Research talk:Measuring edit productivity/Work log/2015-12-02
Add topicWednesday, December 2, 2015
[edit]And I have data! Time for some system-level plots. So, I filtered out all of the tokens that were not words and set "persistence" thresholds at 5 edits and 48 hours. First, I'll be looking at the measures of persistent words added and conceptualizing this as a measure of overall productivity.
OK. Two interesting things that we can see right away:
- unregistered users (anons) add a substantial amount of the tokens that persist
- the overall productivity of registered editors has been holding relatively constant at 125 million persistent words per month, but the productivity of anons has been in steady decline since 2007.
Let's look the proportion of persistent words from anons to see what the decline looks like.
So, we can clearly see a decline here. In 2006, anons were adding almost 18% of persisting word content to Wikipedia. By June of 2015, that figure has declined to ~12%. Proportionally, that's an 30% decay. --Halfak (WMF) (talk) 20:42, 2 December 2015 (UTC)
Reflection
[edit]I just want to take a moment to reflect on something.
Re. the sustained productivity of registered editors, this is a really surprising result. Given that Wikipedia's editor population went through a substantial decline starting in 2007, this result implies that the remaining Wikipedians (and the newcomers who made it into Wikipedia's community of substantial contributors) are very productive. This is also interesting given results that suggest that the total labor hours devoted to editing Wikipedia have been declining along with the size of the volunteer population[1]. Maybe the remaining population is just working more efficiently than the Wikipedians of 2006 and 2007. --Halfak (WMF) (talk) 20:42, 2 December 2015 (UTC)
References
[edit]- ↑ Geiger, R. S., & Halfaker, A. (2013, February). Using edit sessions to measure participation in Wikipedia. In Proceedings of the 2013 conference on Computer supported cooperative work (pp. 861-870). ACM.