Daily top-50 turnover did not increase in eight Wikipedia language editions from 2016 to 2025; it decreased in English and Russian and had no clear trend in six others

Question: does the slowdown of collective attention reported for English Wikipedia (sci:sha256:d88df5ed97e71c69813236e7fc57a74920fb497209326818f8096ed355bf8cdc) occur in other language editions, or does any edition show the acceleration that Lorenz-Spreen et al. (2019) reported for other platforms? Data: Wikimedia daily lists of the 1,000 most-viewed articles (user traffic). English all-access for every day of 2016-2025, and sampled blocks (days 1-28 of February, May, August and November; 1,120 days) for English mobile-web and German, Japanese, Spanish, French, Russian, Italian and Portuguese. Method: after removal of non-article pages, we calculated for the daily top 50 the daily turnover, the fraction of article-days with at least 2x the views of the previous or next day, the fraction of bursts that lose half of their views in one day, and the mean days in the top 50. The measures are exact under the censoring of the lists. Trends are linear fits to ten annual values (t-based 95% intervals). Results: turnover did not increase in any edition. English turnover decreased from 0.456 to 0.389 (-2.1% per year, 95% CI -3.2 to -1.0), and large gains, large losses and one-day burst decay also decreased. This reproduces sci:sha256:d88df5ed97e71c69813236e7fc57a74920fb497209326818f8096ed355bf8cdc with independent code and holds for top 25 and 100, mobile-web data, sampled blocks and novel articles. Russian turnover also decreased (-3.0%/yr, CI -4.9 to -1.2). German, French, Italian and Portuguese had no clear trend in any measure. Spanish had more large gains (+2.1%/yr, CI +0.8 to +3.5) with no turnover trend, and Japanese had weak signs of faster change that do not survive adjustment for multiple tests. Limitations: only the most-viewed articles, daily resolution and Wikimedia user traffic (classification changed in 2020); no bot filter outside English; non-English results use 112 days per year; ten annual points per trend; causes not tested.

collective attention · Wikipedia · pageviews · social acceleration · turnover · burst dynamics · language editions · popularity dynamics · computational social science · replication