Wikidata

Wikidata:Requests for comment/semi-protection to prevent vandalism on most used Items

< Wikidata:Requests for comment
An editor has requested the community to provide input on "semi-protection to prevent vandalism on most used Items" via the Requests for comment (RFC) process. This is the discussion page regarding the issue.

If you have an opinion regarding this issue, feel free to comment below. Thank you!


THIS RFC IS CLOSED. Please do NOT vote nor add comments.

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.

consensus to semi-protect all items used by 500 Wikimedia pages or more. Semi-protection that is only justified by usage should be lifted once the item falls below the threshold. The criteria of Wikidata administrators about usage or visibility on Wikimedia projects may also be enough to semi-protect other items. --Pasleim (talk) 12:31, 24 June 2019 (UTC)[reply]

Data from a few Wikidata Items are shown on many Wikimedia pages (for instance, Wikipedia articles), so vandalism on these Items can cause much damage even when vandalism is quickly detected and reverted. Edits like the one that changed Iraq's label to "Iran", the one that changed Russia's label to "mainkra" or the one that vandalized the label for pneumonia made thousands of pages show wrong countries and absurd causes of death for hours, since problematic versions remain on Wikimedia pages after their reversion on Wikidata. The impact of this vandalism has been expressed on Wikipedias and on Wikidata and its prevention has been requested on Wikimedia's Community Wishlist Surveys in the current and past years. This vandalism usually extends beyond Wikimedia projects, affecting popular tools and services like Google and Siri and being echoed by the press.

This request for comments aims to compile the community's position on using semi-protection as a way of preventing vandalism on the most used Items. This request for comments does not aim to compile comments about other ways of combating vandalism, which may coexist and should be discussed on other pages, nor to discuss the implementation details (how) of the various preferences that can be reflected (what). Decisions should be based on the current resources of the community and the development team, not on those they might or might not have in the future.

Please read carefully the questions and write below the options that most closely approximate those you consider best. Thanks for your feedback! --abián 14:23, 15 February 2019 (UTC)[reply]

Questions edit

1. Should some Wikidata Items be semi-protected according to their usage on Wikimedia pages? edit

This would clarify the page protection policy, which states that page protection "should generally be done after a pattern of vandalism or spam can already be seen in the page's history, but for high-use pages (such as the main page and the community portal) it can be done without such history".
Semi-protection is the intermediate status of page protection, weaker than the full protection but stricter than the default unprotection. To change a semi-protected page on Wikidata you have to be a confirmed user, which means you have to own an account created at least 4 days ago with at least 50 edits or have a confirmed flag granted by an administrator.
Help: Advantages and disadvantages of semi-protecting Items according to their usage on Wikimedia pages
Pros Cons

 ' Semi-protection is the most effective mechanism available so far in Wikidata to prevent vandalism'
  ~95% of vandalism on labels, descriptions and aliases is caused by unregistered users[1]
  Distribution for most used Items exhibits a power-law behavior[2]
 ' It's easy to get autoconfirmed status on Wikidata'
 ' It's possible to grant the confirmed flag to any trusted Wikimedian that is not yet autoconfirmed'
 ' Drastically reduces the risk of wrong data spreading to many Wikipedia articles, so encourages Wikipedias to show data from Wikidata'
 ' Reduces the risk of wrong data spreading to external tools and services, so encourages them to show data from Wikidata'
 ' Encourages editors to use registered accounts'
 ' Encourages some newcomers to edit less visible Items first and make mistakes on them if necessary'
 ' The percentage of semi-protected Items would be low in any case'
 ' Good-faith contributors would be more aware of the problem'

 ' Can not become the general rule, openness is Wikimedia's key value'
 ' Discourages some newcomers from contributing to Wikidata'
 ' It's not easy for newcomers to find a way to request edits on protected Items'
 ' It's not easy to measure the number of potentially good edits that are prevented'
 ' phab:T143486 is not fixed yet'
 ' Vandals would be more aware of the problem'

  1. ↑ Stefan Heindorf, Martin Potthast, Benno Stein, and Gregor Engels. 2015. Towards Vandalism Detection in Knowledge Bases: Corpus Construction and Analysis. In Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR '15). ACM, New York, NY, USA, 831-834. DOI: https://doi.org/10.1145/2766462.2767804.
  2. ↑ phab:T210664#4907866


1.A. Yes, high usage on Wikimedia projects justifies semi-protecting one or more Items. edit

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

1.B. No, high usage on Wikimedia projects does not justify semi-protecting any Item. edit

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

  • I think this needs to wait at least for phab:T143486, and preferably instead of semi-protection, to protect more granularly (allow sitelink changes) with phab:T189412. Or we need some other solution that allows experienced users on other wikimedia sites to automatically be able to bypass semiprotection in Wikidata. ArthurPSmith (talk) 17:28, 16 February 2019 (UTC)[reply]
  • MisterSynergy (talk) 17:28, 17 February 2019 (UTC); After thoughts about the matter in the past days, I do not think that there is a problem severe enough to restrict Wikidata’s openness. I am fine with (long-term, but definite) page protections for frequently vandalized pages, but this proposal seems way too much for me to be acceptable.[reply]
  • Per MisterSynergy, and also on the principle that there is no simple metric that is an accurate predictor of the variable "needs protection".--Jasper Deng (talk) 23:05, 17 February 2019 (UTC)[reply]
  • Frequently-used items should be more closely monitored, not locked down. Protected items will be lower quality, as they prevent contributions. --Yair rand (talk) 01:55, 18 February 2019 (UTC)[reply]
  • Mar del Sur (talk) 03:04, 18 February 2019 (UTC) No, the anonymous contribution should not be prohibited "preventively". The only case in which the pages should be protected or semi-protected is that of repeated and persistent vandalism, really when it has been proven that it can not be controlled in any other way.[reply]
  • I'm on the fence, but leaning towards not supporting for now. I would support this, but the sitelink issue could be problematic, as most Wikipedia editors are not autoconfirmed on Wikidata. If particular items are being vandalized a lot then they can usually be protected on a case-by-case basis; and a lot of potential vandalism targets can be watchlisted in large batches by adding them from Wikipedias' topviews reports (I would think vandalism is more closely correlated to Wikipedia page views than to item transclusions). Protecting important items would also negatively affect some edit-a-thons, particularly those focused on adding labels and descriptions in uncommon languages. I don't think I've ever come across more than one such edit-a-thon, though. Jc86035 (talk) 13:57, 21 February 2019 (UTC)[reply]
  • I think high usage can be a valid reason to justify protection, but the remainder of this proposal seem to take discretion away from administrators altogether, and instead propose that any item with more than X transclusions must be protected and any item below the threshold may never be protected. I don't it's a good idea to remove that flexibility. Deryck Chan (talk) 14:46, 8 April 2019 (UTC)[reply]

2. If so (1.A), should these Items be semi-protected according to a threshold based on the number of uses or according to the various criteria of Wikidata administrators? edit

The number of uses of an Item is the number of Wikimedia pages (for instance, Wikipedia articles) that retrieve one or more data (labels, descriptions, statements, etc.) from the Item. Therefore, the number of uses is the number of Wikimedia pages that could appear vandalized if the Item is vandalized in the worst case scenario. These pages aren't only those connected to the Item by means of sitelinks; an Item can have ten sitelinks while thousands of Wikimedia pages show data from it.

2.A. Every Item above a certain number of uses should be semi-protected, while the criteria of Wikidata administrators about usage or visibility on Wikimedia projects may also be enough to semi-protect other Items. edit

2.B. Every Item above a certain number of uses should be semi-protected and the criteria of Wikidata administrators about usage or visibility on Wikimedia projects can not be a reason to semi-protect any other Items. edit

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

2.C. No threshold based on the number of uses should be imposed, the various criteria of Wikidata administrators about usage or visibility on Wikimedia projects should be enough to determine what Items should be semi-protected for this reason. edit

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

3. If you think that all Items above a certain number of uses should be semi-protected (2.A or 2.B), in terms of which metric should this threshold be set? edit

The number of uses of an Item is the number of Wikimedia pages (for instance, Wikipedia articles) that retrieve one or more data (labels, descriptions, statements, etc.) from the Item. Therefore, the number of uses is the number of Wikimedia pages that could appear vandalized if the Item is vandalized in the worst case scenario. These pages aren't only those connected to the Item by means of sitelinks; an Item can have ten sitelinks while thousands of Wikimedia pages show data from it.

3.A. Every Item used on more than T pages on Wikimedia projects. edit

Since Wikidata usage grows over time, the number of Items semi-protected for this reason should also grow over time.

3.B. T% of the Items used on the most pages on Wikimedia projects. edit

Since the total number of Items grows over time, the number of Items semi-protected for this reason should also grow over time.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

3.C. The T Items used on the most pages on Wikimedia projects. edit

Although both Wikidata usage and the total number of Items grow over time, the number of Items semi-protected for this reason should not change.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4. If you think that all Items above a certain number of uses should be semi-protected (2.A or 2.B), what this threshold should be? edit

The three metrics of each option are roughly equivalent today with some rounded values. These values are based on the quantitative analysis described on phab:T210664. Currently, the 3000 Items used on the most pages on Wikimedia projects are semi-protected. Please choose the option that most closely approximates the one you consider best for the future and, if you want, specify your exact figure next to your signature.

4.A. Each Item is used by 500,000 Wikimedia pages or more (if 3.A) / 0.0001% most used Items (if 3.B) / 65 most used Items (if 3.C). edit

This threshold would currently mean unprotecting several semi-protected Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.B. Each Item is used by 100,000 Wikimedia pages or more (if 3.A) / 0.0005% most used Items (if 3.B) / 260 most used Items (if 3.C). edit

This threshold would currently mean unprotecting several semi-protected Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.C. Each Item is used by 50,000 Wikimedia pages or more (if 3.A) / 0.0009% most used Items (if 3.B) / 500 most used Items (if 3.C). edit

This threshold would currently mean unprotecting several semi-protected Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.D. Each Item is used by 10,000 Wikimedia pages or more (if 3.A) / 0.0025% most used Items (if 3.B) / 1400 most used Items (if 3.C). edit

This threshold would currently mean unprotecting several semi-protected Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.E. Each Item is used by 2500 Wikimedia pages or more (if 3.A) / 0.007% most used Items (if 3.B) / 3600 most used Items (if 3.C). edit

This threshold is close to the current state.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.F. Each Item is used by 1000 Wikimedia pages or more (if 3.A) / 0.014% most used Items (if 3.B) / 7450 most used Items (if 3.C). edit

This threshold would currently mean semi-protecting several Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.G. Each Item is used by 500 Wikimedia pages or more (if 3.A) / 0.029% most used Items (if 3.B) / 15,800 most used Items (if 3.C). edit

This threshold would currently mean semi-protecting several Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.H. Each Item is used by 300 Wikimedia pages or more (if 3.A) / 0.06% most used Items (if 3.B) / 33,200 most used Items (if 3.C). edit

This threshold would currently mean semi-protecting several Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.I. Each Item is used by 200 Wikimedia pages or more (if 3.A) / 0.10% most used Items (if 3.B) / 60,000 most used Items (if 3.C). edit

This threshold would currently mean semi-protecting several Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.J. Each Item is used by 150 Wikimedia pages or more (if 3.A) / 0.15% most used Items (if 3.B) / 80,000 most used Items (if 3.C). edit

This threshold would currently mean semi-protecting several Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.K. Each Item is used by 100 Wikimedia pages or more (if 3.A) / 0.21% most used Items (if 3.B) / 115,000 most used Items (if 3.C). edit

This threshold would currently mean semi-protecting several Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.L. Each Item is used by 50 Wikimedia pages or more (if 3.A) / 0.46% most used Items (if 3.B) / 250,000 most used Items (if 3.C). edit

This threshold would currently mean semi-protecting several Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.M. Each Item is used by 30 Wikimedia pages or more (if 3.A) / 0.84% most used Items (if 3.B) / 460,000 most used Items (if 3.C). edit

This threshold would currently mean semi-protecting several Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

4.N. Each Item is used by 15 Wikimedia pages or more (if 3.A) / 1.74% most used Items (if 3.B) / 950,000 most used Items (if 3.C). edit

This threshold would currently mean semi-protecting several Items.

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

5. If you think that all Items above a certain number of uses should be semi-protected (2.A or 2.B), should a semi-protection that is only justified by usage be lifted once the Item falls below the threshold? edit

Items would be semi-protected with the indefinite option and they may or may not be unprotected once they fall below the threshold. Aspects such as the frequency with which a new ranking of entities can be generated and protection statuses reviewed depend entirely on technical limitations.

5.A. Yes, semi-protection should be lifted unless there are exceptional reasons to keep it. edit

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

5.B. No, semi-protection should remain unless there are exceptional reasons to lift it. edit

If you prefer this option, please add your signature (*~~~~) below and, if you wish, write a brief comment next to it.

Discussion and other suggestions edit

If you have further comments and suggestions on the subject of this request for comments, semi-protection as a way of preventing vandalism on the most used Items, feel free to leave them below. Please don't forget your signature, ~~~~, and try not to repeat the comments you left above.

  • IMHO future developments should be aimed to a more granulated protection (statements vs labels&descriptions vs sitelinks). Even protecting single statements (bot-imported from a credible source and sourced -> no need for changing that, except, maybe, its "ranks"). I do not have a strong oppinion on this mass-protection yet. As an experiment (limited to a short set of items) it has some interest, but reaching previous community consensus would have been nice. Due to this "lacking-consensus" status, it doesn't matter to me whether these protections are mass-reverted or not, but at least they should have a temporal end (1-2 months, maybe). Some statistical evaluation of the results once that period is passed ...will be nice. strakhov (talk) 19:08, 17 February 2019 (UTC)[reply]
    • I do think granular protection would be a far better solution. There are some claims that should almost never be changed, such as the atomic number of an element, but that should not lock down the whole item from editing.--Jasper Deng (talk) 03:13, 18 February 2019 (UTC)[reply]
@Nicereddy: Please read the introduction : "Semi-protection is the intermediate status of page protection, weaker than the full protection but stricter than the default unprotection. To change a semi-protected page on Wikidata you have to be a confirmed user, which means you have to own an account created at least 4 days ago with at least 50 edits or have a confirmed flag granted by an administrator". Snipre (talk) 13:14, 19 February 2019 (UTC)[reply]

References edit

The above discussion is preserved as an archive. Please do not modify it. Subsequent comments should be made in a new section.
Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:Requests_for_comment/semi-protection_to_prevent_vandalism_on_most_used_Items&oldid=1747270330"