arXiv cs.CY· Fengming Liu·· 9 小时前AI 评分30
Reddit 话语同质化研究再分析:保守派“转向一致”是文本长度伪影
Asymmetric Discourse Homogenization on Reddit Was a Text-Length Artifact: A Reanalysis
AI 导读
一篇 Reddit 研究再分析发现,原论文所称“ChatGPT 发布后保守派用户写作趋同”是相似度指标随评论长度波动的伪影。作者用句子编码器在全部 6,056,593 条评论上按长度分层重算离散度,保守派的断点消失,两派组内相似度变化无可靠差异,ChatGPT 日期与两千多个其他切分点无异。唯一稳健的变化是 2019 至 2022 年左右翼评论相互靠拢,且该趋同在 ChatGPT 出现前已完成。
正文 · 原文
Abstract:In the first version of this paper I reported that conservative Reddit users began writing more alike after ChatGPT was released while progressive users did not. That result came from a similarity measure that rises and falls with comment length. I redo every analysis of the original paper on all 6,056,593 comments with a sentence encoder and a dispersion measure computed within length strata, and the conservative break goes away. The two camps show no reliable difference in how within-group similarity changed after late 2022, and the ChatGPT date behaves like any of the more than two thousand other cutoffs I test. The matched flair comparison, the cumulative AI index, the stayer panel and the entropy signature each lose their support once length is held fixed. The one change that survives is older. Left and right comments drew closer to each other between 2019 and 2022, and that convergence was complete before ChatGPT arrived. I end with what this implies for studies that measure homogenization with embeddings.
| Comments: | 18 pages, 4 figures, 5 tables. This version replaces v1 entirely: the v1 headline finding (asymmetric conservative homogenization after ChatGPT) was an artifact of comment length in the similarity measure. The artifact is documented in general form in arXiv:2611.21519 |
| Subjects: | Computers and Society (cs.CY); Computation and Language (cs.CL); Social and Information Networks (cs.SI) |
| Cite as: | arXiv:2608.13674 [cs.CY] |
| (or arXiv:2608.13674v2 [cs.CY] for this version) | |
| https://doi.org/10.48550/arXiv.2608.13674 arXiv-issued DOI via DataCite |
Submission history
From: Fengming Liu [view email]
[v1]
Thu, 13 Aug 2026 18:13:35 UTC (194 KB)
[v2]
Tue, 17 Nov 2026 17:58:50 UTC (126 KB)
来源:arXiv cs.CY · arxiv.org