The slowing growth of Wikipedia (part two): resistance from dominant editors

10 August 2009 | Ed Chi

Part one of this post, which shared findings on the slowing growth of Wikipedia, recently received coverage in the New Scientist (as well as Fast Company, Business Insider/ Silicon Alley Insider, and more). In this second part, we share some more details on the plateauing, more-linear-than-exponential growth rate of Wikipedia. (These findings will be published in WikiSym2009).

Growth among only the most frequent editors

This figure illustrates how the slowdown in Wikipedia activity differs among “editor classes”, as defined by frequency of their monthly edits. [Adhering to power law, we classified users using an exponential scale --  this resulted in 5 classes of users that were recalculated for each month: contributing 1 edit (power of 10), 2-9 edits, 10 to 99 edits, 100 to 999, and more than 1000 edits.]

All editing classes contributing less than 1000 edits slightly decreased their monthly edits beginning in 2007. Only the most frequently editing class (dark blue line at right) show an increase in monthly edits.

Another way to look at this data: analyze relative activity for each editor class. Complementing the figure above, the figure at right shows the percentage volume of edits that each class contributes relative to the total.

The two most-frequently editing classes of editors account for more than half the total monthly edits (56% from 2005 to 2008). Furthermore, the proportional contributions by the most-frequently editing class has increased since 2005. The editors in the 1000+ class have been producing edits at an increasing rate (average monthly edits per editor: 1740 in 2005, 1859 in 2006, 1869 in 2007, and 2095 in 2008).

Why the slowdown?

Let’s examine some specific evidence about what might have contributed to the slowdown for everyone but the most-frequent editors. REVERT is the action of deleting a prior edit. We analyzed the percentage of edits that were reverted monthly for each editor class (vandalism- and bot-related edits were excluded). This figure illustrates two indicators of growing resistance to new content:

(1) The total percentage of monthly reverted edits  (black line at left) has steadily increased over the years for all classes of editors — from 2.9 in 2005 to 5.8 in 2008.

(2) More interestingly: less-frequent or occasional editors (two top reddish lines at left) experience greater resistance compared to most-frequent editors.

The disparate treatment of edits from different editor classes has been widening steadily over the years — and not in favor to the less-frequent editors.

We consider this disparity as evidence of the dominant Wikipedia community’s growing resistance to new content, especially when the edits come from occasional editors.

Comments (3)