From Göbekli Tepe to the World Wide Web, the story of civilization is the story of collaboration—the goal-oriented organization of resources. However, while the quantity, speed, and complexity of collaboration has grown exponentially in recent decades, our core resource allocation algorithms have failed to scale accordingly. Specifically, the current grant-and journal-based status quo evolved to meet the needs of a much smaller research community, which interacted asynchronously and was largely detached from the general public. In contrast, today’s over seven million researchers are distributed globally, and interact broadly and with low-latency.

In this talk, James will introduce the Scaling Science project, which aims to span this gap through the development of data-driven tools for the optimized allocation of scientific resources. By building a multi-source graph with over2 billon nodes and relationships, learning community structure in an unsupervised manner, and combining node centrality, metrics of community permeation, and methods from portfolio theory, we seek to (a) develop a distribution over future impact of new technologies and inventions, and (b) leverage these results in the construction of more optimal funding allocations.