Matthew Effect
A cumulative advantage phenomenon where already highly-cited sources receive disproportionately more citations from LLMs.
Named after the biblical parable of the talents, the Matthew effect in LLM citations describes how sources that are already frequently cited in training data receive systematically more citations in model outputs. Research across 274,951 citations found a median citation count gap exceeding 1,300 between chosen and unchosen sources, creating a self-reinforcing cycle of authority.
Also known as
Matthew effect, cumulative advantage, rich-get-richer effect