Why does attention to web articles fall with time?

We analyze access statistics of a hundred and fifty blog entries and news articles, for periods of up to three years. Access rate falls as an inverse power of time passed since publication. The power law holds for periods of up to thousand days. The exponents are different for different blogs and are distributed between 0.6 and 3.2. We argue that the decay of attention to a web article is caused by the link to it first dropping down the list of links on the website’s front page, and then disappearing from the front page and its subsequent movement further into background. The other proposed explanations that use a decaying with time novelty factor, or some intricate theory of human dynamics cannot explain all of the experimental observations.

💡 Research Summary

The paper investigates how attention to online articles decays over time by analyzing access logs from 150 blog posts and news articles over periods extending up to three years (approximately 1,000 days). The authors find that the access rate A(t) follows a power‑law decline with respect to the elapsed time t since publication, i.e., A(t) = C·t⁻ᵅ, where C is a constant reflecting the initial visibility and α is a decay exponent that varies across sites. Empirically, α ranges from about 0.6 to 3.2, indicating that some articles fade slowly while others lose traffic very rapidly. The power‑law model fits the data remarkably well for the entire observation window, with R² values typically exceeding 0.95, and it captures the long‑tailed nature of web traffic that persists even after many months.

To explain this systematic behavior, the authors propose a two‑stage mechanistic model grounded in the physical layout of a website’s front page. In the first stage, newly published items appear at the top of the front page. As newer content is added, older links are pushed down the list, reducing their visual prominence. Because users tend to click on items that are higher on the page, the probability of a click decays as the link’s rank increases. In the second stage, once an article is displaced far enough, it disappears from the front page altogether and migrates to deeper sections such as archives, category pages, or search results. This “background” placement dramatically lowers its visibility and thus its click probability. The combined effect of these two positional shifts yields the observed power‑law decay.

The paper contrasts this explanation with two dominant theoretical frameworks in the literature. The first is the “novelty decay” hypothesis, which attributes the drop in attention to a diminishing psychological novelty factor. While plausible for short‑term dynamics, this model cannot account for the wide spread of α values across different sites. The second is the “priority‑queue” model of human dynamics, which treats user actions as tasks processed according to a stochastic priority rule. Although such models generate heavy‑tailed inter‑event times, they do not incorporate the explicit influence of link position on a page, and therefore fail to reproduce the observed dependence of α on site architecture.

The authors also note that the power‑law form is reminiscent of self‑organized criticality and other complex‑system phenomena, suggesting that the collective behavior of many users navigating a shared interface may give rise to emergent scaling laws. Nevertheless, they argue that for practical purposes the decay can be effectively modeled by the simple, measurable variable of link rank and its evolution over time.

From an applied perspective, the findings have several implications for content creators and web administrators. First, maximizing the duration an article remains near the top of the front page can substantially increase its total traffic; this can be achieved by controlling publishing frequency or by using “sticky” posts. Second, rather than abandoning older content, sites can periodically resurface it through related‑article widgets, recommendation engines, or periodic “best‑of” lists to restore visibility and mitigate the steep drop associated with stage two. Third, design choices that make links more salient (e.g., larger fonts, contrasting colors, or visual cues) can partially offset the decay caused by lower placement.

In summary, the study provides robust empirical evidence that attention to web articles decays as a power law with time, and it attributes this pattern primarily to the systematic downward movement of links on a site’s front page and their eventual relegation to background sections. This positional‑visibility mechanism explains the variability in decay exponents across sites more convincingly than novelty‑based or abstract human‑dynamics models, offering concrete guidance for optimizing content lifespan in the digital ecosystem.