I spoke to eight early-career social scientists for this post. I tried to cover as wide of a perspective as possible—the interviewees’ experience ranged from prospective PhD students to assistant professors; their fields ranged from economics, political science, to sociology, computational social science, and information science. Of course, I’m drawing heavily from my own experience as a researcher in political science.
# Summary
## What is the problem?
The [credibility revolution](https://en.wikipedia.org/wiki/Credibility_revolution) of the 2000s brought causal inference methods into social science. The data transparency and pre-registration requirements implemented by journals in the 2010s led to the public sharing of data and experimental protocols. The digital revolution happening around the same time made society generally much more legible (e.g. Facebook social graphs, Google mobility trends), and the cost for doing online experiments came down to $1 per respondent.
Taken together, the face of social science has changed a lot—it went from solo-authors building theories to teams of people collecting, analyzing, and sharing data. During this time, scholars’ real-world influence likely increased as well, with policymakers and philanthropists like EA-inspired folks using estimates from empirical studies to compare different proposals.
I argue in this post that more funding should be allocated to radically modernizations of the social science infrastructure so that data can be more easily discovered and re-used, peer review is made more efficient and rigorous, and communication of of study findings is more timely and accessible.
## Who is already working on it?
Various non-profits, for-profits, and academic teams are working on various aspects of modernizing the research infrastructure.
I argue, however, that more is needed because 1) empirical social science requires tools specialized tools; tools built for medical or computer science research aren’t sufficient; 2) private entities and academic initiatives will end up under-supplying infrastructure innovations because the returns are largely public and radical improvements take much more than a few academics working on the side.
## What could a new philanthropist do?
A new philanthropist could fund more organizations such as [Ought/Elicit](https://ought.org/elicit) that rebuilds the research infrastructure from the ground up with 21st-century technologies and 21st-century mechanism designs. For example, we could fund organizations that: - build a new data sharing regime (e.g. create a social science data search engine and standardize data formats); - rethink peer review (e.g. build a platform of semi-anonymous pre-publication review that eventually replaces academic journals and communicates research findings to a broader audience using large language model-powered text summarization; design incentives that reward efficient and expert peer review) - enable more collaborations that bridge geographic distances and disciplinary boundaries - …
Cause Exploration Prize: Modernizing the Social Science Research Infrastructure
# Editorial note
I spoke to eight early-career social scientists for this post. I tried to cover as wide of a perspective as possible—the interviewees’ experience ranged from prospective PhD students to assistant professors; their fields ranged from economics, political science, to sociology, computational social science, and information science. Of course, I’m drawing heavily from my own experience as a researcher in political science.
# Summary
## What is the problem?
The [credibility revolution](https://en.wikipedia.org/wiki/Credibility_revolution) of the 2000s brought causal inference methods into social science. The data transparency and pre-registration requirements implemented by journals in the 2010s led to the public sharing of data and experimental protocols. The digital revolution happening around the same time made society generally much more legible (e.g. Facebook social graphs, Google mobility trends), and the cost for doing online experiments came down to $1 per respondent.
Taken together, the face of social science has changed a lot—it went from solo-authors building theories to teams of people collecting, analyzing, and sharing data. During this time, scholars’ real-world influence likely increased as well, with policymakers and philanthropists like EA-inspired folks using estimates from empirical studies to compare different proposals.
I argue in this post that more funding should be allocated to radically modernizations of the social science infrastructure so that data can be more easily discovered and re-used, peer review is made more efficient and rigorous, and communication of of study findings is more timely and accessible.
## Who is already working on it?
Various non-profits, for-profits, and academic teams are working on various aspects of modernizing the research infrastructure.
I argue, however, that more is needed because 1) empirical social science requires tools specialized tools; tools built for medical or computer science research aren’t sufficient; 2) private entities and academic initiatives will end up under-supplying infrastructure innovations because the returns are largely public and radical improvements take much more than a few academics working on the side.
## What could a new philanthropist do?
A new philanthropist could fund more organizations such as [Ought/Elicit](https://ought.org/elicit) that rebuilds the research infrastructure from the ground up with 21st-century technologies and 21st-century mechanism designs. For example, we could fund organizations that:
- build a new data sharing regime (e.g. create a social science data search engine and standardize data formats);
- rethink peer review (e.g. build a platform of semi-anonymous pre-publication review that eventually replaces academic journals and communicates research findings to a broader audience using large language model-powered text summarization; design incentives that reward efficient and expert peer review)
- enable more collaborations that bridge geographic distances and disciplinary boundaries
- …