How to avoid an explosion of sub-topics in a topic

These are my thoughts, with an implicit question of how members and mods might respond, when a thread starts to go exponential.

Discourse doesn’t offer threaded discussions, although it does have a sort of back-pointer, in the case that a post is a reply to a previous reply. It also allows for quotes from previous replies.

Overall, if a topic has more than a handful of replies already, it’s possible that the next reply will be

  • a general response to the topic, or to the head post
  • a specific response to the immediately preceding post
  • a reply to any other post, with or without a quote
  • something even more complicated, referencing more than one predecessor

It’s also possible that a reply will be

  • addressing the original point of the head post, hopefully as described in the title
  • continuing from a point raised upthread
  • introducing new ideas not yet raised

It’s in that third case, I think, that things can start to get out of hand. If each post introduces one or two new ideas, there’s an exponential increase in the number of ideas the next post might be responding to, and there’s potentially an exponential dilution of the connection back to the original topic title. The title ceases to represent the content of the discussion.

We don’t have threading in Discourse, so we’re left with tactics like

  • being disciplined about staying on topic, avoiding tangents
  • starting new linked threads when we have an idea which seems distinct but substantial
  • using private messaging to discuss or offer feedback on tangents
  • having one or more posts split off into new topics (by mods)
  • having a thread closed (by mods) to encourage new threads which pick off ideas which arose in a branching topic

(I think it’s been noted before that the mechanism of a linked thread isn’t very discoverable in the interface.)

Edit: also implicit in my thinking, is that exploding topics are a bad thing: they can’t come to a conclusion, because they don’t address a single idea; they might never close, because they contain many ideas that someone might later notice and respond to; they are not a useful record, because they are not coherent. It may be that in some forums they are fine, and in other forums they are not fine, because of the aims and culture of the forum.


While it has maybe it’s flaws we do have linked topic options.

A master topic can be a launch point to explore key points. While I agree the link discovery is not very intuitive. Guiding a community to learn how to use tools can be an asset in this.

ie Op creates broad scope topic (Brainstorming topic)

Each time a Topic element is introduced a new topic could be created with a link to explore that branch. First post in such a topic should have Unlimited Edits first post.

Adding each Keypoint introduced with Link to discussion on that particular keypoint.

Ensuring Ops play a key role is paramount

Sample from my site putting responsibility on Op & Participants

We now even have tools available as mods to make things somewhat easier. Like:

  1. Staff Notice Can be handy to highlight a post to needing a new topic or guide to new one.
  2. Reply Thread can help with selecting posts to split into new topic. By only showing replies linked to a post; of course this won’t capture “Reply to topic vs reply to post” so maybe an insert post could be handy ?

This will vary depending on the nature of the topic.

ie The meaning of life is a pi level topic without a foreseeable likely conclusion.

However an Op may feel the topic has exhausted new possibilities and request the “Brainstorming” topic closed.

1 Like

A general policy that can be helpful for this is to monitor specifically what kind of response mentions of sub-topics are creating.

As in, if an established topic with a clear introduction and many replies about that one specific topic starts to get some unrelated/apparently irrelevant comments, those would usually just be ignored by most people and would not be in any way detrimental to the original conversation, unless that’s being printed in an encyclopedia.

However if people start replying with comments about completely different topics and there is some back-and-forth conversation about those, then that could start to overwhelm and disrupt the primary topic. Then it would make sense to split that to a different topic.

It can be a good thing to have some more open-ended topic threads that are not regulated to only be about a specific topic, as that can be a limitation to the general way people use written languages to communicate. There also may be completely legitimate and important reasons why secondary topics are mentioned as they may be directly related to the original topic in a way many people may not be able to easily understand.


People being people, maybe this is the key - that there is a maximum number of replies or participants that can usefully exist in a conversation. I think this is true for in-person conversations. Just because online discussions allow for potentially hundreds of participants and posts, doesn’t mean that discussions should be allowed to scale to this extent.

I can imagine myself enjoying reading a topic that contains (say) 50 replies between two participants. I can also imagine enjoying an AMA topic that has 50 participant’s questions being answered by a single person. I don’t think I’ve ever enjoyed reading a topic with 50 posts created by 25 users. (All numbers here are approximate, but you get the idea.)

One of the stated goals of Discourse is that reading is fundamental: Because Reading is Fundamental. With that in mind, it doesn’t seem unreasonable to make an effort to create readable topics.

At the risk of introducing a sub topic here, I also think that posting in topics that have an excessive number of posts and participants is unsatisfying. There’s a sense that at best, my post will be skim read.

So to answer the question, I don’t think mods closing a topic is the ideal approach, but I understand why it sometimes needs to be done. I think a better approach would be to have the ability to add some constraints to a topic so that its replies and participants could be limited. For example, a limited pool of users could be chosen or apply to be able to respond to a topic that was set to allow a maximum number of replies per participant. Essentially, this would create an artificial scarcity with the aim of increasing the value of the conversation.


I think both of those are fraught with danger.

If you only allow a limited number of users to respond you’ll tend to hear from the loud keyboard warriors who can spend lots of time online and not hear from the quiet knowledgeable people who can only spend a limited amount of time online.

Equally, if you limit the number of replies thatcan be made, then you can lose excellent insight, e.g. I’ve been reading along and have had my 2 allowed replies to clarify what people are saying and then I have the breakthrough thought that has the answer to life, the universe and everything, but no-one will ever hear it because I’ve hit my limit. Sure I can start a new topic, but it’s out of context and even if you point back at the previous topic how many people will read that before replying to you?

1 Like

Bananas :slight_smile:

And now I have to find my half-remembered reference to sourceforge & change it to stackoverflow - Or maybe I won’t I’ll just leave the correction here


I pick the quote above only as one example of the implication of the memes in the minds of those who wrote the software which now shapes the possible evolution of conversations. Maybe I should have quoted "avoid” in the title & proposed it be changed to ”encouraged” (or “managed”) ?

I could postulate from the quote “therefore there’s a design flaw in this software platform…”
Or "This platform is suited to x but not to y (When with more insight it might have been engineered to be fit for XY and Z)”
Or “as experience builds emergent needs are found that must be accommodated if civilisation is to prosper in the transforms & extensions to digital…”

AND I could assume

"built for the next decade of the Internet.
The goal of the company we formed, Civilized Discourse Construction Kit, Inc., is exactly that – to raise the standard of civilized discourse on the Internet through seeding it with better discussion software:”

is still a current aspiration? Source

Currently, rightly or wrongly the tone/ atmosphere / tolerance/ culture is not conducive to the sort of conversations that derive guiding principle as opposed to code blocks. I say “not conducive” in the spirit - I hope - of your edit that recognise the fact that you had embedded a mindset in your first draft that at least needs recognised and explored somewhere and somehow.
I posit meta is the place to do it. If the discussion is pushed off into another forum then maybe it ceases to be something that will eventually extends competitive advantage to CDCK (?)

Someone said to me recently [and apologies to the author of this comment because I’m adapting slightly] that “programmers write community software without insight into sociology and sociologists don’t want to participate in the conversations about design principles” (and is it any wonder with extant moderation criteria?)

Ed, your “implicit” is correct I think - IMHO - the rest of this quote needs to be examined and the lens used will change the results of analysis.

I pick out your use of the words “can’t”, “don’t”, “not” I agree your final sentence relating to aims and culture. I think if we drew a Boston Square with the correct axes (and the might need more than two axis and or divisions of each axis so it might be a Boston cube of 27 subcubes¡!) We would see your assumption and your choice of words fitted perfectly into one quarter (1/27th) but not the other three quarters.


I think the answer to that for the at least one other quarter is “get out of the way” or “check back later”. Maybe when we have the last “n” posts to read over we will start to see the emergence of now I can see that all of this is related… we distill out of it themes for more focused debate. finding those themes if they are only detectable with hindsight requires leaving a thread because “can’t”, “don’t” and “not” are the wrong words for its characteristics

I’ll leave you with a graphic I drew 20 years ago and have seen the other people have derived it too since (and drawn better than me! So I use theirs) My caption is what do you mean it’s a square It’s plainly a circle…

We only get to a full understanding of the reality when we have many points of view that are accepted as simultaneously valid and apparently contradictory. That tells us that we are still questing for something unseen and when it is seen it might be a breakthrough


That’s a good metaphor with the shape diagram. Also auspicious is how those two dimensional renderings appear is dependent on where is the source of light.


This may be flagged as off-topic and disappear, but here are some solar mapping diagrams that architects use for things like that:


Maybe. I’ll admit that my idea might sound a little weird. There’s a similar solution built into Discourse though. Groups can be given specific permissions about how they can interact with a category’s topics: See, See Reply, See Reply Create. There are a few ways that these group memberships could be updated over time to prevent replies from being dominated by a subset of users.

X/Twitter has a feature where a user can restrict who can reply to their posts on a per post basis. For example, replies can be restricted to users who were mentioned in the top level post.

The essential point I’m trying to make is that the physical world imposes restrictions on conversations. An un-moderated real world conversation rarely involves more people than could comfortably sit at a table. When there are larger gatherings of people, for example at a party, people will naturally divide themselves into smaller groups. There are also time restrictions placed on conversations. They rarely go on for more than a few hours.

Yes, this could happen. The idea of self-imposed arbitrary restrictions is a bit of a hobby horse for me. One of my favorite things on YouTube at the moment is a channel where two musicians impose a 1 hour time limit for writing and recording a song. On occasion they get super inspired by an idea and allow themselves to break the limit.

I didn’t make it clear, but I’m not suggesting applying these limits to all topics.

Another approach for preventing an explosion of sub-topics that’s used on Meta is setting topic timers that delete all replies after 30 days. This is used for documentation topics. The idea is that if a post asks a relevant question that isn’t answered in the topic’s OP, the answer to their question gets incorporated into the OP. This was implemented to avoid having documentation topics with hundreds of replies and multiple sub-discussions.


This is important to consider, and can be very helpful for online communities to have regular meetings over zoom or another platform if people cannot see each other in person.

Forums can be good for curating a community, however it can be difficult to know if people are on the same page about things with only written text/posts/emojis.

Sometimes people just aren’t interested in having any kind of meetings, it can be difficult to organize those across different time-zones. Another way is to do question answer calls to answer people’s questions they’ve written in over a video presentation that can be live-streamed/recorded.


I’m very much with you on this observation and in thinking of ways to apply it online - as you say, selectively.

A crucial point here is that not every conversation has to cover all possible ground. A conversation is about something.

With most forum software including Discourse it’s very cheap to make a new thread aka topic. A topic has a title, a category, and a head post, all of which serve to frame a conversation. One of the tactics we have, which I believe is underused, is to spin off a new topic for a new idea.


These are all valid points. However in a discussion of philosophies all pov are always valid.

Your initial Op statement only covers in one sense a small element as @simon has further expanded.

With Discourse after years of resistance added option to view threaded response to a post(community persistence will eventually succeed in influencing needed change) can help with Simon’s presented issue.

It is quite true when there becomes a large volume of participants even on topic it can be hard to follow with ppl exploring different branches even without having a broad spectrum topic.

Ppl within a topic will often pool together exploring a particular element

Which is exactly why a broad brainstorming topic can work quite well with a bit of guidance and teaching ppl how to manage and encouraging a topic that will spawn new sub topics with SoGs and SoPs helping to make it work.

New topic branch? The post with the link can be highlighted with Staff Notice for example…Op Statement can be editted to identify where to discuss a specific element…Etc…

It is about removing barriers not imposing them.

@simon mentioned the 30 day reset replies topics. Reset Category in a free for all motive can work as well. Start a topic to fletch out an idea. Just before 30 days end. Copy key points modify op statement start a new 30 day brainstorm. Just an idea of sorts.

Which reminds me iirc there was maybe not long ago a plugin adding a form of topic threading. Will look and see if can :wink:

1 Like

Except it’s not cheap :frowning: !!

The cost - as evidenced by the history that brings this thread into existence - is in the moderation against benchmarks or guidelines that don’t acknowledge the technical facility and desirability… So much good faith effort has been spent by all sides :frowning: hence this suggestion

@Heliosurge - I think what you’re proposing is valuable, already technically available,… But doesn’t it require a culture change in outlook by the complete community? Therefore is it feasible? In what time scales with what effort and what preliminary steps - EG create a group or groups ?:slight_smile:

1 Like

We regularly struggle with exploding threads.

Two concrete feature ideas we’ve had that we think would help manage these are:


Splitting a thread at the point where someone has replied with two or three ideas is currently next to impossible. I very much like the idea of duplicate and split, with fine surgery to dissect out the different points.

There are people who don’t like having their posts edited. If somehow they could learn not to combine diverse ideas in a single reply they might end up happier.


Of course splitting is already very labor-intensive, so it’s not perfect, either. I’ve also created target posts that summarize (with my own account/voice) and selectively quote mixed grab-bag or otherwise problematic posts, which then allows me to use that new post as a target for the replies in a manner that feels ok to most participants in the discussion. It ends up with a starting post that is “newer” than its replies, but it’s understandable… and can allow you to also close the original topic in some cases. It’s still so much work, though.

The other strategy here, of course, is just around setting expectations and norms. Clear and concise rules around what makes for an on-topic topic is absolutely essential and can completely do an end-run around “but my free speech” sorts of rants.


I think there’s another independent potential issue: each forum has a culture, and when a person doesn’t pick up on that, and adapt, they may keep coming up against disapproval or moderation. One forum might welcome exploding threads and another might prefer single-topic topics. Some might like essay sized posts while others might like concise single point posts.