As the fundamental engine of economic growth and human prosperity, science and technology not only advance in breadth, opening up more novel areas for exploration, they also progress in depth, developing increasingly sophisticated solutions for specific tasks. Yet our quantitative understanding of the record-breaking dynamics remains limited. The fishing model of innovation assumes that new ideas are sequentially drawn from a fixed distribution and records are exponentially hard to break, while the cumulative cultural evolution theory predicts rapid incremental improvements through social learning. Here we collect over 6M records from diverse domains to systematically quantify the record-breaking dynamics in science and technology. Our empirical analysis reveals three key results:
- (1) The number of record-breaking events grows systematically faster than a logarithm function, which rejects the fishing model of innovation.
- (2) At the same time, the record-breaking dynamics is highly punctuated, where the waiting time until the next record follows a power-law distribution with infinite expectation, suggesting long streaks of stagnation that cannot be explained by the cumulative cultural model.
- (3) The empirical record-breaking sequence exhibits systematic temporal correlation, where recent progression (or stagnation) predicts future progression (or stagnation), featuring short-term predictability and long-term uncertainty of technological progress.
Together, these results offer a new quantitative basis to understand and predict advances in science and technology, which not only deepens our understanding of complex social systems, but also holds important implications for innovators and policymakers.
This event is open to faculty, doctoral students, academic researchers, and graduate students.