Thanks for the reply! I agree having a directory seems potentially useful, and also that there could plausibly be some cases where having familiarity with EA could be particularly beneficial. Hopefully you're documenting such cases and can point to examples. I'm just a bit wary that sometimes there seems to be a reluctance to use outside experts.

I liked this and would encourage you to publish it as a top-level post.

How did you identify "services that there is a high demand for but not enough supply"? Is it simply based on the "quick look" you did, or is there some other evidence? 

The absence of EA services could simply be evidence of sufficient non-EA services, in which case it's probably worth thinking about the pros and cons of having EA services. 

The most obvious justification seems to be to keep money in the community, and/or to provide services at a relative discount. 

However, by relying on EA services there is a risk of missing out on the highest quality services that already exist: I can't think of any particular reason why EAs would necessarily be better than the rest of the world at providing finance, legal, or tech services. Though perhaps in many cases this doesn't matter - maybe EA orgs merely need 'adequate' rather than 'best'.

Why is it inadequate to use language associated with Bayes in an informal analysis? Are you suggesting that when people communicate about their beliefs in day-to-day conversation, they should only do so after using Dirichlet or another related process? Can you see how that is, in fact, extremely impractical? Can you see how it is rational to take into account the costs and benefits of using a particular technique, and while empirical robustness may sometimes be overwhelmingly important in some contexts, it is not always rational to use a  method in some contexts such as if there are too high costs associated with using it?

Please keep taking screenshots! I'm sure you wouldn't want to mislead your audience by only showing part of the discussion out of context :)

While I expect some EAs and rationalists do actually use Bayes formally in their analysis, a lot of its use is informal, using language associated with Bayes to communicate an approach to updating beliefs.

From this informal perspective, clarity and conciseness matters far more than empirical robustness.

Given that this only involved one week of research, this is a strong starting point for more in-depth work - which, as you have indicated, will likely result in a less cost-effective estimate.

One quick comment:

DALYs are bad. They measure the burden of disease. They represent years lost to mortality, and years lost to poor health. We want to avert DALYs. In a couple of places this post talks about averting DALYs, but it mostly seems to talk as if DALYs are a good thing being lost to diabetes. 

Similarly your evaluative framework states that 'A full, healthy life is equivalent to 61.09 DALYs'. This seems to be conflating DALYs with QALYs; QALYs are good - we want more QALYs.

Agreed that it doesn't solve that specific problem, but it serves the same end goal: making things easier for the reader.

This post has some additional helpful tips, in particular having a summary/putting key points up front.

I'm not convinced that is a reasonable proxy. There seems to be a big difference between 'believe that races differ genetically in socially relevant ways' and 'believe that there are population-level differences in average intelligence between ethnic groups'. The latter statement is much more specific.

For example, I would agree with the first statement on the grounds that certain genetic diseases are more prevalent in some ethnic groups than others (e.g. cystic fibrosis is higher in white people, sickle-cell disease is more prevalent in people of African and African-Caribbean origin), and this is socially relevant for how healthcare resources are used.  This seems to be scientific consensus - and indeed I think I first learnt about these differences as a biology student, either in high school or university.