The Alien Space of Science: Sampling Coherent but Cognitively Unavailable Research Directions
Summary
The paper argues that scientific discovery is constrained by what is cognitively available to researchers and proposes an ‘alien space of science’—directions that are scientifically plausible but unlikely given current research communities. It decomposes papers into concept units (idea atoms) and uses a coherence model to score viable directions while an availability model scores whether any existing author community is positioned to pursue them, then ranks combinations to maximize coherence and minimize availability. Empirical results on 16,068 peer-reviewed AI/ML/NLP papers show more diverse research portfolios and potential AI-assisted ideation that complements human science.