Page MenuHomePhabricator

statistics about edit conflicts according to page type
Open, NormalPublic

Description

This is a follow up to a discussion from Wikimania 2016.

Nobody likes edit conflicts.

To make them cause less suffering, it would be useful to know how often do they happen and where.

The patterns of editing on talk pages and on articles are quite different, and so are the patterns of edit conflicts. Analyzing in which namespaces do conflicts happen may show, for example, that 75% of conflicts happen in Talk:, User talk:, and Project: namespaces, and less than 20% in the main space. But it's just a wild guess, and the real numbers might be totally different.

So—do we log edit conflicts on Wikimedia sites? If we don't, can we?

Ideas for things to include in the analysis:

  • namespace
  • general frequency of edits on a page
  • number of unique editors per page
  • page size
  • maybe other things

Thanks.

(Tagging Flow because it is a solution to edit conflicts on talk pages, and this issue may be interesting to Flow developers. Tagging Editing-(Department|Analysis) because it seems clearly related to both. Tagging Analytics because it's related as well, although I'm not sure who should own it exactly. Please feel free to adjust the tags.)

Event Timeline

Amire80 created this task.Jun 30 2016, 7:06 AM
Restricted Application added subscribers: Zppix, Aklapper. · View Herald TranscriptJun 30 2016, 7:06 AM

Tagging also Collaboration-Team and Community-Tech with the hope that it will interest them. Again, feel free to adjust the tags.

Amire80 updated the task description. (Show Details)Jun 30 2016, 7:08 AM
Amire80 updated the task description. (Show Details)

Finally, tagging TCB-Team, German-Community-Wishlist and Community-Wishlist-Survey, because this task is a follow up to T137318.

Restricted Application added a subscriber: JEumerus. · View Herald TranscriptJun 30 2016, 7:11 AM
Elitre added a subscriber: Elitre.Jul 4 2016, 5:36 PM

We have one dashboard for edit conflicts : https://grafana.wikimedia.org/dashboard/db/mediawiki-edit-conflicts
It shows how many conflicts appear, everywhere. @Addshore has also made a patch to differentiate more in which namespace the conflict actually occurs, which should be added to the dashboard soon.

Nice! What is the exact definition of Subject, Talk and Project pages?

Amire80 updated the task description. (Show Details)Jul 5 2016, 2:12 PM

Nice! What is the exact definition of Subject, Talk and Project pages?

Talk pages are all pages in a talk namespace
Subject pages are all pages that are not in a talk namespace
Project pages are all pages in the project namespace (the namespace id is defined by NS_PROJECT)

The patch that introduced this can be seen at https://gerrit.wikimedia.org/r/#/c/289837/

Lea_WMDE added a comment.EditedJul 5 2016, 2:16 PM

talk page: every talk page, examples:

project page: project subject page namespace

subject page: everything that is not a talk page, so:

  • main namespace project pages,
  • help pages,
  • category pages
  • image pages

Please note that "subject page" includes "project page"

TheDJ added a subscriber: TheDJ.Jul 5 2016, 2:23 PM

Would be nice if we could also gather statistics on 'follow through' though:

  1. How often do people just give up, vs have a successful save after an edit conflict
  2. Do people accept the 3rd party version, amend the 3rd party version, or overwrite the 3rd party version.

Please note that "subject page" includes "project page"

Hmmmm. Project pages come in a lot of varieties. There are probably many more edit conflicts on Village Pumps, Noticeboards and other discussion pages many of which are under the Project namespace than on essays which are not edited much. It would be nice to at least separate the Project namespace from the Subject namespace. If articles and Village Pumps are lumped together, the picture may be not as precise as it should be.

Putting this on our Radar until the project is better defined (right now it's a bit more like a research project than an infrastructure project). I'll also add that we have edit conflicts logged for a sample of 25% of our users via the Schema:Edit event logging instrumentation, I can point you to that data if you're interested.

Milimetric moved this task from Incoming to Radar on the Analytics board.Jul 7 2016, 5:01 PM
Zache added a subscriber: Zache.Jul 13 2016, 2:58 PM

Hmmph, in context of current edithaton it seems that it would be very nice that in case of edit conflict system would save the revision to log (like Abuse Filter log) or to some special page like sandbox where it could be recovered. It should not be too hard to implemet afaik and this would fix some saddest feelings which users will get from edit conflict and losing the content. (in this case back button didn't save the day)

Neil_P._Quinn_WMF triaged this task as Normal priority.Jul 26 2016, 5:38 PM

fyi: @Addshore just updated https://grafana.wikimedia.org/dashboard/db/mediawiki-edit-conflicts to show edit conflicts divided by name spaces. In short: Most edit conflicts appear on "Real Content, Article" pages (ID 0)

Neil_P._Quinn_WMF lowered the priority of this task from Normal to Lowest.
Neil_P._Quinn_WMF raised the priority of this task from Lowest to Needs Triage.
Neil_P._Quinn_WMF triaged this task as Normal priority.
Restricted Application added a project: Growth-Team. · View Herald TranscriptJul 5 2018, 8:24 PM