A study spanning BBC sports coverage, Reddit worldnews discussions and US box-office records found little sustained evidence that a new hit displaces attention from existing items. The overall hit effects were small and not statistically significant, although Reddit showed an approximately 40% drop in background activity at the peak of a hit.
The work is an arXiv version-1 preprint dated 28 September 2026. Its central question was whether high-attention new items reduce attention received by existing items, tested through a natural-experiment comparison of hit periods with baseline periods in which no new item appeared.
How the comparison worked
Attention was measured as new interactions in two consecutive windows. The analysis summarized the change with a multiplicative slowing factor, which describes the change in expected activity from one window to the next. News was measured in 10-minute windows, while cinema was measured in one-week windows. Cinema baseline periods excluded weeks with a new movie that later succeeded at or above the 80th percentile.
To quantify uncertainty, before-and-after interaction counts were modeled with a Poisson likelihood. The slowing factor was estimated by maximum likelihood, confidence intervals were obtained by non-parametric bootstrap, and hit-versus-baseline differences were assessed with a z-score and two-tailed p-value.
The BBC dataset contained 3,087 sports articles, 852,400 comments and 67,527 readers collected from 1 October 2018 to 30 June 2019; comments were time-stamped at one-minute resolution. The Reddit analysis covered approximately 56,000 worldnews posts and 2.3 million comments, defining hits as the 25 most-commented submissions in the three-month period. The cleaned US box-office dataset covered 1983 to 2018 and included 14,000 movies; weekly data were used to smooth within-week variation.
The crowd usually stayed with the hit
In the BBC data, the baseline slowing factor was close to 0.95, with a characteristic decay timescale of 195 minutes. Hit articles received twice as many comments as other items in the first 10 minutes, but the hit-period slowing factor was only about 2% smaller than baseline. The difference was not statistically significant: p = 0.67.
Reddit produced a large rise in total activity, but it was concentrated on the hit post. Total comment rate increased by approximately 3.5-fold, while non-hit posts remained near 20 comments per minute. Baseline and hit slowing factors were approximately 0.97 and 0.95, a difference of about 2% that did not amount to a statistically significant sustained effect.
At the hit peak, however, the background ratio fell to 0.60, compared with 1.97 in baseline, an approximately 40% drop.
Cinema showed a smaller shift
Cinema showed a modest pattern. The baseline slowing factor was 0.58, with a 12.9-day decay timescale. In hit periods, the immediate decrease was less than 0.02 and not statistically significant; differences at later time points became significant from week 3, while the compounded difference over five weeks was about 5%. Weekly movie gross income served as a proxy for attention or the number of interactions.
To test the sensitivity of the measurement, the researchers ran synthetic FAE-model simulations calibrated to the BBC data. They varied the model's collective-attention elasticity from 0, representing no elasticity, to 1, representing perfect elasticity. The measurement detected baseline-versus-hit differences across most settings, but the difference became non-significant at elasticity values approximately 0.96 or higher, where p-values were greater than 0.05.
What the result does not settle
The findings need to be read as associations. The comparison used naturally occurring hit and baseline periods rather than randomized assignments, so the analysis cannot establish that a hit itself changed activity around other items. The non-significant overall results should not be read as proof that competition is absent, especially when the news comparison used 10-minute windows.
Taken together, the evidence supports a restrained picture of collective attention. Hits became much more active themselves in the BBC and Reddit measures without a clear sustained reduction in activity elsewhere. The Reddit peak shows that a brief point in the trajectory can look different from the broader comparison, while the cinema result points to a small measured difference over five weeks. The preprint therefore offers evidence about three systems, not a universal rule for every cultural or economic setting.
Paper data and sources
Original title: Measuring the impact of hits
Authors: Matúš Medo, Liudmila Rozanova
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-09-28
DOI: Not available
Original paper · Full text