Benefit: 243k increase in counterfactual funds raised for effective charities
Report: 411k in counterfactual matched donations …
275k from 100% match. Previous report suggests no EA donors would get the 100% match without the EA GT org. This seems wrong to me. Maybe EAGT made this happen in the past, but in the future many of the institutions and knowledge are in place … so let’s say 100k would be raised from this without EAGT going forward.
--> 175k counterfactual impact of 80k
$136k of which was from 10% match … ~ 40% of which obtained without EA GT (see discussion in that report) -->
--> 81k counterfactual impact of 80k
81 + 175 = 256k
Less 17% lost tax benefits, which I’ll estimate at 25% … so subtract about 5% of the total
$25,000 in costs (Hours organizing, other expenses)
466 + 297 = 763 paid and unpaid hours organizing this.
I suspect that as we learn more, get better, have the sheets etc. in place, it will take fewer hours.
… So let’s say 450 total hours going forward.
I’m not sure what type of labor goes into this. Let’s say 150 hours of ‘managerial and tech time’ valued at $100 per hour, and 300 hours of ‘volunteer/student time’ valued at $30/hour.[1]
150 * 100 + 300 * 30 = $24,000
Nonlabor costs … about $1000
Not considered:
chances Facebook continues GT matches
extent to which this leads counterfactual donations to be made
less tangible benefits,
cost of the time spent by donors (est: 411 donors spending ~30 min each = 200 hours = maybe 10k in value of time?)
Overall first-pass assessment
This seems like a potentially good use of resources. ~243k in increased amounts received by to EA charities per year. Let’s say these have 10x the value of the counterfactual matched charities, so a this is worth $219k per year.
Relative to perhaps $25-35k in time costs? Or, if I’m wrong about the ‘learn by doing time savings’, maybe $50-60k in time costs.
Probably worth doing, or worth further investigation (including perhaps a MonteCarlo Fermi using Squiggle or something).
(Of course some might say EA hours are super valuable, on the other hand people get something out of this, it’s social, and it may not substitute for time spent solving X-risk issues etc.
Hi David! I apologize for the very slow response. A few points: - Your analysis makes me upgrade how important I think diligent time tracking is on this project in future years, segmented by e.g., ‘managerial and tech time’ vs ‘volunteer/student time’ - I don’t have a go-to answer for you on the time costs for EA GT 2021. We had 2 Ops Specialists (Aisha and Mac) each work ~200 paid hours; I worked about 350 paid hours (including hiring and training); Avi worked probably a few hundred volunteer hours (including hiring and training); Gina and a few others worked a small amount of volunteer hours. - Can the project’s time costs decrease via “learn by doing?” I am somewhat optimistic about this. But it’s tricky because historically, new people have had to be trained on the systems and context every year. So processes can be improved, but a big thing is getting the same people to contribute to the project year after year. And this is tough, because it’s uncertain the project will run any given year, and it’s only seasonal. Ideally, the “institutional knowledge” would sit at an EA org (ideally, with the same people) over the long term. - Thanks again for your BOTEC, I enjoyed reading it and I imagine it has helped folks in the community evaluate the projects’ value.
Q: Where can we see the ‘bottom line’ on the impact for the most recent years?
I’m looking for.
Additional amounts raised/diverted relative to counterfactual (with no EA GT org)
minus
Cost (money and value of time) of this
As I didn’t see anything in the linked posts etc., I sketched one below (which took about 20 minutes):
Proto-BOTEC sort of for 2020, sort of going forward
A quick skim and proto-Botec from the recent linked report for 2020
Benefit: 243k increase in counterfactual funds raised for effective charities
Report: 411k in counterfactual matched donations …
275k from 100% match. Previous report suggests no EA donors would get the 100% match without the EA GT org. This seems wrong to me. Maybe EAGT made this happen in the past, but in the future many of the institutions and knowledge are in place … so let’s say 100k would be raised from this without EAGT going forward.
--> 175k counterfactual impact of 80k
$136k of which was from 10% match … ~ 40% of which obtained without EA GT (see discussion in that report) --> --> 81k counterfactual impact of 80k
81 + 175 = 256k
Less 17% lost tax benefits, which I’ll estimate at 25% … so subtract about 5% of the total
$25,000 in costs (Hours organizing, other expenses)
466 + 297 = 763 paid and unpaid hours organizing this. I suspect that as we learn more, get better, have the sheets etc. in place, it will take fewer hours.
… So let’s say 450 total hours going forward.
I’m not sure what type of labor goes into this. Let’s say 150 hours of ‘managerial and tech time’ valued at $100 per hour, and 300 hours of ‘volunteer/student time’ valued at $30/hour.[1]
150 * 100 + 300 * 30 = $24,000
Nonlabor costs … about $1000
Not considered:
chances Facebook continues GT matches
extent to which this leads counterfactual donations to be made
less tangible benefits,
cost of the time spent by donors (est: 411 donors spending ~30 min each = 200 hours = maybe 10k in value of time?)
Overall first-pass assessment
This seems like a potentially good use of resources. ~243k in increased amounts received by to EA charities per year. Let’s say these have 10x the value of the counterfactual matched charities, so a this is worth $219k per year.
Relative to perhaps $25-35k in time costs? Or, if I’m wrong about the ‘learn by doing time savings’, maybe $50-60k in time costs.
Probably worth doing, or worth further investigation (including perhaps a MonteCarlo Fermi using Squiggle or something).
(Of course some might say EA hours are super valuable, on the other hand people get something out of this, it’s social, and it may not substitute for time spent solving X-risk issues etc.
Hi David! I apologize for the very slow response. A few points:
- Your analysis makes me upgrade how important I think diligent time tracking is on this project in future years, segmented by e.g., ‘managerial and tech time’ vs ‘volunteer/student time’
- I don’t have a go-to answer for you on the time costs for EA GT 2021. We had 2 Ops Specialists (Aisha and Mac) each work ~200 paid hours; I worked about 350 paid hours (including hiring and training); Avi worked probably a few hundred volunteer hours (including hiring and training); Gina and a few others worked a small amount of volunteer hours.
- Can the project’s time costs decrease via “learn by doing?” I am somewhat optimistic about this. But it’s tricky because historically, new people have had to be trained on the systems and context every year. So processes can be improved, but a big thing is getting the same people to contribute to the project year after year. And this is tough, because it’s uncertain the project will run any given year, and it’s only seasonal. Ideally, the “institutional knowledge” would sit at an EA org (ideally, with the same people) over the long term.
- Thanks again for your BOTEC, I enjoyed reading it and I imagine it has helped folks in the community evaluate the projects’ value.
Thanks for taking the time to complete and share a first-pass assessment, David! I’ll follow up with a bit more info when I’m able.
And thanks for all the work you have done on this project!
I appreciate that—thanks! I have worked a lot on it. A lot of the credit goes to my great EA GT teammates, in present and past years.