Research talk:Active editor

Differences

The page should highlight what the differences are compared to the traditional definition (the one that WikiStats used). Nemo 08:59, 21 December 2016 (UTC)

• @Nemo bis: I decided to keep that information on Research:Defining monthly active editors, 2016 and other similar pages. You're welcome to duplicate some of that here if you want to.Neil P. Quinn-WMF (talk) 23:45, 21 December 2016 (UTC)

foundationwiki

It's inconsistent to include foundationwiki but not the chapters wikis. What's the rationale? --Nemo 08:59, 21 December 2016 (UTC)

• Because Wikistats included it. It's a very minor point and I saw no reason to challenge it. Neil P. Quinn-WMF (talk) 23:45, 21 December 2016 (UTC)
• foundationwiki was added long ago on request (iirc by comms). Exclusion of chapters wikis wasn't a very deliberate decision. If my memory serves well some chapter wikis were hosted on Wikimedia servers, some weren't. At the time it seemed a situation that might change.

Redirects

Do I get Research:Defining monthly active editors, 2016 right that the redirects are included just because the new tools are incapable of distinguishing redirects? Given [1], it's troubling if the new systems are so limited. The impact on the active editors count may be limited, but I don't understand if there was some consideration of how desirable it is to consider redirect edits. Nemo 08:59, 21 December 2016 (UTC)

• the new tools (i.e. the MediaWiki SQL databases) are capable of distinguishing redirects, but it adds significant additional complexity to the definition which we decided was unnecessary.

We didn't discuss desirability, but I would say it is desirable to include redirect edits, for the same reason that I'd prefer to include non-content namespaces: activity is activity whether it happens on articles or redirects or talk pages or whatever. However, we didn't even consider changing the inclusion of non-content namespaces because this project was only about agreeing on minor technical details that had previously only been decided implicitly by the implementers of these metrics. If I ever decided I felt strongly enough about including non-content namespaces, I would absolutely propose that on the mailing lists first. Neil P. Quinn-WMF (talk) 23:45, 21 December 2016 (UTC)

• Interesting read 'Consider the redirect'.

I discussed this first with Analytics Team, and explained what it took to recognize internationalized redirect tags. Before 2009 Wikistats harvested aliases from php message files for each language. In 2009 this functionality was 'delegated' to dump scripts (at least for redirects, not for other internationalized tags): a new <redirect title="some name"/> tag was introduced for the xml dump (once per article, not per revision).

I said something like for 'Wikistats 2 rebuilding this functionality of detecting internationalized redirect tags seems more trouble than it's worth, just to exclude not so many edits'. But thinking again, we want to exclude redirects from article counts anyway (or else English Wikipedia would have 13 million 'articles'), so that detection logic is needed anyway, right? Erik Zachte (talk) 14:10, 22 December 2016 (UTC)

Deleted edits

It seems strange to count deleted edits, since vandalism etc. will be counted as a desirable thing. (Sure, most reverted edits are counted anyway because they don't get deleted.) It will also make it impossible for communities to correct mistakes of the past: if there was a bot import at some point and then thousands of articles have to be deleted (which happens regularly), the statistics will be forever polluted. Nemo 08:59, 21 December 2016 (UTC)

• The thing to keep in mind is that active editors measures activity, not productive activity. Erik Zachte originally felt the way you did, but changed his mind when he found that the number of deleted edits was far smaller than the number of reverted edits. Neil P. Quinn-WMF (talk) 23:45, 21 December 2016 (UTC)
• I can confirm I changed opinion recently. Deleted edits is 50 to 100 times less than reverted edits. So where I considered 'deletion drift' a move towards better numbers over time, it doesn't bring much improvement really. I'd say not enough to be a reason for ever-changing numbers Erik Zachte (talk) 14:29, 22 December 2016 (UTC).