User:Aquillion/What is a revert

This is a loose sandbox of thoughts on what defines a revert. Feel free to edit it with additions or thoughts.

The increased use of WP:1RR restrictions makes it more important to nail down the precise definition of a revert; 3RR restrictions are rarely breached by accident.

Note that these do not cover 3RR exemptions, which are never considered reverts for 3RR/1RR purposes.

What is a revert

• Most edits made using the undo, rollback, or restore-specific-revision functionality are trivially reverts.
• An edit that is clearly intended to undo the actions of other editors is generally a revert.
• An edit that restores or removes a particular disputed version is always a revert, regardless of the age of the dispute.
• An edit whose substance is clearly functionally identical to restoring such a disputed version is usually a revert.
• Removing recently-added text is almost always considered a revert.
• When an article is the subject of an overarching dispute between two states, an edit that clearly moves it back from one state to the other is usually considered a revert.
• Altering disputed text in a way that undermines the intent behind it is usually considered a revert, but this can get complicated; see below.

What isn't a revert

• Almost any removal or revision of text could notionally be considered a revert (since some past editor's text is being revised or removed.) However, if an edit doesn't meet any of the definitions above, and is not part of any sort of current or previous dispute, it's not usually considered a revert simply to modify or even remove older text in an uncontroversial fashion.
• Edits that uncontroversially modify or remove parts of existing text in ways that do not change its meaning or intent are usually not considered reverts, but some caution is required here, because editors in disputes may disagree about what the essential meaning of the text is and what the intent behind it was; if the text has already been clearly disputed, it is safest to assume that any further edits to it might be considered reverts.
• The initial addition of dispute tags is generally not considered a revert (although of course restoring them once they've been removed is.) This is true even if the text in question was previously "disputed" in the more colloquial sense by removal or revision.
• Edits that solely add new material, that was not previously present in the article in any form, are generally not considered reverts.

Uncertain

• If someone removes some text, and it gets restored, would it be a revert for them to then add an in-text expression of doubt, or to otherwise make an addition that effectively seems to downplays the related text?

• Example: An editor removes a sentence saying "X is true", and is reverted. Then they add attribution to it, making it "Professor Y says X is true." Is the last edit a revert?
• One possibility is that this is contextual and depends on how severely the text is contradicted (and therefore by how much the intent is undermined.) But editors in a dispute can and will claim that the simple addition of attribution undermines their addition by converting it into an opinion.

• What if, in a similar situation, they add notionally unrelated text that effectively waters it down? What if one editor makes a big addition intended to shift the article's weight one way, and another editor makes another big addition to shift it back the other way, effectively undoing the first editor's efforts?

• Example: An editor removes a sentence saying "Professor Y says X is true", and is reverted. Then they add another sentence immediately before or after it saying "Professor A says X is false." Or even just add a big unrelated block of text elsewhere in the article, shifting the balance of the article as a whole back towards where it was before.
• This example is almost certainly not a revert (by this logic, once an editor in a dispute over WP:WEIGHT has made any addition to an article at all, they could frame any additions that differ in perspective as reverts, forever.) But it shows how the line here has to be handled cautiously.

Thoughts

It is helpful to look at the intent behind the WP:3RR. The core purpose of it is to prevent edit warring - that is, a situation where the article is forced rapidly back and forth between two states. This is why the WP:1RR makes it tricky; for 3RR violations, it is usually glaringly obvious that the article has started to swing back and forth, but for the 1RR it is not.

This is why the example of "adding new text that changes the balance of the article" is not a revert. Even if the article's balance is swinging back and forth (and the editors are undermining each other's intents with their addition), the article itself is progressing, rather than cycling disruptively.

This is also why the "adding attribution" case is so thorny. In an article with 3RR restrictions, it would be obvious if editors were going back and forth revising a sentence over and over in ways that shift it from authoritative to non-authoritative; but as a simple sequence of just three edits, "fine, let's at least add attribution" intuitively seems like behavior we would want to encourage - the article hasn't really started to fluctuate; the addition of attribution represents genuine progress rather than just spinning in place.