Wikipedia top-list attention did not accelerate from 2016 to 2025 in English and German pageviews, and new English mobile bursts decayed more slowly

We tested whether collective attention on Wikipedia became faster from 2016 to 2025, as Lorenz-Spreen et al. (2019) reported for Twitter hashtags (top-50 residence time decreased by about 12% per year from 2013 to 2016). We used the Wikimedia Pageviews API daily top-1000 lists (user agent, 2015-07 to 2026-10) for English (four access types) and German (two access types). We removed non-article pages and measured the mean residence time in the daily top k (k = 10 to 500) and the daily turnover. For the trend we fit OLS on log values for the ten years 2016-2025. For English mobile-web, the top-50 residence time changed from 3.44 to 3.66 days (+1.5% per year, 95% CI -0.3 to +3.2). For German mobile-web it changed from 2.53 to 2.41 days (-0.4% per year, CI -1.1 to +0.3). Daily top-50 turnover decreased in both editions. Only the top-500 list showed a small decrease in residence time (-1.3% and -2.4% per year). We also sampled 2,351 new bursts (articles that entered the English mobile-web top 50 after 28 days outside the top 1000). The median peak-day share of excess views decreased from 0.62 to 0.47, and the median next-day retention ratio increased from 0.21 to 0.34. Thus, in these two editions, the data show no acceleration and show slower decay after peaks. Limits: undetected automated traffic (a large 2021-2023 peak on English desktop), only ten yearly points without a model of serial correlation, changes in how people access Wikipedia, and a language set that we reduced to two editions because of interrupted jobs.

collective attention · Wikipedia pageviews · social acceleration · popularity dynamics · attention bursts · top lists · turnover