User:Bawolff/commons/ImagePatrol

Problem

• New things get uploaded
• Some of the things are bad (Usually misunderstandings about licensing, but also vandalism, etc)
• No systemic way of evaluating new uploads to check which are good

Current tools

• Just the basic list of new file type special pages - Special:Newfiles, Special:ListFiles, Special:RecentChanges, Special:RecentChangesLinked.
• Commons has various user scripts to help sort out files. Most notably the one-click nominate this file for deletion in the toolbox.
• There's a previous attempt at Commons:Commons:CUP.
• Various "tools" at commons:Commons:Tools#Patrolling_and_notification
• No equivalent of new page patrol for images, let alone Page Curation (using the PageTriage extension, see New Pages Feed on enwiki). Users cannot mark off what Images they have checked, so hard to prevent duplication of work.

What to do

• First step could be to make a very basic Image patrol feature.
  • User can check off that a particular image is checked (Mark image as patrolled).
  • Would want an option for pages like Special:newimages to not show patrolled images
  • Would also want a "recent images in category X" feature, so people could patrol images from a specific category (Like images uploaded via mobile web). Something similar to commons:Special:RecentChangesLinked/Category:Uploaded_with_Mobile/Web but image specific
  • Such a patrolling feature should be integrated into the api so that local js-wizards could easily integrate patrolling features into their tools where it makes sense to.
• Second step would perhaps be a more integrated tool that allows proposing for deletion, adding {{nsd}}, automatically flagging suspicious pages. Basically the things PageTriage does but for images.
  • I would propose that step 1 be implemented first, but implemented in such a way that it can be folded into step 2 (So as to avoid having to re-do patrolling system like PageTriage and traditional new page patrol)
  • In particular I imagine that means not storing patrolling data in RecentChanges table. (Which really is not a good place to store it for a number of reasons). I would suggest perhaps putting in image/oldimage table directly. (Would need to make sure that that can be done efficiently, especially for a "recent images in cat X" feature).
  • PageTriage involves a lot of cultural assumptions. To some extent that is very difficult to avoid. Thus it would be good to have the two steps integrate with each other, but also be separate so the basic image patrol feature could be used on any wiki, regardless of their cultural traditions.

Additional things to consider (vs en Wikipedia)

• People come from all over, and sometimes don't speak english. Its hard to evaluate if a file is ok when its description is written in Chinese and you are zh-0.
• There is much more bot uploaded content (from glams and things). These probably don't generally need to be patrolled in the same way that user content does.