The General Longtermism team at Rethink Priorities is interested in generating, fleshing out, prioritizing, and incubating longtermist megaprojects.
But what are longtermist megaprojects? In my mind, there are tentatively 4 criteria:
Scale of outcomes: The outputs should be large in scope, from a longtermist point of view. A project at scale should have a decent shot of reducing probability of existential risk by a large amount. Perhaps we believe that after considerable research, we have a decent chance of concluding that the project will reduce existential risk by >0.01% (a “basis point”).*
Cost-effectiveness: The project should have reasonable cost-effectiveness. Given limited resources, we shouldn’t spend all of them on a project that cost a lot of money and human capital for merely moderate gain. My guess is that we ought to have a numerical threshold of between 100M and 3B (best guess for current target: 500M) for financial plus human capital cost per basis point of existential risk averted.
Scale of inputs: The inputs should be fairly large as well. An extremely impressive paper or a single conversation with the right person, no matter how impactful, should not count as a longtermist megaproject. My guess is that we should have a numerical threshold for minimal investment before something counts as a megaproject, perhaps between the 50M and 1B region (best guess: 100M).
Type of inputs: Finally, for the project to qualitatively “feel” like a megaproject, I think the highest-cost inputs should be mostly (>70%?) financial capital and junior human capital, rather than senior human capital.
If people are interested, I’m happy to elaborate in more detail about any of these criteria, or explain why I came to conclude that they should be the right criteria, and not others.
*though arguably relative percentage reduction of x-risk is a better metric than absolute reduction. I find myself confused about this.
The General Longtermism team at Rethink Priorities is interested in generating, fleshing out, prioritizing, and incubating longtermist megaprojects.
But what are longtermist megaprojects? In my mind, there are tentatively 4 criteria:
Scale of outcomes: The outputs should be large in scope, from a longtermist point of view. A project at scale should have a decent shot of reducing probability of existential risk by a large amount. Perhaps we believe that after considerable research, we have a decent chance of concluding that the project will reduce existential risk by >0.01% (a “basis point”).*
Cost-effectiveness: The project should have reasonable cost-effectiveness. Given limited resources, we shouldn’t spend all of them on a project that cost a lot of money and human capital for merely moderate gain. My guess is that we ought to have a numerical threshold of between 100M and 3B (best guess for current target: 500M) for financial plus human capital cost per basis point of existential risk averted.
Scale of inputs: The inputs should be fairly large as well. An extremely impressive paper or a single conversation with the right person, no matter how impactful, should not count as a longtermist megaproject. My guess is that we should have a numerical threshold for minimal investment before something counts as a megaproject, perhaps between the 50M and 1B region (best guess: 100M).
Type of inputs: Finally, for the project to qualitatively “feel” like a megaproject, I think the highest-cost inputs should be mostly (>70%?) financial capital and junior human capital, rather than senior human capital.
If people are interested, I’m happy to elaborate in more detail about any of these criteria, or explain why I came to conclude that they should be the right criteria, and not others.
*though arguably relative percentage reduction of x-risk is a better metric than absolute reduction. I find myself confused about this.