Thank you for the excellent content that you produce for the EA community, especially the podcasts.
There is one issue that I want to raise. I gave serious thought to raising this via your survey, but I think it is better raised publicly.
In your article “The case for reducing extinction risk” (which is linked to in your “Key ideas” article), you write:
Here are some very rough and simplified figures to show how this could be possible. It seems plausible to us that $100 billion spent on reducing extinction risk could reduce it by over 1% over the next century. A one percentage point reduction in the risk would be expected to save about 100 million lives among the present generation (1% of about 10 billion people alive today). This would mean the investment would save lives for only $1000 per person.
At the top of the page, it says the article was published in October 2017 and last updated in October 2017. There are no footnotes indicating any changes were made to that section.
However, an archived copy of the article from June 2018 shows that, at the time, the article read:
We roughly estimate that if $10 billion were spent intelligently on reducing these risks, it could reduce the chance of extinction by 1 percentage point over the century. In other words, if the risk is 4% now, it could be reduced to 3%.
A one percentage point reduction in the risk would be expected to save about 100 million lives (1% of 10 billion). This would mean it saves lives for only $100 each.
I think it would be helpful to members of the community to indicate when and how an article has been substantively updated. There are many ways this can be done, including:
an article explaining how and why your views have changed (e.g. here, here/here, and here);
linking to an archived version of the article (as you do here) ideally with a change log; and
a footnote in the section indicating what it previously said and why your views have changed.
I understand that you have a large amount of content and limited staff capacity to review all of your old content. But what I’m talking about here is limited to changes you choose to make.
I’m sure it was just an oversight on the part of whoever made the change. You all have a lot on your plate, and it’s most convenient for an article to just present your current views on the subject.
But when it comes to something as important as the effectiveness of spending to reduce existential risk and something as major as a shift of an order of magnitude, I really think it’d be helpful to note and explain any change in your thinking.
Thanks for pointing this out (and for the support).
We only update the ‘Last updated’ field for major updates not small ones. I think we’ll rename it ‘Last major update’ to make it clearer.
The edit you noticed wasn’t intended to indicate that we’ve changed our view on the effectiveness of existential risk reduction work. That paragraph was only meant to demonstrate how it’s possible that x-risk reduction could be competitive with top charities from a present-lives-saved perspective. The author decided we could make this point better by using illustrative figures that are more conservative than 80k’s actual rough guess and made the edit. We’ve tinkered with the wording to make it clearer that they are not actual cost-effectiveness estimates.
Also, note that in both cases the paragraph was about hypothetical effectiveness if you only cared about present lives, which is very different from our actual estimate of cost effectiveness.
Something else I hope you’ll update is the claim in that section that GiveWell estimates that it costs the Against Malaria Foundation $7,500 to save a life.
The archived version of the GiveWell page you cite does not support that claim; it states the cost per life saved of AMF is $5,500. (It looks like earlier archives of that same page do state $7,500 (e.g. here), so that number may have been current while the piece was being drafted.)
Additionally, the $5,500 number, which is based on GiveWell’s Aug. 2017 estimates (click here and see B84), is unusually high. Here are GiveWell’s estimates by year:
2017 (final version): $3,280 (click here and see B91)
2018 (final version): $4,104 (click here and see R109)
2019 (final version): $2,331 (click here and see B162) (downside adjustments seem to cancel with excluded effects)
2020 (Sep. 11th version): $4,450 (click here and see B219)
Once the AMF number is updated, the near-term existential risk number is less than five times as good as the AMF number. And if the existential risk number is adjusted for uncertainty (see here and here), then it could end up worse than the AMF number. That’s why I assumed the change on the page represented a shift in your views rather than an illustration. It puts the numbers so close to each other that it’s not obvious that the near-term existential risk number is better and it also makes it easier for factors like personal fit to outweigh the difference in impact.
Hi Arden and the 80,000 Hours team,
Thank you for the excellent content that you produce for the EA community, especially the podcasts.
There is one issue that I want to raise. I gave serious thought to raising this via your survey, but I think it is better raised publicly.
In your article “The case for reducing extinction risk” (which is linked to in your “Key ideas” article), you write:
At the top of the page, it says the article was published in October 2017 and last updated in October 2017. There are no footnotes indicating any changes were made to that section.
However, an archived copy of the article from June 2018 shows that, at the time, the article read:
I think it would be helpful to members of the community to indicate when and how an article has been substantively updated. There are many ways this can be done, including:
an article explaining how and why your views have changed (e.g. here, here/here, and here);
linking to an archived version of the article (as you do here) ideally with a change log; and
a footnote in the section indicating what it previously said and why your views have changed.
I understand that you have a large amount of content and limited staff capacity to review all of your old content. But what I’m talking about here is limited to changes you choose to make.
I’m sure it was just an oversight on the part of whoever made the change. You all have a lot on your plate, and it’s most convenient for an article to just present your current views on the subject.
But when it comes to something as important as the effectiveness of spending to reduce existential risk and something as major as a shift of an order of magnitude, I really think it’d be helpful to note and explain any change in your thinking.
Thank you for reading, and keep up the good work.
Hi RandomEA,
Thanks for pointing this out (and for the support).
We only update the ‘Last updated’ field for major updates not small ones. I think we’ll rename it ‘Last major update’ to make it clearer.
The edit you noticed wasn’t intended to indicate that we’ve changed our view on the effectiveness of existential risk reduction work. That paragraph was only meant to demonstrate how it’s possible that x-risk reduction could be competitive with top charities from a present-lives-saved perspective. The author decided we could make this point better by using illustrative figures that are more conservative than 80k’s actual rough guess and made the edit. We’ve tinkered with the wording to make it clearer that they are not actual cost-effectiveness estimates.
Also, note that in both cases the paragraph was about hypothetical effectiveness if you only cared about present lives, which is very different from our actual estimate of cost effectiveness.
Hope this helps clear things up.
Thanks Howie.
Something else I hope you’ll update is the claim in that section that GiveWell estimates that it costs the Against Malaria Foundation $7,500 to save a life.
The archived version of the GiveWell page you cite does not support that claim; it states the cost per life saved of AMF is $5,500. (It looks like earlier archives of that same page do state $7,500 (e.g. here), so that number may have been current while the piece was being drafted.)
Additionally, the $5,500 number, which is based on GiveWell’s Aug. 2017 estimates (click here and see B84), is unusually high. Here are GiveWell’s estimates by year:
2017 (final version): $3,280 (click here and see B91)
2018 (final version): $4,104 (click here and see R109)
2019 (final version): $2,331 (click here and see B162) (downside adjustments seem to cancel with excluded effects)
2020 (Sep. 11th version): $4,450 (click here and see B219)
Once the AMF number is updated, the near-term existential risk number is less than five times as good as the AMF number. And if the existential risk number is adjusted for uncertainty (see here and here), then it could end up worse than the AMF number. That’s why I assumed the change on the page represented a shift in your views rather than an illustration. It puts the numbers so close to each other that it’s not obvious that the near-term existential risk number is better and it also makes it easier for factors like personal fit to outweigh the difference in impact.