Li & Cao's IKN/CCL model shows AI intermediaries distort citation metrics

Li Li and Yu Cao's new paper, "Citation Pathways in the AI Era," identifies a long-overlooked dimension in scientometrics: the intermediate nodes through which knowledge flows from source to citing author. The authors coin two key concepts: Interpretive Knowledge Nodes (IKN) — papers that reorganize classic works — and Citation Compression Layers (CCL) — stable publication identities that emerge when these intermediaries enter formal citation networks at scale. The central claim is that AI hasn't altered citation rules themselves, but has profoundly transformed the cost structure of producing such citable intermediaries. Under conditions of full compliance, the network-position effects of these pathways become a salient variable that can undermine the validity of standard impact metrics.

To substantiate their theory, the authors present a thought experiment involving a hypothetical journal R and a parsimonious "Citation Gravity" model. This model demonstrates how extreme scenarios — like widespread use of AI-generated summaries as references — can lead to institutional risk of incentive misalignment, where scholars are rewarded for citing intermediaries rather than original sources. The paper, submitted to arXiv on July 20, 2026, spans 9 pages and sits at the intersection of physics and society (physics.soc-ph) and computers and society (cs.CY). Its implications reach beyond academia: if citation counts can be artificially inflated through AI-compressed pathways, the entire system of research evaluation — from tenure metrics to funding decisions — faces a measurement boundary that current tools are unequipped to handle.