Comment author: joshjacobson  (EA Profile) 13 November 2017 09:02:32PM *  1 point [-]

This looks great! Looking forward to doing a more detailed read when I have more time, but I already see some resources and techniques I wasn't aware of or have failed to fully implement thus far, so this will serve as added motivation and a nice reference.

Sci-Hub is another resource that is likely to be highly useful to those without institutional access to journal subscriptions. And I find that the archive of Data Is Plural is a great source for data on a wide variety of topics: http://bit.ly/2h3bNzQ

Comment author: Joey 05 November 2017 01:13:32PM 2 points [-]

The use of the term talent constrained vs talent limited was not intentional.

Overall I think salary is not a large factor in our talent concerns. We have experimented with different levels of salaries between 10k and 50k USD and have not found increasing the salary increases the talent pool in the traits we would like to see more of. It could be that 50k is still too low or that we are not marketing our jobs in communities that are very income sensitive. I would guess that normally we are looking for pretty hardcore/dedicated EAs and that tends to correlate very strongly with people who take low salaries.

Comment author: joshjacobson  (EA Profile) 06 November 2017 10:21:50PM 2 points [-]

What was the communication of this like? As someone who I believe has monitored CS pretty closely, I can't remember a time a salary approaching $50k was communicated.

Comment author: joshjacobson  (EA Profile) 03 November 2017 06:56:29PM *  2 points [-]

I see you didn't call Charity Science 'talent constrained' but rather 'talent limited'. Was this intentional? Because it does seem Charity Science is an org that would get much better access to talent with more funds for salaries.... and that that is a likely major factor in your talent shortage.

Comment author: joshjacobson  (EA Profile) 30 October 2017 06:52:31PM *  2 points [-]

(x-post from FB, so phrasing is written more directly as a comment to Scott)

I think this is mostly spot on. There's one or two additional things I might have included based on my experience (would probably emphasize warm introductions more and mention the value in getting on their radar early).

Also just noting that I think the email could have been improved upon, but I'm interested in whether you share this belief. Top suggestion would have been to have one of the key attention-grabbing names in the subject line of the email, and to prioritize brevity a bit more.

I'm glad you wrote this... I do get questions in this vein a lot and expect it to be a helpful resource for many.

Comment author: oliverbramford 23 October 2017 03:20:36PM 0 points [-]

I had written a much longer piece, and had feedback from a number of people that it would be best to split it up.

Comment author: joshjacobson  (EA Profile) 23 October 2017 04:43:00PM 0 points [-]

I can't imagine why. Even all 3 together are shorter than many posts on here. And they really don't have much standalone value IMHO (i.e. the first).

Comment author: joshjacobson  (EA Profile) 23 October 2017 02:00:23PM 0 points [-]

I don't think there was any reason for this to be split into 3 posts? It'd be better to condense it into one.

Comment author: joshjacobson  (EA Profile) 26 September 2017 10:55:16PM 5 points [-]

The vast majority of large institutional spending is somewhat static. When there have been major shifts, it is usually in response to the combination of highly successful marketing campaigns and new events.

Malaria has been largely ongoing, without much newsworthiness (to regular media outlets) or specific press. It's funding therefore is likely to have stayed at a somewhat static level in most organizations.

In contrast, HIV/AIDS was emergent in previous decades. It went from nothing to being highly prominent in a short time period. Relatively large budgets were allocated against it because:

  1. It showed a pattern of significant growth, and there was significant fear that not containing it could lead to runaway growth.

  2. It emerged from 0 cases to being prominent, which was highly newsworthy.

  3. There was a strong coordinated marketing campaign to get governments and IGOs to strongly address it.

HIV/AIDS funding came at a relatively high level as a result, and because funding is largely static and the problem remains, it has stayed that way.

Comment author: joshjacobson  (EA Profile) 21 September 2017 02:46:09PM 7 points [-]

I only skimmed this, but I think the majority of EAs don't actually look into the how and why of GiveWell's recommendations. And even less go into the processes and publications that lead to the numbers that GiveWell eventually uses. An indirect result is that GiveWell doesn't get as much feedback as it could likely benefit from, and too many EAs can't speak to M&E professionals in international development at a meaningful level.

What's explained here, and alluded to here, as well as the criticisms, is important basic info for many EAs who are unfamiliar with it. The various methodologies for costing and discounting (both included here and others), in particular, are definitely worth investigating further for those who haven't.

Comment author: Peter_Hurford  (EA Profile) 12 September 2017 03:02:56AM 2 points [-]

This is covered in detail in the methodology section. We try not to talk about statistical significance much, we try to belabor that these are EAs "in our sample" and not necessarily EAs overall, and we try to meticulously benchmark how representative our sample is to the best of our abilities.

I agree some skepticism is warranted, but not sure if the skepticism should be so significant as to be "quite misleading"... I think you'd have to back up your claim on that. Could be a good conversation to take to the methodology section.

Comment author: joshjacobson  (EA Profile) 12 September 2017 02:59:16PM 0 points [-]

I haven't looked there yet, so I'm flagging that my comment was not considering the full context.

(I think that the end links didn't come up on mobile for me, but it could also have been an oversight on my part that there was supporting documentation, specifically labelled methodology.)

Comment author: joshjacobson  (EA Profile) 11 September 2017 03:03:53PM -1 points [-]

I think it's quite misleading to present p-values and claim that results are or aren't 'statistically significant', without also presenting the disclaimer that this is very far from a random sample, and therefore these statistical results should be interpreted with significant skepticism.

View more: Next