Jeff Kaufman

Software Engineer @ Nucleic Acid Observatory
14151 karmaJoined Working (15+ years)Somerville, MA, USA



Software engineer in Boston, parent, musician. Switched from earning to give to direct work in pandemic mitigation. Married to Julia Wise. Speaking for myself unless I say otherwise.

Full list of EA posts:


Raising my hand for an even more niche category: people who likely would have attended LessOnline had their partner not been attending EAG.

there are certainly cases where it doesn’t make sense to give annually

Does GWWC have recommendations on how to handle inflation? For example, if I earn $X and then wait a couple years during which we have 10% cumulative inflation, do I now donate 10% of 1.1 * $X? Or (in my case) if I gave more than my pledged amount for multiple years while earning to give, building up a 'surplus', and am considering giving less than my pledged amount for a bit while I work a lower paying direct job, should I account for that my donations were of more valuable dollars? Should we handle this asymmetrically (yes for #1 and no for #2) to stay conservative and above reproach? Does the dashboard tracker account for inflation?

(I think the "correct" way to handle it is to do all the calculations in constant dollars, adjusting for inflation, but since this could look to the general public like motivated reasoning for giving less in case #2 perhaps GWWC thinks we shouldn't do it?)

It can be disheartening for people to be stumble across 80,000 hours content and be inspired but then be left a bit disappointed by the practical options available to them.

To me this is the key point in your post: if you try to bring people into a movement, saying "X is the thing that most needs doing" and then when people decide "ok, I'll go work on X" you say "actually we have a lot of people already, probably go do something else" they will understandably be disappointed!

But I'm also a little confused by your calculations. You say:

If we assume fifty percent of the attendees are interested in EA roles (100) then there is roughly 10x as many people interested in EA roles as there are available domestically.

The 10 roles available in Australia sounds like it's coming from adding up the number of Australians working in EA community building in Australia, is that right? If the number of people in EA community building were on the same scale as the number of people attending EA conferences, that would actually be very worrying -- if people count impact by getting others into the movement, but then the only impact those others have is getting still more people to join the movement you're not actually doing anything to make the world better.

When I look over the 80,000 Hours list of pressing problems I see lots of things that it should be possible to work on from many places, including Australia?

(Aside: I saw you used both "give to earn" and "earn to give". I'm used to seeing the latter -- was this a typo or is there a distinction you're trying to draw between the two?)

Can you say more about what your "Household Income Per Impacted Child" metric is? Searching online, I only see it in this post and the linked spreadsheet.

Changing the language they used

My dance group switched from gendered terms for the roles to non-gendered (blog post) and from calling one of the dance moves "Gypsy" to "right shoulder round" (blog post). This didn't involve strife in our specific community, though other dance communities had serious rifts over these two issues.

In the gendered terms case, our transition was the outcome of a long process with the community, including talking about various term options, trial dances, and then polling. We needed to do it this way because role terms are very visible.

In the "Gypsy" case the board and callers did it much more quietly. The caller booker told callers individually that we'd prefer "right shoulder round" if they were comfortable with it, then later that we strongly preferred it, and then eventually started asking callers not to use "Gypsy".

Communities where these two issues went poorly often had the people who were pushing for change taking a confrontational attitude, and accusing people who disagreed with them of being prejudiced. I think they typically also involved advocates pushing hard on this before the community was ready, not building support, and not getting buy-in from respected members.

there seems to have been a trend of former leaders and managers switching back to object level work

Guess: people who enjoy object level work ended up doing leadership and management because EA catastrophic risk work was so low on experience there. As this crunch has somewhat resolved those people are able to go back to the work they like and are good at?

(But you probably know more than I do about whether the management/leadership experience crunch has actually lessened)

Thanks for the feedback!

Why not just specify a distribution with some parameters rather than list lots of possible values drawn from that distribution?

The values in the list aren't drawn from a parametrized distribution, they're the observed values in a small study.

Maybe rather than have the line go back to 0, just stop it when it hits 30%


the y-axis numbers are cut off


if for whatever reason you run lots of scenarios, where the whole bottom half of the graph disappears

This was due to me not testing on monitors that had that aspect ratio. Whoops! Fixed by allowing you to scroll that section.

For qPCR or other targeted detection approaches wastewater has quickly become a very common sample type, mostly because (a) it was very successful for covid, (b) a single sample covers hundreds of thousands of people, and (c) it's an 'environmental' sample so it's easy to get started (no IRB etc). And targeted detection is generally sensitive enough that the low concentrations are surmountable.

There isn't really a status quo for metagenomic monitoring: everything is currently in its early stages. There are academics collecting a range of samples and metagenomically sequencing them, but these don't feed into public health tracking, partly because they're not running their sequencing or analysis in a way that would give the low sample-to-results times you'd need from a real-time monitoring system.

Load more