Wikipedia:Does Wikipedia traffic obey Zipf's law?

WikiStats
Main
General info Status statistics WikiStats project WikiProject edit counters
General statistics
Awareness† Editors Total size Size comparisons Growth Records Zipf's law comparison
Breakdowns
Editors by number of edits Editors by article count Time between edits Administrator actions
Notes
† Inactive / archived for history
v t e

If accesses to Wikipedia's article pages obey Zipf's law, we can expect a roughly linear relationship between log(hits) and log(hit rank) for Wikipedia pages. (Note: the hit data in the graph has been scaled in such a way that 10000 hits are equivalent to 1% of the total access rate.)

This appears to be the case in practice for pages with rank between 5 and 1000, based on data from WikiCharts, as of September 2006.

The five most popular pages deviate significantly from the straight-line curve, but the approximation is pretty accurate from then on. The slope of this part of the log-log graph is approximately 1/2, suggesting that the hit rate is inversely proportional to the square root of the page rank,

${\displaystyle {\begin{aligned}\log({\text{hits}})&\approx -{\frac {1}{2}}\log({\text{hit rank}})+\log H_{0},\\[8pt]{\text{thus, hits}}&\approx {\frac {H_{0}}{\sqrt {\text{hit rank}}}}\\[8pt]{\text{or, }}&{\frac {{\text{number of hits for article ranked}}\;n}{{\text{number of hits for article ranked}}\;m}}\approx {\sqrt {\frac {m}{n}}}\end{aligned}}}$

Note: These scaled hit rates are derived from actual hit data counts over a particular period, and thus reflect actual hit counts for a statistical sample of user hits over that period, rather than statistical estimates of a theoretical underlying constant hit rate from those hit counts. The error bars in the WikiCharts data apply to the hit rates as an estimator of an underlying hit rate, and do not apply here.