“What I would do to improve this” should read “what I would do to build upon this”. I think that most of the value of this particular estimation lies in the scenario where I/others use it to build a selection of impacts for various career paths, which allows for relative value comparisons between career paths.
Ok, so here is how I am thinking about this.
What Joel is doing
You can divide your estimate between:
The impact per year of an Open Philanthropy-equivalent grantmaking role.
The chance you can get one such, and the number of years.
To estimate 1., you look at:
The number of funds you would be responsible for
The share of the impact that those funds cause that can be attributed to you, e.g., vs your counterfactual
Other benefits from being a good grantmaker.
Their conversion between funds affected and basis points of x-risk.
To estimate 2., you look at:
The number of organizations, and their relative value vs Open Philanthropy
The probability that you would get hired by each
Nitpicks:
I am not sure whether you are considering that if you arrive at a smaller org, you will probably be responsible for a greater % of their funding.
Probabilities of getting hired are not independent, e.g., a “desirable grantmaker” would probably be
Probability of getting hired probably should be a beta, rather than I lognormal? Feels right, though I don’t really have many intuitions for the beta
What I would do to improve this.
I think I’d first simplify the problem and look at only 1., and leave 2. (probability of getting a job, number of such jobs available) for later.
Then, I’d want to:
Have a really clear model, which can be very well maintained.
Get better estimates of how much money each grantmaker at Open Philanthropy is moving
Have the model output all of:
Dollars of money the grantmaker will move
Dollars the grantmaker would be responsible for (probably including Shapley values)
Expected reduction in x-risk, but based on looking at past OP grants, rather than relying on Linch’s estimates
“What I would do to improve this” should read “what I would do to build upon this”. I think that most of the value of this particular estimation lies in the scenario where I/others use it to build a selection of impacts for various career paths, which allows for relative value comparisons between career paths.
Thank you for feedback, Nuno!
I totally agree that part 2 is an unhelpful distraction at this stage. (Agreed with your other points at time of writing!)