Should Discourse make an effort to become a viable comment platform?

I’ve created a few posts recently about using Discourse as a comment platform and improving its existing functionality for embedding comments:

Bases on the lack of response to those posts, my suspicion is that there isn’t much interest from the Discourse team or community in pursuing the use of Discourse as a comment platform. That’s fine from my point of view, but I’d like to make an attempt to clearly state my case.

At some point in the 2010s, news organizations pivoted from hosting their own comment systems to relying on social media for discussions about their articles: The Trend Of Killing News Comment Sections Because You 'Just Really Value Conversation' Stupidly Continues | Techdirt. It turns out that comment sections do provide value for news organizations though: Killing Website Comment Sections Wasn't The Brilliant Move Many Newsroom Leaders Assumed | Techdirt.

Recent changes to Facebook and X/Twitter are highlighting other problems with relying on social media for news article discussions:

Facebook is making an effort to get out of the news business: Facebook looks ready to divorce the news industry, and I doubt couples counseling will help | Nieman Journalism Lab. As a response to a recent Canadian law “News links and content posted by news publishers and broadcasters in Canada will no longer be viewable [on Facebook] by people in Canada.

The issue with X/Twitter will be obvious to anyone who’s on that platform. The recent change in ownership has entirely changed the tone of the site - any main stream media posts are flooded with “lol” type comments. I suspect the posts generate some engagement, but the responses must be disheartening for the journalists who write the articles.

It seems obvious to me that Discourse could offer a potential solution for this issue.

The closest competitor I can see in this space is Coral (an open source project funded by Vox Media.) Coral’s philosophy is outlined here: Five Myths of Community Design - Coral by Vox Media. I’m seeing very little that Coral is doing that couldn’t be accomplished with a few improvements made to Discourse’s embedded comment functionality. On the back end, in terms of moderating and structuring conversations around articles, I think Discourse could easily improve on what Coral is offering.

I’m not sure that promoting the use of Discourse as a comment system would generate a huge amount of revenue for the Discourse team, or for some other group that wanted to take it on. I do think that it could help to improve online journalism and democracy though.

For full disclosure, in my free time I’m currently working on a WordPress plugin for improving comments that builds on some of the ideas I mentioned here: Active moderation mode - a slow mode alternative. Even if I can pull that off, it won’t have the same kind of impact as a Discourse commenting system could have - it might get used by blogs, but it’s unlikely to be used by MSM news sites.

16 Likes

Now that the bot knows how to read:

I think this is a somewhat even analysis by GPT-4.

Building a proper read/write commenting system is a huge undertaking. Easily filling an entire cycle (6 months) for a 5 person team.

I simply do not see us as having bandwidth for this at the moment, coupled with a huge amount of JS modernization work that is still in flux.

Given this is very unlikely to be picked up by the team at Discourse in the upcoming 12 months, I think the only viable path forward would be for a third party to prototype / build a solution. With the correct scoping and team on it I would consider sponsoring at least some of this.

But the hard thing here is that there is lack of interest, this would be highly speculative.

9 Likes

I don’t think it should be a read/write system. Comments should be created on the forum, just as they are now.

The For arguments are failing to note that this is timely due to recent changes at Facebook and X/Twitter. There’s currently a lot of discussion, at least in Canada, about how news organizations are going to survive in the digital age.

I think the Against arguments are overstating the technical difficulty. Possibly there would be a need to add an oEmbed endpoint for sites that didn’t want to display comments in an iframe. (Coral gives the option of either displaying comments in an iframe, or pulling them in from an oEmbed API endpoint.) Otherwise it should just be a matter of improving the existing comment embed functionality.

2 Likes

I am confused though, if the proposal is not for read/write. What makes it viable? Friction is ultra high for commenting if it is not read/write.

Is this merely about fixing bugs?

3 Likes

Some friction is good. It reduces the likelihood of low effort commenting.

Existing comment systems that allow users to comment directly on the article may be introducing friction via the moderation process anyway. For example, the CBC comment system moderates all comments and may hold a comment for 30 minutes or more before approving or rejecting it. A Discourse powered comment system could allow for a more natural, real time conversation to occur on the forum, while introducing a time lag between when comments were posted on the forum and when they appeared on the website. This would give moderators a chance to work with commenters to clarify what they are trying to say, consider other points of view, consider if what they are posting is really something they want to be sharing with the whole internet, etc. Essentially, the best posts that were created on the forum would get published to the website.

Having comments published first on the forum also solves potential SEO issues that can come from having lots of irrelevant comments published directly on a web site.

It’s more about improving existing features. Also, possibly adding some moderation features, and developing and advertising a method that can be used for Discourse powered comment systems.

Feature improvements:

  • fix issues with embedding external posts on Discourse. This is currently too difficult to configure.
  • for topics created from embedded posts, allow staff to mark which posts should get pulled to the website
  • improve the existing comments template
    • add tabs to the template to allow selection between “featured”, “all” (a sanitized version of all comments in chronological order), and possibly “related” (allow moderators and trusted users to branch off a discussion into a related topic.)
    • add a “show more” button to the comments template to allow comments to be pulled to the website in batches
  • allow comments from a restricted category to be pulled to a website. This would give the option to have comments from private discussions on Discourse get embedded on a website.
  • make sure comments are being cached on Discourse to ensure that anon users visiting the website doesn’t put a huge load on Discourse

Possible new features:

  • allow moderators (and possibly trusted users) to create topics where they can restrict who can reply to users who are mentioned in the OP. Here’s a use case: Fred and Bill are arguing endlessly about gun control in the main discussion. Allow a moderator to create a separate topic for Fred and Bill that sets them a task, for example, have them follow Dennett’s rules for critical commentary to discuss the issue. If they can pull that off, allow them back into the main discussion. Maybe even mark parts of their related discussion to be published to the website under the comment template’s “related” tab.

  • allow moderators to easily exclude certain users from a discussion, without having to suspend them from the site as a whole. For example, Fred and Bill in my above example could be temporarily excluded from the main discussion until they’d worked out their differences.

  • create a slow-drip AMA plugin if one doesn’t already exist. For example, instead of having an AMA all in one go, allow users to submit questions or comments to an expert and have the questions and replies published to Discourse (and the related website) over a period of a week or so. (For Canadians, my idea here is a Discourse powered Cross Country Checkup.)

  • add an oEmbed API endpoint for retrieving comments

Developing and advertising a method for Discourse powered comment systems:

This seems like the most difficult part of the task to me. Figure out how embedded comments, Discourse’s moderation features, the Perspective API plugin, the Subscriptions plugin, etc, can best be configured to meet the needs of a mainstream news organization. Talk with journalists, publishers, and news readers to figure out how to best meet their needs. Try to address current issues related to polarization, lack of consensus, and lack of trust in a sensitive but effective way. Always keep in mind the idea that journalism is an essential pillar of democracy.

After having done the above, advertise Discourse as a commenting and community platform for news organizations.

The technical aspects of what I outlined above don’t seem overly technical to me. Unless I’m missing something, it’s all within the range of what I’d be able to pull off myself.

I’m involved with a news site which has trialled utilising Discourse for commenting.

We have a well established and active (but closed) Discourse instance, so were very keen to bring the two together via commenting.

While there were many benefits (especially by drawing community members into making high value comments on news articles), I’m sad to say that we abandoned it after a solid trial. The key issues for us:

  1. Configuration was tricky (not surprising with two complex platforms involved!)

  2. Friction from users having to leave the news site in order to comment

    • we streamlined our joining process to facilitate this, but it was still too much friction
  3. Insufficient tools for highlighting the discussion within WordPress

    • the rather rubbish native WP comments showed a speech bubble and count on discovery pages; this was sorely missed
    • the obvious appearance of active engagement on as many topics as possible was very important to the news team
  4. Difficulty in getting the news team to actually engage with Discourse

4 Likes

I wonder if this would be as big of an issue for an extremely busy MSM news site. For example, a site that’s reluctant to display comments because they’re getting too many comments from trolls and haters.

I wonder if any UI changes could be made that would make it clear that comments being displayed on the website were highlights from a larger discussion that was occurring on the forum. That way, readers wouldn’t have the expectation that they’d be able to comment on the website, then be disappointed to see they had to visit the forum. In either case (commenting on the website vs commenting on the forum) users are going to have to create an account and login before leaving a comment, so the friction seems more perceived than real from my point of view.

That’s a good point, especially in terms of embedded comments that aren’t adding any data to the news site’s database. I can think of a WordPress specific solution to this though. It would be somewhat complex to setup. (I can give you a hand with this if you’d like to try it again.)

Yeah, I think the news team would have to be onboard for this to be worth pursuing. It might also be the answer to the friction issue. If potential commenters knew they’d have a chance of engaging with the article’s author on Discourse, they might be more likely to visit the forum.

2 Likes

In confused by this, did WP-discourse drop comment counts recently? I’ve implemented Discourse commenting on Wordpress sites and they were definitely a thing unless the Wordpress theme in question bypassed the supported method to get them.

3 Likes

3 posts were split to a new topic: Trouble using embedding in dark mode

I’m not sure - to be honest, this came up as an issue late in the piece and we never pursued making it work. The actual dealbreakers were really these two points:

But it is good to know that it is in there if we go in that direction once again!

At a U.S.-based local news publication planning to use Discourse for our journalism, I’d like to voice support for this general direction.

If the goal is to improve the discourse around civic issues covered by news media, and the primary obstacles identified are onboarding staff and user friction: I’d rather think about the solution as Discourse being used as a CMS, where a category can publish to a separate domain for the “news publication.”

The technical aspects discussed here around embedding comments, interface, authentication, and user friction between two sites seem at least structurally helped by natively using one system for both publication and community/comments.

As an internal collaboration tool for a newsroom, Discourse is amazing. As a tool for healthy discussion, Discourse is amazing. But actually creating a distinct publication with content from Discourse is difficult and what I see as our biggest pain point.

I recognize it is not what Discourse is designed for. I’m sharing some thoughts mostly because I appreciate any thought of the news industry here.

Understanding that this isn’t at all attractive for CDCK, I would think this could be handled in improving the Wordpress or WooCommerce plugins, which post articles to Discourse and embed replies on Wordpress articles. I think @simon would have mentioned that if it was relevant.

That would be my preference. Specifically, two-way posting and improving the user-authentication pathways.

Staff onboarding seems like it’s not a Discourse problem, but it would have a greater chance of success if staff are using the same system for their CMS instead of having to use two separate systems. As mentioned, if staff are more engaged, then commenting users may be.

As a separate note, I think it’s relevant to point out that there are legal and resource concerns newsrooms have around user-generated content.

Many newsrooms are under-funded and the nature of their coverage (public, controversial, general interest) mean that discussions around it are more likely to tip into toxic areas. Staff moderation may not be able to keep up. I think the signal to noise ratio is different than what you see with most Discourse uses. It seems to me that healthy communities need to have some degree of exclusivity and the news media you have in mind are probably not exclusive in the topics they cover.

This is all to say that newsrooms may be reluctant to dive deep into any commenting system because of the legal risks of not getting it right. If I post an untrue comment on a newspaper’s article, it could be considered misinformation, while an untrue reply on meta might just seem amateurish. A lawyer would say that we should review and edit each comment before it’s actually posted. Basically: Write a Letter to The Atlantic - The Atlantic

In my opinion the comparisons to social media are fraught, but building community around journalism is something I am very keen on.

2 Likes

Interestingly, I’ve recently been wondering about reducing the scope of this project to making Discourse a viable commenting system for WordPress based news sites. With that approach, most of the technical objections that have been raised in this topic could be easily addresses - including adding the functionality to allow users to comment directly on the main article. The existing WordPress plugin could serve as a starting point for the project. Within the next few days I’ll update this topic with more details about how it could be implemented.

Do you have any idea if the legal concerns would be less if user generated content that contained potential misinformation was allowed to exist and be discussed in a private Discourse topic, but was not displayed in the public facing article’s comment section?

Yes I think so, but doesn’t that conflict with your proposal? Aren’t you looking to feature comments on the publication site, with the article?

I am not a lawyer, but my reference was largely to Section 230. Publishers understand that they could be sued for defamation if they edit UGC. If they don’t touch it, they aren’t liable. At the same time their core principles — informing large audiences, rigorous accuracy — discourage them from exposing any inaccurate information to their readers, even if they didn’t create it. This is why I say they are different from social media.

If a publication did want to have a “don’t touch it” approach to comments, I have to imagine hosting it on a separate domain would be the legal advice.

I don’t think the public vs private distinction is so relevant. It’s more about whether it’s third-party content, which is muddied if it’s featured as part of the publication.

I want to consider all options. My initial thinking was that conversations should occur on Discourse and that the “best” aspects of the conversations should be pushed to the main website. This would mean that moderators would be vetting comments. That might open them up to liability, but I think vetting comments is already a common practice for most mainstream news sites. (I’m largely using the site at cbc.ca as my example.)

Focusing on using Discourse comments on WordPress based sites opens up the option of allowing users to create a comment directly on the article’s page. I still have some questions about the best way to implement that. For example, should users be allowed to reply directly to another user’s comment directly on the article’s page, or should they be limited to commenting directly on the article from that page? By this I mean that only top level comments could be created from the article’s page. Users would be prevented from reply to or quoting another user’s post from that page.

Discourse’s trust level or groups system offers some possibilities for streamlining the vetting process for the main page’s comments. For example, users could be added to a “trusted users” group based on some criteria. Member of this group could be have their comments appear immediately on the main page. I’m getting ahead of myself though.

Any thoughts you have about this would be greatly appreciated.

My first thought is that those are all really good ideas.

You’re talking about promoting comments, as I understand it. Curating the comments displayed on the publication site seems like an appropriate and powerful tool for a moderator.

I’m not sure about the implications of nested replies versus direct replies. I suppose direct replies could potentially reduce the risk of arguments.

One possibility is that keeping it as direct replies would make it clear that the Discourse site is the place for discussion and interaction with other users. If their main interface is the curated comments, they might never see their comment appear and get frustrated.

With direct replies and curation, it starts looking more like a letters section.

If AI spam accounts can be used to earn trust levels, I would be wary.

1 Like