2009/7/18 Carcharoth carcharothwp@googlemail.com:

On Sat, Jul 18, 2009 at 9:53 AM, David Gerarddgerard@gmail.com wrote:

...

http://stats.wikimedia.org/EN/TablesWikipediaEN.htm

Cool! I'm too lazy to look. Anything there worth discussing?

Look at that URL at least. You'll see the data for our recent discussion on article numbers laid out in a nice table, for instance.

- d.

On Sat, Jul 18, 2009 at 7:20 AM, Carcharothcarcharothwp@googlemail.com wrote:

Cool! I'm too lazy to look. Anything there worth discussing?

To me, the data is really encouraging. Take a look at the charts for New Wikipedians vs. Active Wikipedians. We knew before that both of those peaked in early 2007. But now it seems that the decline has more or less stabilized, and the decline in active Wikipedians was less severe than new Wikipedians. Edits per month, and maybe new articles per month, look to be stabilizing as well.

Broadly speaking, there are two possible explanations for why community activity level peaked and then declined: market saturation (just about everyone likely to edit was exposed to Wikipedia by mid 2007) or project maturity (activity declines because people can't find things to do). Obviously there are elements of both at work, but comparing the new and active charts suggests to me that market saturation has been the dominant factor, and that editors are not having too much trouble finding things to work on. That's much more cause for optimism than if people were leaving simply because they were satisfied with a 'good enough' Wikipedia (which everyone here knows has a long way to go yet).

In another thread, Will Johnson (I think) argued that activity levels (new articles, in particular) would continue to decline rapidly in the next few years and that by Christmas we would have fewer than 1000 new articles per day. Looking at the new stats, I'm more confident that en-wiki can maintain a steady state of activity something close to the present level (especially as the usability efforts begin to make it easier for newbies to edit, after years of increasingly complex markup that did the opposite).

-Sage

Sage Ross wrote:

To me, the data is really encouraging. Take a look at the charts for New Wikipedians vs. Active Wikipedians. We knew before that both of those peaked in early 2007. But now it seems that the decline has more or less stabilized, and the decline in active Wikipedians was less severe than new Wikipedians. Edits per month, and maybe new articles per month, look to be stabilizing as well.

This also was my first impression: the last 12 months have not been so bad at all. I'd like to be able to combine this with a continuing thought of mine: WP's model is beginning to "bite", in the sense that it has not proved really problematic to bring new areas of content along, and there has also been progress in upgrading lower-quality existing articles. I think both points are still somewhat debatable; but if both of these are granted in a general sense (dodging say round BLP and a few vexed areas where edit wars are still typical) there is scope for optimism.

Charles

2009/7/20 Michael Peel email@mikepeel.net:

...

In another thread, Will Johnson (I think) argued that activity levels (new articles, in particular) would continue to decline rapidly in the next few years and that by Christmas we would have fewer than 1000 new articles per day.  Looking at the new stats, I'm more confident that en-wiki can maintain a steady state of activity something close to the present level (especially as the usability efforts begin to make it easier for newbies to edit, after years of increasingly complex markup that did the opposite).

Perhaps rather than counting new articles, we should be looking at the amount of text added to articles? As I understand it, there's a bit of a shift going on from creating new articles to improving existing ones, and that seems like a reasonable metric to measure that from (especially if only article text, rather than templates, was counted). Of course, it's also a lot harder to calculate...

"edits per article" is the best statistic we have for that at the moment (and it's fairly steady). Once we have a full dump we'll be able to get better statistics. I don't know what the timeframe for that is, but I understand it is being worked on.