easier for other groups to coordinate with you.

If you’re considering going ahead with a new organisation to solve a gap, we’d love to hear about it, in case it’s a gap we can quickly plug. e.g. we’re open to consider putting up extra articles, making edits to old articles, or making tweaks to our programmes (though we don’t have a ton of spare capacity and aren’t likely to change any of our main focuses).

I'm still a bit worried about this.

It would have been reasonable for them to use the mean global income as the baseline, rather than dollars to the mean US citizen.

If I understand correctly, that would boost things by about a factor of 3 in favour of climate change (mean global income is about $20k, vs. mean US income of about $60k). Though, I suppose that's a fairly small uncertainty compared to the others listed here.

Yes, unfortunately the leaders forum survey didn't ask about it as its own category, so it's merged into the others you mention.

I agree, I don't like the near-termism vs. longtermism terms, since I think it makes it sound like a moral issue whereas it's normally more about epistemology or strategy, and like you say for most people it's a matter of degree. I hope we can come up with better terms.

I also agree people should be clear about 'causes' vs. 'worldviews'. You could be longtermist in your worldview but want to work on economic empowerment, and you could be near termist but want to work on AI GCRs.

I did my analysis in terms of causes, though my impression is that results are similar when we ask about worldviews instead (because in practice causes and worldviews are reasonable correlated).

I'm not aiming to take a stance on how important representativeness is. My goal is to get people to focus on what I see as the bigger issue we face today: how should we design a community when the new members and the "middle" have mainstream cause priorities and the "core" have (some) unusual ones?

Hey Khorton, I didn't mean to imply that. I think the last paragraphs still stand as long as you assume that we'll want some of the core of EA to work on unusual causes, rather than 100%.

I'm also not sure I know what you mean.

Hi Denise, on the second point, I agree that might be a factor (I mention it briefly in the article) among others (such as people changing their minds as in David's data). My main point is that this means the problem we face today is more like "people are bouncing off / leaving EA because the most engaged ~2000 people focus on unusual causes" rather than "the leaders don't represent the most engaged 2000 people".