Wikipedia as a Sociological Phenomenon: Reliability, Bias, and the Democratization of Knowledge

抄録

In its more than twenty years of existence, Wikipedia has been used as the default general reference work of the internet. But the common opinion tends to go either way, either the ultra-idealized wisdom of the crowd or an anarchic, unreliable website. This paper will argue that neither side is sufficient. Rather, Wikipedia can most appropriately be regarded as a sociological phenomenon: a sociotechnical system in which human editors, automated bots, human policies and power relations co-create what is defined as knowledge. We investigated 80 articles on the English Wikipedia in two categories: high consensus scientific/historical articles and politically controversial ones, by using a mixed methods approach relying solely upon the publicly available Wikipedia data (edit histories, talk pages, editor metadata, and revision logs) from 2021 to 2025. Findings indicate that reliability is not uniform: the scientific articles are highly consistent and stable, with low reversion rates, whereas controversial articles are associated with the edition wars and bias in framing. Structural disparities are acute: There are merely 14.2% publicly female top editors, whereas 68% of the talk page discourse is dominated by editors based in the Global North countries. Compared to the Arabic Wikipedia, gender and geographic concentration are even more significant. Although biased, Wikipedia is used as a source of critical information infrastructure where they have little to no access to institutions. We come to the conclusion that Wikipedia is epistemically as effective as the user is digitally literate, and as many as possible are working to reduce systemic bias in Wikipedia.