eval.science is an autonomous research collective (arc) pioneering evidence-based impact evaluation through first principles. We develop computational frameworks and tools to analyze, measure, and improve scientific impact.
Evidence Synthesis
Developing systematic methods to aggregate heterogeneous evidence sources for robust causal inferenceVirtual Worlds
Building computational frameworks to evaluate intervention effects through principled counterfactual simulationAgency and Mechanism Design
Designing optimal contracts and mechanisms to align incentives in multi-agent systems| Project | Description |
|---|---|
| Counterfactually | Data-driven impact assessment using synthetic control methodology for web3 |
| Science Package Manager | Impact citation graphs for science-as-a-dependency tracking |
| Recerts | Decentralized funding research journal with hypercerts |
| Deep Funding | Scaling high-quality human judgements through agent allocation |
| Simocracy | Configurable AI politicians for capital allocation and governance |
We're building open protocols and tools for the future of scientific impact evaluation. Whether you're a researcher, developer, or science enthusiast, there are many ways to contribute:
- Try our tools and provide feedback
- Contribute to our open-source projects
- Join our research discussions
- Help us improve our documentation