Plugin: Topic Summary Section

Note: There’s a seemingly similar topic about a topic summary here, but it’s actually quite different from what I’m proposing, although some elements might be similar.

Would having a summary section at the top of a topic be something that sounds useful to you?

This is something that’s been on my idea list for ages, and I believe it could possibly transform discussions.

I’ve always found that reading through all the posts in a thread can be way too much if I only want to know what’s the current status of things. (Nice to follow the evolution of thoughts and get deeper insight, but not everybody has time for that). And just skipping to the end also doesn’t work, because the actual important posts might be somewhere in the middle.
Especially in forums that are supposed to discuss about improving something, there might be a lot of ups and downs and backs and forth’s, and what I’d really love to have is something that summarizes all that, preferably at the very beginning, so that I know what’s going on and only need a minute.

How would this summary work?

I imagine the summary to be an editable text field just like any post, and anyone can go and make changes, just like wikipedia. Possibly changes could or should only be allowed when you’re also posting something to the thread, which would kinda work like the justification for your change to the summary.

Let me illustrate it by using this very conversation as an example:
Say, the first person (me) starts a new topic with an initial post (this).
Discourse would then prompt me: “Do you want to edit the summary?”
I’d go sure, and add the first text to the summary:

A summary could help new people to catch up quickly and will help the community to come to better conclusions.

Ok, so far so good. Now person B comes in, and might say as a response:
“I’m not sure whether it’s worth doing that. I’d imagine it to be quite difficult to implement, and wouldn’t it just add to the noise?”, post it and then be prompted whether to update the summary:

A summary could help new people to catch up quickly and will help the community to come to better conclusions. It might add to the noise though, and not be easy to implement.

3rd person C: “I think it would be a great idea, but I agree with B, might not be worth the effort. … Unless… Could this be used in our proposals and votings [or something else that the particular community might be interested in] as well? Hm, maybe we should have a list of pro’s and cons for this, what do you think?” C decides not to update the summary, he’s not sure what he could change that would actually make it better.

4th person D: “Yes that could totally work for our user engagement and during the process of coming up with proposals. And yes, I second C’s idea to make some pro’s and cons, I have some more:
We could make it so that each edit is linked to the editor, and when you click on it it will jump directly to that post, it’s basically a quick navigation. Of sorts, right?”

A summary could:

  • help new people catch up quickly
  • help the community to come to better conclusions
  • act as ‘quick navigation’
  • Not limited to this discourse forum

However, possible downsides are:

  • It might add to the noise
  • Probably not easy to implement.

B jumps back in: “Hold on, hold on a minute. Before we go down that rabbit hole, what is that summary actually supposed to summarise? Is it just a rephrasing of every single post that’s been done? That wouldn’t make all that much sense, right? Could we have a definition of that first? I think it should only summarise what is actually important, i.e. only what is contributing to answering the OPs question, right? Otherwise I stand by my earlier statement that it will just generate more noise, or more overhead in what users need to type, and UI overhead, etc…
Btw, D, I found your ‘quick navigation’ was a bit hard to understand, so I modified it a bit, hope that’s still ok now?”

A summary should summarise all the relevant content in a post that contributes to answering the authors question.
It could:

  • help new people catch up quickly
  • help the community to come to better conclusions
  • sections in the summary can be hyperlinked to the editor
  • Not limited to this discourse forum

However, possible downsides are:

  • It might add to the noise
  • Probably not easy to implement.

User D: “Yes thanks B, you’re right that’s not necessarily obvious what I meant. I think I’d still like to have the quick navigation mentioned there though, because it is a somewhat separate point IMHO. Linking something up gives a bit of … verification? Authenticity? Not sure what the word is there…
Oh, and just rephrasing my initial point about using it outside of this discourse forum a bit…
What I actually meant with that is that I think it can also be used outside of our specific focus of proposals and voting. E.g. it could be used to keep track of progress of tasks, of the status of something, could be a more comprehensive insight to any kind of question, … the possibilities are endless.”

A summary should summarise all the relevant content in a post that contributes to answering the authors question.
It could:

  • help new people catch up quickly
  • help the community to come to better conclusions
  • sections in the summary can be hyperlinked to the editor
  • Quick navigation to details for individual statements
  • Can be super useful not only for proposals & voting

However, possible downsides are:

  • It might add to the noise
  • Probably not easy to implement.

Then user X comes in, sees this last version of the summary, everything’s intuitive except the ‘hyperlinked to the editor’ bit, so he’d hover over it, see that user D’s post was the first linked entry, and user B’s post the second. (Some kind of info on hover popup). He can directly click on the link to D’s post and read the details from there on, with B’s directly afterwards (there might have been others in between that would be collapsed because they didn’t contribute to the change).

… ok, I hope you got the picture. :wink:

Super breezy to catch up, isn’t it? (if you imagine only having to read that summary!)

Well, it ain’t gonna be easy to write this plugin in all the glory details that I imagine (There’s a lot more that this plugin could be doing, some of my yet unnamed faves would be to increase or decrease font size/weight/color based on votes/reactions associated with the contributing blog posts), and some of it might be a bit tricky logic (e.g. correctly attributing the changed bits to the author, and visualizing what happens if someone just deletes a ‘not’, etc… ) but I’m sure it’s all feasible and can contribute immense value to many discussions.

(This was much more than I intended to write, but it was fun coming up with that example :laughing: )

What are your thoughts?
Is this kind of plugin actually technically be feasible in the discourse ecosystem?
Do you need more information? More serious mockups? (I’m not good at that, but could try creating some)

Disclaimer:
I could try to develop this myself, and I might, but I don’t have any ruby / ember / discourse experience so far, so would take me a while.
Also, the idea for this is a bit older but got a bit more attention in a community group that would find this super helpful, so I might also do it there and do a bit dogfooding… but again, it would take me a while, and if it’s of value for discourse then having some of you involved would be super cool!

Cheers :slight_smile:

6 Likes

Sounds great! I think it’d work well in conjunction with the posts-focused mechanic I proposed in the previous topic.

Might you be able to make a few doodles/mock-ups of the essential UX this new feature would introduce?

You might get quite far with a prototype made as a #theme-component.

3 Likes

Thanks for pointing me to #theme-component , that looks interesting.

Is it possible to create a theme-component and then hook that into a custom plugin for backend functionality and some other things that might be out of scope for theme-components themselves?

Also, yes, it might work really well in conjunction with what you proposed previously.
From skimming over the discussion in that topic the main takeaway as I understood it was that most people think the ability to edit the first post would give you already enough control to keep the topic ‘clean’.
My proposal is similar to editing the first post, however, there are a few big differences that I believe make a major difference:

  • Posts can only be edited by the original poster (or mods/admins), and the idea of the summary field is explicitly to expand beyond that limitation
  • There is no association of these edits to who edited it and why, which is something that could (should) be done for the summary. (I’ll try to mock that up a bit more soon)
  • There should be a history for the summary so that people can see what’s changed when and how (wikipedia-style)
  • Ultimately there could (should?) be more fine-grained notifications associated with edits. E.g. I want to be notified if one of my contributions to the summary was edited.

The whole idea comes out of the desire to enable online voting / decision making, and to do that people will need to be able to discuss proposals in a meaningful and safe (transparent about edits) way.

2 Likes

Yes, topic summaries would be an excellent way for new readers to get up-to-speed. Plus, with an ability to view a wiki-like revision diff of a summary since your last visit, a quick way to catch up.

As you’ve written, summaries should include permalinks to individual replies, so that people can read the sources of the aspects of the summary that interest them.

Two suggestions:

  1. A summary should be associated with a list of the replies it summarises. This makes it easy for those writing summaries to filter replies by their currently unsummarised (and unedited) status, and so easily incrementally summarise replies from only this pool, including new replies as added.

  2. Instead of a single summary at the top, allow multiple summaries, which can each be normal replies that have a summary tag. Multiple summaries are useful when there’s more than one side of a contentious argument (preventing edit-wars on a single summary), or when there can be sub-topics (in which case replies to a sub-topic summary can form the root of that sub-topic’s second-level thread).

1 Like

Yes, summaries would really make it much easier for people to digest and participate in long, large, complex, and contentious debates. Without them, it’s not feasible to democratically construct pro and con cases for something like a plebiscite. The aim should be to provide an environment that helps participants get closer to consensus on the balance of facts and to the nub of differences, and hone one or more of the most supportable conclusions.

One thing that would help here would be to add Genius annotation to topic posts. This would allow annotations (each with a Genius comment thread) to be attached to fragments of topic text. Such close scrutiny should be a powerful incentive for the topic text to be edited to address rebuttals, setting off further rounds, until the topic and rebuttal texts are perfected. The final rebuttals can be used to build a rebuttal summary reply.

This would unfortunately create a duplicate comment system. A native annotation system that integrated with Discourse comment threads would be preferable.

1 Like

Hey mrj, thanks for those ideas, I think they’re great!

I can see how having multiple summaries, one for each ‘branched out’ conversation, might be great!
Possibly it could even have some ‘levels’, i.e. one master summary and a few sub-summaries (and possibly sub-sub-…), and contributions in each sub summary could flow back to the parent to some degree.
One could also collapse all blogs and only show each summary, and then expand whatever ‘belongs’ to that summary for more detail without getting distracted by posts that talk about other parts of the conversation.

Genius annotations (funny btw, I’m working for a company who’s main product is called ‘Geneious’ ;-)) could be useful in any case, not only for summary-containing topics.
I could see that also working quite well together with summaries, if it’s possible to use these annotations as ‘suggestions’ when you edit the summary, or maybe some kind of quick link to ‘add highlight to summary’. And once there’s also weighing, annotated bits could weigh more.

1 Like

Here’s a SVG with the basics…

Discourse Summarizer

Not too sure about the highlighting based on post interactions & markup. I think it might be quite good sometimes to have it automatically bolder if there’s a lot of interaction, but then it’s hard / impossible to differentiate from normal markup. (Also, it might be sometimes bad having that, another path to game the system…?)
Possibly markup shouldn’t allow bolding & different font sizes?
Or just no automatic weighing.

3 Likes

Yes, all good. There would however have to be some thought on how to best handle sub-sub-topics in two-levels of comments.

That’s a good idea to use annotations of summaries as edit suggestions. The advantage of that would be that the big Discourse change of allowing a comment to have multiple editors wouldn’t have to be implemented. The initial writer of a summary would be the assigned editor of that summary, and those who have suggestions would add annotations. Anyone who disagrees with an editor’s decision can write their own summary. It may also be easy to allow ownership of a summary to be passed around, even though there’s only one at a time.

As for a more widespread use of Genius, I’d have to see whether the Genius system allows comments on annotations to be disabled. Comments on annotations would then be forced to use the main reply threads, hopefully with a way to link to individual annotations.

I’m not sure that the ability to edit a/the summary of the discussion (meaning all replies, not including the post to which they’re attached), should be limited to those who’ve added replies. Just like Wikipedia benefited from open editing, I think it should be open to all. But it would be wise to give mods the ability to ban users for malicious edits, plus allow mods to limit edits by user reputation/badges, like Wikipedia’s “semi-protected” article status.

And rather than display per-sentence hover diffs, I think Wikipedia-like colour-highlighted whole-summary diffs would be easier to scan. This would be the hardest thing to implement, so could be deferred until the basic summary mechanism works.

Differentiating links to individual comments by their rating is a good idea. But displaying the score of each comment would be better than messing with readability through shading.

1 Like

I just found out that comments can be made into wikis by both the commenting user and admins, with the ability to compare revisions. So the hard work is already done, and a summary plugin just has to make them easier to create and use.

1 Like