Community Wishlist Survey 2022/Wikidata/Automated page protector

Automated page protector

  • Problem: Sister projects, especially large ones like the English Wikipedia, do not want to use Wikidata because it is prone to vandalisation. By exposing Wikidata to projects as big as the English Wikipedia, vandalism could increase sharply and human editors would not be able to keep up with reverts and page protection requests would overflow admins.
  • Proposed solution: An automated page protector bot or system that recognizes if a page is repeatedly being vandalised such as by tracking the number of reverts of edits made by inexperienced users and their ORES scores. It should then apply the appropriate page protection automatically after a number of identified forms of vandalism have occured.
  • Who would benefit: Everyone because using Wikidata has the power to significantly strengthen sister projects and the movement overall.
  • More comments: Not addressing the vandalism issue is a serious and cyclical problem that is holding back the huge potential for Wikidata. It goes like this: Wikidata isn't used much because it could be vandalised if exposed to the masses. Because it isn't used much it isn't vandalised and the tools to prevent it are not developed. We need to develop the tools to stop vandalism now so that other projects recognize Wikidata can handle it, and only then will they be willing to use it.
  • Phabricator tickets:
  • Proposer: Lectrician1 (talk) 06:45, 11 January 2022 (UTC)

Discussion

  • Wikidata changes for statements that are used other projects' pages should show up on that page's Watchlist Are you aware that this already happens? --Lucas Werkmeister (talk) 09:26, 11 January 2022 (UTC)
    Okay cool :) I wasn't sure or not. Lectrician1 (talk) 12:45, 11 January 2022 (UTC)
  • Note that heavily used items are already automatically protected by a bot: d:Wikidata:Requests for permissions/Bot/MsynABot. --Matěj Suchánek (talk) 09:31, 11 January 2022 (UTC)
  • Why not make ClueBotNG work on Wikidata? NightWolf1223 (talk) 15:50, 11 January 2022 (UTC)
    @NightWolf1223 That should probably be a separate proposal. Also, developing a system that can recognize the vandalism of pure data would be very hard because data points have little recognizable context and structure compared to human sentences. Providing page protection is a much more easy to implement and effective as it significantly reduces the chance of vandalism occuring in the first place. Lectrician1 (talk) 18:57, 11 January 2022 (UTC)
  • Currently ORES scores for Wikidata edits are not even remotely good enough to use them for such a task. In general, the entire revision-based scoring approach is not ideal for Wikidata with its atomic edits.
    We need to hope/wait for better scoring systems, and—for the time being—use the conventional patrolling function as a community much more that we currently do. —MisterSynergy (talk) 20:13, 12 January 2022 (UTC)
    @MisterSynergy I mean, I'm not looking for a system that scores Wikidata edits (though it could be involved). I'm looking for a system that tracks vandalism and does something about it. All it needs to do is track reverts on a page that were marked as vandalism and then protect that page after a number of occurrences. This shouldn't be that hard... Lectrician1 (talk) 05:29, 13 January 2022 (UTC)
    @Lectrician1: Could you clarify the proposed solution above to make it clear that this isn't only about using ORES scores? Thanks. —SWilson (WMF) (talk) 05:09, 14 January 2022 (UTC)
    @SWilson (WMF) done. I myself actually don't even know if ORES scores are available for Wikidata at the moment or if we have a way to mark reverts as vandalism... Can we do either of those? Lectrician1 (talk) 05:21, 14 January 2022 (UTC)
    It looks like ORES scores are a thing so I added that back in. Lectrician1 (talk) 05:31, 14 January 2022 (UTC)
    @Lectrician1: Looks good, thanks. I think the specific details can be figured out later once support for the general idea is established. — SWilson (WMF) (talk) 06:27, 14 January 2022 (UTC)
  • @Peter coxhead, Finnusertop, Sdkb, Bluebaor, and Fram: @Blueboar: You conveyed in either this village pump discussion or this one that Wikidata is prone to vandalism and that something needs to be done about it to allow for it's use on the English Wikipedia. This proposal seeks to build a tool to prevent vandalism for exactly that reason and I'd recommend you vote for it. Lectrician1 (talk) 19:41, 29 January 2022 (UTC)

Voting

  •   Support Sea Cow (talk) 19:03, 29 January 2022 (UTC)
  •   Support Silver hr (talk) 19:45, 2 February 2022 (UTC)
  •   Support Wikidata and Commons can be prone to vandalism. Thingofme (talk) 14:26, 4 February 2022 (UTC)
  •   Support Daniel Case (talk) 23:08, 5 February 2022 (UTC)
  •   Support --Ciao • Bestoernesto 16:49, 6 February 2022 (UTC)
  •   Support Ayumu Ozaki (talk) 04:11, 7 February 2022 (UTC)
  •   Support this is needed Carn (talk) 15:21, 11 February 2022 (UTC)
  •   Support You can take in consideration also blocks from sitelinks. Valerio Bozzolan (talk) 17:22, 11 February 2022 (UTC)