Wikipedia talk:Reviewing pending changes

Hello everyone, I have created a page in my userspace at User:DreamRimmer/Reports/PendingChanges to recognize and award the top pending changes reviewers from the last 30 days. Anyone interested can join me in this effort. – DreamRimmer (talk) 14:37, 25 November 2023 (UTC)

I responded to 2 pending changes by the same IP to Ojamajo Doremi, selected the diff of both edits (linked) to understand the totality of the effort for context, concluded that it was good to go, and then was unsure if accepting the changes would function as one might expect.

I came here and read through the apparently related parts several times, and although will admit to being in the middle of other things too, could see no clear statement about the ideal, best practice, procedure. I chose to trust in the developers and accept both together – only the latter was accepted. I came back here and re-read it all again several times, and am still not sure. I went back to the page and manually also accepted the earlier of the two changes, since it just seems logical.

Could someone who knows the best practice, or required procedure, please make it clear in this documentation, or make clear where the information is if it's somewhere else (I would have to wonder why in that case)? Fred Gandt · talk · contribs 07:15, 16 September 2025 (UTC)

@Fred Gandt. I think I wrote a little bit about this at Wikipedia:Tips for pending changes reviewers#Other tips. In short, I agree with your technique of reviewing one editor's group of edits at a time. As long as a bunch of edits in a row are by the same editor, it should be safe to review them together. Just be careful if edits are by DIFFERENT editors, or not in a row. Hope that helps. –Novem Linguae (talk) 10:05, 16 September 2025 (UTC)

I appreciate the response. The technical aspect to this process is my main concern. Apparently if a reviewer accepts multiple edits (as linked: two edits by the same editor viewed as the difference they both made compared with the previous version), the system only marks the latest of the edits as accepted, leaving the intermediate edits as unreviewed. Should we mark each and every edit that leads to an acceptable result as accepted, or as the system does, just the latest of those edits? i.e. if an editor fumbles their way through, say 9 bad edits, then on their tenth, it all comes together and the whole is acceptable: do we go through the history marking each and every one of the 10 edits as accepted, or just the tenth (as the system does if a diff with intermediate edits is accepted)? Clearly marking 9 edits as unaccepted, but the tenth as accepted would be bizarre. This is not explained in this project page. Fred Gandt · talk · contribs 16:02, 16 September 2025 (UTC)

I remember being confused by this too. I concluded that if there are multiple pending edits, accepting the most recent one accepts them all, but this wasn't clear from the documentation.

There's no need to explicitly accept the interim edits.

It would be nice if this was explicit in the documentation. If nobody else replies I'll try adding something. Dan Bloch (talk) 17:39, 16 September 2025 (UTC)

You can leave some revisions unreviewed. PC protection shows to the public the newest reviewed revision. Anything older than that doesn't matter from a "display to the public" perspective, and is only important from a "make sure a reviewer checked it" perspective. You as the reviewer checked it even if it doesn't get marked (since you viewed a big diff with the unchecked revisions in it). –Novem Linguae (talk) 17:46, 16 September 2025 (UTC)

Okay, that's what I was thinking, and it did appear that with the most recent edit accepted, the previous was implicitly accepted (while that doesn't seem technically sound to me). And thank you Dan; I think this needs to be properly explained in Wikipedia:Reviewing pending changes § Reviewing edits by a single user. Fred Gandt · talk · contribs 18:21, 16 September 2025 (UTC)

Okay, this is done. Danbloch (talk) 01:41, 18 September 2025 (UTC)

Very good Dan, thank you :) One point I think could be highlighted is that part about all earlier pending changes being implicitly accepted, making it essential to verify that all currently pending changes earlier than the one you're accepting are good to go. While it is clear that this note is referring to reviewing changes by one user, it is possible (correct me if I am wrong) that: if a block of edits by user B came after any number of edits by user A, and the reviewer is checking user B's changes as a block, then accepts the most recent, user A's get implicitly accepted too – all, by any editors. Fred Gandt · talk · contribs 04:52, 18 September 2025 (UTC)

Yes, this is right. Are you saying you think this should go in the Reviewing edits by a single user section, or the Reviewing edits by multiple users section? Danbloch (talk) 18:05, 18 September 2025 (UTC)

I'd say it forms a crossover between the two conditions, and might require a restructure to (1) make clear and (2) not require potentially confusing repetition of a process description from two perspectives. I see you're currently cleaning up the page, and I don't want to throw you off too much. Ultimately it's that all pending changes prior to any accepted change will be implicitly accepted that needs to be clear, and a process of reviewing all pending changes in chronological order from least-to-most recent to avoid inadvertently accepting unreviewed changes that perhaps should be suggested, no matter if there are multiple changes by any one editor. Fred Gandt · talk · contribs 19:55, 18 September 2025 (UTC)

Hi everyone, I'm from the Russian Wikipedia and made a small contribution here by writing, for example, the article Rocketbank. Since I don’t fully understand how the English Wikipedia works, I wanted to ask: if I don’t have the appropriate permissions and I publish an article, is it automatically unreviewed (i need the reviewer flag for that) or is it considered reviewed? Senya48 (talk) 12:36, 9 December 2025 (UTC)

@Senya48. Hey there. We have our mw:Extension:FlaggedRevs to be configured for pending changes protection only. We use mw:Extension:PageTriage for new article reviewing, but the process is invisible to most people (it doesn't hide the article from anything except search engines). The WP:USERGROUP that lets your new articles skip the review queue is WP:AUTOPATROL. Hope that helps. –Novem Linguae (talk) 15:57, 9 December 2025 (UTC)

Alright, thank you. And can I obtain the autopatrolled flag here without formal procedures, if I already have it on the Russian Wikipedia (where I also have the reviewer flag)? Senya48 (talk) 16:07, 9 December 2025 (UTC)

Unfortunately not. You'd want to apply at WP:PERM/AP after getting a good track record (dozens) of articles created without being sent to WP:AFD. Please also keep in mind that enwiki has stricter notability standards than many other wikis. –Novem Linguae (talk) 17:32, 9 December 2025 (UTC)

Thank you very much . Senya48 (talk) 17:35, 9 December 2025 (UTC)

The page here seems to indicate that there are only a few ennumerated circumstances where a pending change should be reverted, namely the four bullets listed under "Requirements to accept an edit." But this seems to exclude edits that would likely be reverted on an unprotected page. For instance, good-faith edits that introduce incorrect information, disrupt templates, or add unencyclopedic language. I was also under the impression that an edit being unsourced or poorly sourced on its own was sufficient grounds for reverting since, per WP:BURDEN, it isn't the responsibility of other editors to get find sources for other people's edits. TornadoLGS (talk) 02:11, 21 December 2025 (UTC)

Howdy. The 4 bullets are the list of things that it is mandatory for a PCR to check. They can of course check other things if they want. But if it's not on the list of 4 bullets, then they can click "accept" and let other editors deal with those things if they so wish. –Novem Linguae (talk) 02:57, 21 December 2025 (UTC)

I see. I read the language as implying that unsourced edits should be accepted if they are not vandalism or BLP violations. TornadoLGS (talk) 03:02, 21 December 2025 (UTC)