Jump to content

Research:Editor classes

From Meta, a Wikimedia project coordination wiki
Contact
Zack Exley
This page documents a completed research project.


Topic

[edit]

The Editor Trends Study showed that the number of active editors (editors making at least 5 edits a month) on many of the top Wikipedia projects is in decline. This sprint is intended to build on the findings of that study to include different "classes" of editors: those in the 1-9, 10-99, 100-999, 1000-9999, and 10,000+ edit ranges. How has the recruitment of editors who go on to make a significant contribution to the project changed over the years? Are there fewer heavy editors or fewer casual editors recruited since 2007?

Process

[edit]

We took the number of editors on the English Wikipedia from years 2011 to either 2010 or the present (including 2011 may produce a misleading appearance of further decline, since the year is not complete) and divided them up by those from each month who would go on to make 1-9, 10-99, 100-999, 1000-9999, and 10000+ edits in their first year as a Wikipedian, starting from the date of their first edit.

This does not address retention of different ages or classes of editors at all, but instead demonstrates how many new editors who join the project become very active Wikipedians.

Results and discussion

[edit]

By charting the number of editors who would go on to make a certain amount (1-9, 10-99, 100-999, 1000-9999, or 1000+) of edits in their first year of editing, we discovered marked differences between editor classes.

English

[edit]

While the recruitment of new Wikipedians in the 1-9 and 10-99 classes began to decline in 2007, the year indicated by the Editor Trends Study for the outflow of new editors, heavier editors were being recruited less and less into the project starting in 2006 and their numbers have been dropping much more dramatically each year than the other editing classes.

This suggests that a decline in very active editor recruitment preceded the overall decline in retention of new editors, and that Wikipedia is not recruiting heavy editors at nearly the same numbers as light and moderate editors. Given the findings of other sprints, which indicate that many important activities on Wikipedia such as new page patrol and vandal-fighting follow a power law curve (where most of the work is done by the small fraction of top heavy-duty editors), the sharper decline in recruitment of very active English Wikipedians demonstrates an even greater potential challenge for the health and growth of the project.

Recruitment of different editor classes into the English Wikipedia over time Recruitment of different editing classes by year, log scale

EN Wikipedians by year who made n edits in their first year
Year 1-9 edits 10-99 edits 100-999 edits 1,000-9,999 edits 10,000+ edits
2001 414 120 90 25 0
2002 998 501 248 115 4
2003 4927 2377 887 293 29
2004 24623 11295 3587 1058 72
2005 104113 39572 11041 2822 169
2006 512037 130223 21591 3926 239
2007 747853 130706 14807 2231 148
2008 678789 105564 11042 1603 105
2009 645375 96021 9693 1219 68
2010 588566 84187 7728 902 44
2011 277079 35639 2678 165 2

Russian and German

[edit]

Similar trends hold true when you look at projects such as German, even on a different time scale, such as here where we measured amount of edits in the first 3 months of participation. Note however that there is only a very slight decline in recruitment for these two editor classes in Russian, though retention trends in Russian may still be in decline.

Recruitment by year of German and Russian Wikipedians who went on to make 1 or more edits their first 3 months Recruitment by year of German and Russian Wikipedians who went on to make 100 or more edits their first 3 months

Future work

[edit]

Using Toolserver, the same data as above has been generated for all language projects of Wikipedia other than English, and needs to be charted and shared.