A service/consultancy that calculates the value of information of research projects
Epistemic Institutions, Research That Can Help Us Improve
When undertaking any research or investigations, we want to know whether it’s worth spending money or time on it. There are a lot of research-type projects in EA and the best way to evaluate and prioritise them is to calculate their value of information (VOI). However, VoI calculations can be complex and we need to build a team of experts that can form a VoI consultancy or service provider.
Examples of use cases: 1. A grant maker wants to know whether it’s worth spending 0.5FTE on investigating cause area Y vs cause area X. 2. A thinktank has generated a list of policy ideas to investigate but is uncertain which to prioritise. 3. A research org also has a list of research questions but want to know which one has the highest VoI.
In each of this use case, I suspect a VoI consultancy can be extremely valuable.
I think there might be harder meta-problem: should we even spend time and money on calculating the VoI of certain investigations? A failure mode is where the VoI consultancy calculates a bunch of research projects that turn out to have very low VoI.
I guess figuring out baseline, the cost of doing VoI calculations, and having a cheap heuristic as a preliminary calculation could help, but I’m highly uncertain.
A service/consultancy that calculates the value of information of research projects
Epistemic Institutions, Research That Can Help Us Improve
When undertaking any research or investigations, we want to know whether it’s worth spending money or time on it. There are a lot of research-type projects in EA and the best way to evaluate and prioritise them is to calculate their value of information (VOI). However, VoI calculations can be complex and we need to build a team of experts that can form a VoI consultancy or service provider.
Examples of use cases:
1. A grant maker wants to know whether it’s worth spending 0.5FTE on investigating cause area Y vs cause area X.
2. A thinktank has generated a list of policy ideas to investigate but is uncertain which to prioritise.
3. A research org also has a list of research questions but want to know which one has the highest VoI.
In each of this use case, I suspect a VoI consultancy can be extremely valuable.
David Manheim has written more about VoI here.
I think there might be harder meta-problem: should we even spend time and money on calculating the VoI of certain investigations? A failure mode is where the VoI consultancy calculates a bunch of research projects that turn out to have very low VoI.
I guess figuring out baseline, the cost of doing VoI calculations, and having a cheap heuristic as a preliminary calculation could help, but I’m highly uncertain.