Abstract:Scientific progress is widely understood as combinatorial, yet existing measures focus on the inputs a paper draws together, saying little about how knowledge is supplied for downstream reuse. Analyzing citation networks from tens of millions of publications in OpenAlex and the Web of Science, we decompose scientific contributions into three functional types, foundations, extensions, and generalizations, distinguishable by the roles they play in future science. Foundational and extensional works, which build and elaborate ideas within disciplines, dominated the post-war decades but declined steadily after the 1990s, while generalizations that catalyze innovation across many contexts rose sharply. Challenging the narrative of declining innovation inferred from the disruption index, our findings suggest the locus of innovation has shifted from within disciplines to between them, a transition the Internet and large language models may have facilitated. Our indices disentangle innovation modes conflated by existing measures, enabling reinterpretation of landmark findings of 'disruptive' science.
From: Hongbo Fang [view email]
[v1]
Mon, 22 Sep 2025 18:56:23 UTC (4,991 KB)
[v2]
Fri, 17 Oct 2025 23:13:44 UTC (5,371 KB)
[v3]
Fri, 10 Apr 2026 16:56:31 UTC (8,254 KB)
[v4]
Tue, 14 Apr 2026 13:00:47 UTC (8,386 KB)
[v5]
Sat, 8 Aug 2026 21:51:19 UTC (21,161 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。