The discussion highlights a UX inconsistency in the new unified review queue, where the meaning of Yes and No actions flips depending on the context. sableraph explained that for flagged posts (e.g., “Is this post spam?”), clicking Yes confirms the issue, whereas for topics needing approval (e.g., “Approve this topic?”), clicking Yes approves the content. This inconsistency creates dangerous muscle memory risks for moderators, potentially leading to accidental approvals of spam or accidental rejections of valid content.
sableraph and awesomerobot agreed that standardizing the language so that Yes consistently means “confirm a problem” would be safer. ToddZ expanded on the issue, noting that the terms “Approved” and “Rejected” also have opposite meanings for flags versus queued posts (e.g., approving a flag confirms a violation, while rejecting a post confirms a violation). ToddZ argued this semantic inversion adds unnecessary mental overhead and makes the system intimidating for new users.
Andrew_Rowe shared a personal anecdote about accidentally approving a post he meant to delete, acknowledging that adding a confirmation step would mitigate this risk. He praised Discourse’s continuous UX improvements despite the initial confusion.
sableraph later bumped the topic, emphasizing that the rise in AI-generated spam increases the volume of review items, thereby heightening the risk of such errors, especially for new moderators. Andrew_Rowe directed users to a related topic for further discussion on swapping the approve/reject buttons.
I love the new unified review queue but I noticed a slightly confusing UX pattern when processing a lot of items in a row.
You might review several flagged posts that ask “Is this post spam?” where clicking Yes means taking action against the post (with a confirmation). Then the next item asks “Approve this topic?” where clicking Yes does the opposite (without confirmation).
When you’re scrolling through the queue quickly, it’s easy to build up muscle memory and accidentally click Yes and approve a topic when you actually meant “Yes, this is spam.”
Thanks for the feedback, I can see how that could happen… do you think it would make sense to flip things around so we’re more consistently asking to confirm a problem?
We already ask: Is this post [spam, inappropriate, etc]?
And to flip around the “approve this topic?” situation… I guess we would instead ask “Is there anything wrong with this topic?” then yes would consistently be “yes something’s wrong” right?
Definitely. I’m not sure what the best copy would be, but flipping the language so Yes always means “take action against this post” feels like the best approach.
Yes, please! And while we’re on it, there may be related opportunities to make the review system feel more consistent and approachable.
We have two basic types of review items:
“Post flagged as _______” and
“Queued post Needs Approval”
And the resolutions Approved and Rejected have opposite meanings for each class:
Issue confirmed with a flagged post? → “Approved.”
Issue confirmed with a queued post? → “Rejected.”
The colored resolution pills are clearly labeled, which is nice: ,
…etc. But these are located fairly far from the “Timeline & notes” area. When scrolling items and scanning notes, my brain wants to think in the context of posts and what’s done with them, but instead I have to think about what Approved or Rejected means in context of the review item type:
note
meaning - if flag item
meaning - if queued post item
“Approved by”
Flag approved (problem confirmed)
Post approved (no problem confirmed)
“Rejected by”
Flag rejected (no problem confirmed)
Post rejected (problem confirmed)
And a little thing, in the filter drop-down:
Approved and Rejected just find Flag approved and Flag rejected items.
only All reviewed or Everything will find Post approved and Post rejected items.
None of this makes the system unworkable – it’s just a bit of mental overhead that, IMHO, contributes to the review queue system being a little confusing and intimidating at first.
Maybe this should be a separate post: “Inconsistent meaning of resolutions in the review queue” – but the matters seem pretty closely tied.
I found the review queue a little confusing at first but quickly learned to look at my choice twice to make sure I was doing what I intended.
that caught me out once at first but I meant to delete the post and instead approved it so I went directly to the topic list, found the topic and manually deleted the post. Then I went back into the queue and explained my actions in case anybody (staff) was wondering wtf
anyhow, providing just the confirmation would help
I tend to just figure out how things work and except it and one of the things that blows me away about Discourse is how it is constantly being refined especially the UX
Just a friendly bump to ask whether this still being considered. With the increase in AI generated spam, the issue is becoming more pressing: more items to review make it more likely that someone will click the wrong button by mistake, especially for new moderators who are not familiar with this quirk of the UI.
,
