The curated weekly English Wikipedia Top 25 Report shows no faster turnover in 2016–2026, and its 2013–2017 increase matches source changes and the loss of perennial pages

Question: did the turnover of the most viewed English Wikipedia articles become faster at a weekly time scale from 2013 to 2026, in particular in 2013-2016, when Lorenz-Spreen et al. (2019) found faster attention on Twitter (top-50 dwell about -12% per year)? Data: 717 consecutive weekly lists of the Wikipedia Top 25 Report (January 2013 to September 2026), parsed from the English Wikipedia dump. Editors make this list and remove entries with suspect automated traffic. The report text documents its source periods: desktop views (to September 2014), added mobile views (October 2014), new WMF pageview data (August 2016) and pageview tools (from January 2020). Method: weekly turnover and one-week entries for the top 10 and top 25; OLS trends with Newey-West errors, trends inside source periods, steps at source changes, and a variant without perennial pages (listed 26 or more weeks in a year). We compared the lists with weekly lists that we computed from Wikimedia daily top-1000 data, and we repeated the daily-list analysis of earlier SCI studies for five editions. Results: in 2016-2025 weekly turnover decreased slightly (top 25: -0.36 pp per year, 95% interval -0.73 to +0.02; top 10: -0.61, -1.20 to -0.01). In 2013-2017 top-25 turnover increased (+2.46 pp per year, +1.26 to +3.66), but no source period had a significant increase inside it; with steps at the source changes the trend was +0.53 (-3.20 to +4.26), and without perennial pages (Facebook, Google, YouTube and others held 17.6% of places in 2013 and about 4% from 2017) it was +1.06 (-0.11 to +2.23). The top 10 had no increase. Even without corrections the upper bound (+6.0% of the mean per year) is less than half of the Twitter-equivalent turnover rate (+13.7% per year). 82% of the curated top 25 was also in the API weekly top 25 (214 weeks). The daily lists of en, de, fr, es and ja (mobile web, 2016-2025) showed no faster turnover; the largest upper bound was +1.3% per year. Limitations: one curated weekly list of one edition; curation rules changed; trend and source steps are not separable in 2013-2017; the Twitter rate is only a reference from another platform and time scale.

collective attention · Wikipedia · pageviews · Top 25 Report · popularity turnover · attention acceleration · social dynamics · data source changes · replication · time series