Issues when using slow mode on a topic forever

Oh, just to clarify: I posted just to inform that we were also using slow mode permanently, not to second @dfabulich’s request – I’m also not exactly clear on what he’s requesting.

We actually had indeed increased the editing grace period, for the reason you said.


Let me begin by making a general remark about the “editing grace period” when slow mode is off, i.e. in the normal case.

The editing grace period is a really weird feature, because its normal/official purpose is to prevent the creation of a history entry for “fast” edits. I honestly don’t even know why that would be wanted… don’t we basically always want history entries for edits?? Like, even if somebody just fixes a typo ten seconds later, I don’t understand why we wouldn’t want a history entry for that, or at least a little “edited” indicator. (Even in chat clients that support editing, they normally say “(edited)” on all edits, no matter how fast.)

I think it’s especially confusing to have an editing grace period that’s long enough for someone to reply with a correction, and for someone to edit their post in response to the reply. This invalidates the correction, with no evidence of the original comment.

Anyway, in the Covid thread, the edits that folks wanted to make were not “oops, typo” edits, but substantial “oops, I got that wrong” edits, which typically arrived in response to corrections or requests for clarification, e.g.

  • “I see now that this is fake; sorry for posting it. Unfortunately, I can’t now edit my post to remove the link.”
  • “What I meant by that is _____. I wish I could edit my earlier post to clarify.”
  • “I no longer think this is right; see my post #77 from below. I’ve pinged the moderators to update my earlier post, because I can’t do it myself.”

I guess we could crank the editing grace period up to something astronomical, like hours or days, but (since the grace period is a site-level setting) that would completely eliminate all edit history entries on the entire forum. So, I claim, we really don’t want to do that!

To be clear, when I said “overblown” I don’t mean “less likely than you think,” but rather “less serious than you think.” If/when folks started abusing the “edit loophole” to turn the Covid thread into a chat room, we’d temporarily suspend the people who did so. A firm hand of moderation can easily solve social problems that can only be clumsily solved with technology, IMO.

Which feature decision are you referring to here? Whose confusion?


I definitely see @dfabulich’s point here. It is a bit weird that editing outside of the grace period resets the posting timer.

1 Like

This is my thought on it. Slow mode exists (in my mind) to make moderation easier, not replace moderation. IT’s for when things get heated. People who abuse features on the site of any kind are moderated in other ways. Play nice, or get off the playground, as it were.


One case that can happen is that a post gets flagged and the flag accepted. The poster then gets notified that they should edit their post to guidelines and so it will be automatically restored, but because of slow mode they can’t do this for an hour or whatever the delay on the topic is.

That’s an exceptional case, but we’ve hit it a couple of times as we tend to engage slow mode in response to heated posts where a little bit of clean up is needed.

Slow mode in the whole is still an excellent and worthwhile feature though, and has improved discussion on some topics where mods have been biting their fingernails, but slow mode has ‘saved the topic’ and things stayed constructive and civil.


No, we don’t, because it’s pointless to have a permanent record of a simple spelling mistake. The grace period is there so you can fix mistakes that you didn’t notice until you pressed the “submit” button. This always happens. Once you see it “printed in bright lights”, the little errors and mistakes suddenly jump out in a way that they didn’t in the editor.

And they couldn’t do that in a normal reply, like a normal person, in a normal conversational flow… because…? Perhaps because the posting interval is way too high, and should be reduced? I think that should be your first port of call here. Your users are telling you the post interval you’ve selected for slow mode is too high, and should be reduced. The new version of the UI is going to RADICALLY emphasize this by front-loading the time intervals to much shorter times, and removing the “optional” part of the slow. (Don’t worry, you can still do forever, but you must scroll down and select it, to add friction.)

I’m sympathetic to the “oh no, I have posted a link to what is clearly dangerous misinformation” case – but if it’s really THAT dangerous, why not flag it for moderator attention?

And you can’t suspend people who use the “grace period edit loophole” for the exact same reasons, because…?

I’m definitely open to the idea of adding one allowed edit in slow mode, but I want to roll out the other UI changes first… and in the meantime from what I’ve read, your community is telling you the slow mode interval you’ve selected is a just a bit too long. Try reducing it.

Completely fair point, and dovetails with the “oh no, I’ve posted a link to REALLY DANGEROUS misinformation” scenario.


I think @dfabulich touches on a critical distinction: there are (at least) two different kinds of edits, which the Discourse software could potentially treat differently in the future:

  1. Minor corrections—e.g. typos and spelling.
  2. Content correction and clarification—i.e. “setting the record straight.”

For (1), the e.g. 5-minute editing grace period makes sense. I personally make heavy use of this feature, and agree heartily with @codinghorror about the “printed in bright lights” effect.

However, I think using the editing grace period to address edits of type (2) is not the best solution. Instead, imagine if Discourse had a sort of "track changes" feature where edits beyond the grace period did not erase anything, but instead struck things out and highlighted new things with e.g. yellow-background CSS. This would be amazingly useful for addressing misinformation and clarifying previously unclear statements, in a way that did not invalidate later posts offering that feedback in the first place. It would also assuage the fears of those with “edit feature hesitancy” who (at least in my community) have pushed back against the fact that posts can be edited at all, for exactly the reason that erasing past statements can create more confusion than it solves.

As for why not “do that in a normal reply, like a normal person”: in my experience, normal people are reactionary. They read something at the top of the thread, and don’t always notice corrections in later posts—especially in dense/busy topics. It would be much, much better to correct the OP itself, nipping problems in the bud so to speak.


This feature already exists, and has since the earliest days of Discourse; click or tap the pencil icon at upper right of any edited post. The pencil will be bright red if the edit was recent, and fades to black as the edit becomes less and less recent.


Note that grace period edits (first 5 mins after posting, by default) don’t create edit revisions.

This is also why I am VERY skeptical that people will notice hour-later edits in a topic that’s set to Slow Mode. Especially if a few people have already posted after you. Those readers are really gonna scroll back up and see, oh, I noticed you revised your opinion on this? That’s… not how people work, in my experience.

1 Like

The scenario here isn’t people who have already read the post & need to scroll up, but new readers who haven’t yet read everything. The edited post will thus remove the incentive to pile on to the “person who is wrong and hasn’t yet edited out the wrongness and so is clearly an <insert inflammatory word> and needs to be called out on it”


Right. But what I’m speculating about here is: what if it weren’t hidden behind the pencil button? What if edited posts just showed the changes visually in the post itself when they were done beyond the grace period, and/or if they were too substantive—without users needing to explicitly seek that out by clicking something? I think there could be a lot of value to that. It would help combat the misinformation scenarios. Better balance concerns by showing the provenance of what was said when, while still discouraging people from reacting to obsolete statements. I also expect it would help reduce the phenomenon of people using edits on slow mode topics to circumvent the time gating restriction: posts would quickly become very ugly if repeatedly edited, and I think most people would naturally eschew that.

I’m not necessarily advocating strongly here, and I certainly don’t think Discourse should always work that way—e.g. it wouldn’t work well for “guide”-style topics used in many communities where the OP does edits over time to expand and revise the material. Just offering food for thought.


That is an excellent point. However. It’s also true that we expect people to read conversations, and we expect positions to evolve in conversations through multiple responses. So expecting that someone has a single post in the topic with precisely and exactly their final understanding … is unrealistic.

I definitely hear you on the triggering XKCD 386 factor though. If the triggering bit can be edited out, then all subsequent replies will be better ones, and the whole topic improved … :thinking:

I want to reiterate that I am firmly a fan of “allow one edit”, especially in the case of “oh my, I’ve posted very dangerous misinformation”, but I feel the risk of this is being overstated by folks a bit, and they are understating the risk of people turning the topic into a live editing chat session… but allow one edit is a good middle ground here, too.


Part of what we are worried about here is people editing in bad faith, and part of it is people anxious to correct an error in their post but not concerned about breaking the logic of subsequent responses.

Sometimes when I’m wrong (on the internet) I will post a correction or acknowledgement of correction, and also add “Edit: oops, see below” or similar to the post with the error. (I always act in good faith, of course.) Just possibly, if this situation is common, it would be a reasonable feature request to make a ‘correcting reply’ or similar, which is a new post at the bottom of a thread, but which annotates the errant post with a postscript. (And it does that without allowing an edit, in the case where an edit would be disallowed.)

A second point: some in this thread are using slow mode as a means to wait for moderators to come on duty, so they can monitor and react as a thread develops. That may be valid, but there’s another model too, which is the model where slow mode is a means to prevent rapid-fire responses which might raise the temperature, but allow slower and hopefully more considered responses - and all this, without moderator presence. Personally, I’d expect to use the second mode, and for that an hour should be long enough to cool down. If people are still hot-headed after an hour, they need a conversation with a moderator.

I have myself temporarily closed posts, which didn’t go down well, and now that slow mode is available I will in future use that.


Well good. Let us know how it goes. I’m becoming more and more open to “allow one and only one edit after the grace period”. The examples have convinced me:

edit: this task has been added to the 2.8 release list in #releases