Washington Post integrates Talk – Mozilla’s open-source commenting platform(blog.mozilla.org)
blog.mozilla.org
Washington Post integrates Talk – Mozilla’s open-source commenting platform
https://blog.mozilla.org/blog/2017/09/06/mozilla-washington-post-reinventing-online-comments/
334 comments
Can't speak for WaPo but the New York Times comments sections are absolutely dominated by a handful of designated 'top commenters' who rush to post on any major story.
Sometimes their comments are good, but then there's the guy who writes a poem every single time, or 'Socrates' who posts the same cookie-cutter 'blame-the-GOP-all-the-way-back-to-bush' comment with the most dramatic rhetoric possible, trolling for likes of course.
Having said that, I'm pretty satisfied with the Times comment system, everyone does get a voice even though the top comments are absolutely fixed.
Sometimes their comments are good, but then there's the guy who writes a poem every single time, or 'Socrates' who posts the same cookie-cutter 'blame-the-GOP-all-the-way-back-to-bush' comment with the most dramatic rhetoric possible, trolling for likes of course.
Having said that, I'm pretty satisfied with the Times comment system, everyone does get a voice even though the top comments are absolutely fixed.
After reading this article over and over I am still clueless.
What is this all about? Feature-wise it doesn't sound better than a comment system from HN, Reddit or StackExchange.
Besides, Mozilla is really lacking communicational skills. This blog post feels like the documentation at MDN, tedious and unclear.
What is this all about? Feature-wise it doesn't sound better than a comment system from HN, Reddit or StackExchange.
Besides, Mozilla is really lacking communicational skills. This blog post feels like the documentation at MDN, tedious and unclear.
As a DC-area resident and frequent reader of the Washington Post, the biggest problem with their comments section is the commenters. I'm not sure how this helps.
I would love to see a big site try to use Slashdot's old commenting system. Commenting felt like a privilege and so did moderation.
Here's a slightly off-topic dream about online comments:
In an ideal comment system I believe that articles, comments and moderation events should come from three different, decentralized streams (like Atom) that the end user can subscribe to individually and that are joined at the end users client. That would would provide transparency to the moderation process, ability to comment anywhere, and it would allow moderators to become effective spam-filters without giving them the power of censorship. Now, imagine if this system was built into the browser and it became the default commenting platform for all websites...
In an ideal comment system I believe that articles, comments and moderation events should come from three different, decentralized streams (like Atom) that the end user can subscribe to individually and that are joined at the end users client. That would would provide transparency to the moderation process, ability to comment anywhere, and it would allow moderators to become effective spam-filters without giving them the power of censorship. Now, imagine if this system was built into the browser and it became the default commenting platform for all websites...
Mozilla Talk could be useful if it let me apply my own moderation schema (filters, blocklists, etc). It looks like another totally centralized moderation tool:
https://coralproject.net/products/talk.html
The only explanation of their approach is that they did "an enormous amount of research". This sounds about as convincing as someone claiming "Oh, I took a class in that."
https://coralproject.net/products/talk.html
The only explanation of their approach is that they did "an enormous amount of research". This sounds about as convincing as someone claiming "Oh, I took a class in that."
Let me distract you from the actual content of the article first (I am sorry), because I am a little frustrated (I usually don't, and always almost give credits to the awesome works done by Mozilla; disclaimer: used to be a Mozilla intern).
This article is too shy from actually showcase the product. Half of the article is devoted to PR announcement (which IMO is really a waste of time for most readers, even for the non-technical ones). Just first paragraph into the article, I had already given up reading through, but instead put my effort in spotting keywords to understand the motivations behind this new system, instead of jargons.
The second half is a brief overview of the new system, but only two lines are relevant to what I was looking for.
I suggest next time the writer should:
* tell readers the challenges/flaws in existing system concretely and perhaps showing examples
* the finding from their research participants, the hypothesis ,and how they conducted their experiments
* explain concretely how the proposed system will help address the prominent flaws and issues
This perhaps should turn into a research paper, but even as a blog post it should have more substances. At best, this article feels like a company quarterly performance announcement, which I just scan for keywords and move on, then I forget about all of it. So, what is the point of reading this anyway? README would have been far more interesting. I am disappointed.
----
Anyway, I don't really read comments on news sites. I can't explain why exactly but I feel disconnected and buried (much like what @creagpatr said about fixed commentary). I'd rather go on FB and comment there, even though there is a fairly good chance no one can read mine. Perhaps the social part of FB feels more lively and often comments are funny as hell. Yes, I know that comments on FB are just as echo chamber as many other places do.
There is a site allows people to take side and debate to death. I forgot the name though...
This article is too shy from actually showcase the product. Half of the article is devoted to PR announcement (which IMO is really a waste of time for most readers, even for the non-technical ones). Just first paragraph into the article, I had already given up reading through, but instead put my effort in spotting keywords to understand the motivations behind this new system, instead of jargons.
The second half is a brief overview of the new system, but only two lines are relevant to what I was looking for.
It’s filled with features that improve interactions, including functions that show the best comments first,
ignore specific users, find great commenters, give badges to staff members, filter out unreliable flaggers,
and offer a range of audience reactions.
I suggest next time the writer should:
* tell readers the challenges/flaws in existing system concretely and perhaps showing examples
* the finding from their research participants, the hypothesis ,and how they conducted their experiments
* explain concretely how the proposed system will help address the prominent flaws and issues
This perhaps should turn into a research paper, but even as a blog post it should have more substances. At best, this article feels like a company quarterly performance announcement, which I just scan for keywords and move on, then I forget about all of it. So, what is the point of reading this anyway? README would have been far more interesting. I am disappointed.
----
Anyway, I don't really read comments on news sites. I can't explain why exactly but I feel disconnected and buried (much like what @creagpatr said about fixed commentary). I'd rather go on FB and comment there, even though there is a fairly good chance no one can read mine. Perhaps the social part of FB feels more lively and often comments are funny as hell. Yes, I know that comments on FB are just as echo chamber as many other places do.
There is a site allows people to take side and debate to death. I forgot the name though...
>New comments and reactions update instantaneously
Killer feature - in that its going to kill the process.
This is chat. Or close enough to chat that its going to be used as chat.
Chat is terrible for comments. A flame war on chat escalates at the rate at which someone can hit enter.
like the wake of a ship, the comment area will expand, and defeat moderation ability to clean it up.
Flame wars are bad enough, flame wars on chat which persist like comments are a bad idea.
Make it update in non-real time.
Killer feature - in that its going to kill the process.
This is chat. Or close enough to chat that its going to be used as chat.
Chat is terrible for comments. A flame war on chat escalates at the rate at which someone can hit enter.
like the wake of a ship, the comment area will expand, and defeat moderation ability to clean it up.
Flame wars are bad enough, flame wars on chat which persist like comments are a bad idea.
Make it update in non-real time.
Can a client make a request to fetch all the things from a specific comment thread?
Example: I go to a mailing list for pick-your-favorite-old-grumpy free software project. I click through the ancient, craggy mailman web interface. Hey, there's a button that says, "download the full archive (x MB)." Now I have all the things.
Real World Use Case: The Guardian once deleted some comments on a Greenwald op-ed because of insert-old-craggy-British-law-here. Archive access would have empowered users to periodically archive the comments so that they have control over content they want to read, while the publisher is still able to remove comments from flow and archive when forced to censor.
Example: I go to a mailing list for pick-your-favorite-old-grumpy free software project. I click through the ancient, craggy mailman web interface. Hey, there's a button that says, "download the full archive (x MB)." Now I have all the things.
Real World Use Case: The Guardian once deleted some comments on a Greenwald op-ed because of insert-old-craggy-British-law-here. Archive access would have empowered users to periodically archive the comments so that they have control over content they want to read, while the publisher is still able to remove comments from flow and archive when forced to censor.
If anyone was looking for it, the source is here: https://github.com/coralproject/talk
Why not use machine learning and a sort of "reputation system" to filter out the noise and trolls? The system could analyse comments, but the users could also, instead of upvote/downvote buttons, have buttons with reactions such as "Interesting", "Insightful", "Not relevant", etc...
Some people have informed opinion/interesting things to say and some people just contribute noise to the discussion. The system can learn both from the language and reactions from people.
Just like in "real life", if someone is always talking out their ass, you learn to listen to them less and less because they don't have anything meaningful to say. It's just noise.
The comment system should do the same. It learns from multiple comments and their reactions to determine a reputation. Obviously, if your comments are always seen as "not-relevant" and "toxic", then your contribution is worth less than other "insightful" and "interesting" contributions, so past a certain threshold, you stop being able to comment.
Some people have informed opinion/interesting things to say and some people just contribute noise to the discussion. The system can learn both from the language and reactions from people.
Just like in "real life", if someone is always talking out their ass, you learn to listen to them less and less because they don't have anything meaningful to say. It's just noise.
The comment system should do the same. It learns from multiple comments and their reactions to determine a reputation. Obviously, if your comments are always seen as "not-relevant" and "toxic", then your contribution is worth less than other "insightful" and "interesting" contributions, so past a certain threshold, you stop being able to comment.
Doesn't work if you have Ghostery turned on. The Washington Post has 15 trackers, and the comment thing has a cross-site scripting request.
This thing writes many messages to the browser console. The last line apparently prevents even displaying comments.
This thing writes many messages to the browser console. The last line apparently prevents even displaying comments.
STARTING Paywall c4d7fa1f35.js:41:468
PAYWALL CHECK STATUS: c4d7fa1f35.js:42:331
"PWAPI URL: https://pwapi.washingtonpost.com/news/josh-rogin/wp/2017/09/12/state-department-official-quietly-visits-moscow-to-discuss-north-korea/?rplapisplit=1&referrer=https://www.washingtonpost.com/" c4d7fa1f35.js:42:405
PAYWALL ACTION: 0 c4d7fa1f35.js:43:138
PAYWALL URL: N/A c4d7fa1f35.js:43:218
PAYWALL SUB: Unsubscribed c4d7fa1f35.js:43:292
PAYWALL METERED: Metered c4d7fa1f35.js:43:393
PAYWALL METER COUNT: 4 c4d7fa1f35.js:43:498
PAYWALL MODE: 1 c4d7fa1f35.js:44:49
PB Debug: Fired comment_count_target.init:[object Object] c94e09db1d.js:381:443
boomerang.nt: [info] This user agent supports NavigationTiming. wprum.min.js:1:10101
pt.fetchStart: 1505251033854 wprum.min.js:1:10803
boomerang: [debug] Checking if we can send beacon wprum.min.js:1:10101
boomerang: [debug] Sending url: //rumds.wpdigital.net/metrics/?rum=%7B%22redt%22%3A0%2C%22nl%22%3A285%2C%22dns%22%3A0%2C%22tcp%22%3A0%2C%22rest%22%3A195%2C%22domt%22%3A904%2C%22uplt%22%3A3052%2C%22url%22%3A%22https%3A%2F%2Fwww.washingtonpost.com%2Fnews%2Fjosh-rogin%2Fwp%2F2017%2F09%2F12%2Fstate-department-official-quietly-visits-moscow-to-discuss-north-korea%2F%3Futm_term%3D.b5cb34fc201d%22%2C%22site%22%3A%22www.washingtonpost.com%22%2C%22title%22%3A%22State%20Department%20official%20quietly%20visits%20Moscow%20to%20discuss%20North%20Korea%20-%20The%20Washington%20Post%22%2C%22domn%22%3A940%2C%22section%22%3A%22opinions%22%2C%22cntt%22%3A%22blog%22%7D
v=0.9
u=https%3A%2F%2Fwww.washingtonpost.com%2Fnews%2Fjosh-rogin%2Fwp%2F2017%2F09%2F12%2Fstate-department-official-quietly-visits-moscow-to-discuss-north-korea%2F%3Futm_term%3D.b5cb34fc201d
if= wprum.min.js:1:10101
PB Debug: Fired sticky-vertical-sharebar.collides_with_main_content:false c94e09db1d.js:381:443
PB Debug: Fired nav.start_close:true c94e09db1d.js:381:443
PB Debug: Fired nav.finish_close:true c94e09db1d.js:381:443
Comments load: timer startedtwp-comments.js:1
[Echo.UserSession] error : Unable to initialize user session, error response from IdentityServer API received | args: Object { response: Object } 499c2b4a02.js:294:297How do users authenticate? Do they need to register for an account and give their email address to every website they want to comment on?
This seems like it would have been a good use-case for Persona, but that project was cancelled; is there any good replacement these days, besides the ubiquitous "sign in with your facebook account"?
This seems like it would have been a good use-case for Persona, but that project was cancelled; is there any good replacement these days, besides the ubiquitous "sign in with your facebook account"?
Comment systems are tricky, everyone has an opinion on how they should work and how they should be moderated. Its very similar to forums and community building.
I was so opinionated about my comment system I decided to build my own called Remarkbox (https://www.remarkbox.com)
Its embed-able like discus but I want it to feel more like hacker news or reddit (deeply nested threads). My system is still a work in progress but I'm actively dog fooding it on my own websites and so far its working well.
One major take away I've learned from this thread is that people really like to use the collapse feature here on hacker news and reddit to narrow in on the conversations they care about. I actually never used this feature on either site (I didn't even know what that [-] was on hacker news until today)
I was so opinionated about my comment system I decided to build my own called Remarkbox (https://www.remarkbox.com)
Its embed-able like discus but I want it to feel more like hacker news or reddit (deeply nested threads). My system is still a work in progress but I'm actively dog fooding it on my own websites and so far its working well.
One major take away I've learned from this thread is that people really like to use the collapse feature here on hacker news and reddit to narrow in on the conversations they care about. I actually never used this feature on either site (I didn't even know what that [-] was on hacker news until today)
I've always been frustrated by the Washington Post comment system. Many articles have hundreds, perhaps more than a thousand comments. When I revisit an article, I can sort by newest/oldest/highest replies or something like that. I can go to my user profile, select "my comments" and view my recent comments (though not in the context of the article they came from). For the comments I left where someone replied to me, I have no way to reply to them, other than to return to the article and hit "show more comments" 100 times and locate my comment and the reply so I can reply in turn.
I hope the new system fixes that. It is otherwise about impossible to hold a conversation, so it reduces the incentive to leave a comment in the first place.
I hope the new system fixes that. It is otherwise about impossible to hold a conversation, so it reduces the incentive to leave a comment in the first place.
> Talk is small — about 300kb — and lightweight
This makes me feel old. Is 300kb really considered small these days?
This makes me feel old. Is 300kb really considered small these days?
Shouldn't they be focusing on the browser?
There is literally a good hundred of solid solutions out there for web commenting - hosted, embedded or app libraries -, and Talk doesn't seem innovative at all.
There is literally a good hundred of solid solutions out there for web commenting - hosted, embedded or app libraries -, and Talk doesn't seem innovative at all.
I think to have good comments, you need to build a good community. I sometimes see the same article shared across HN, Reddit and Facebook. Some sites have better comments than others..
I remember a time when commenting seemed like a weird idea. Why would someone write on your site instead of on their own site? And then dynamically generated pages and eventually social networks took over. I wonder if the internet would be a better place if everyone had their own websites and RSS/Atom feeds rather than the centralized services we have today.
Wouldn't the easiest approach to this whole "issue" be to give every user access to individualized blacklisting tools?
Just put a "This comment makes me angry" button next to each one and people who feel offended can click it to hide said comment and check a box to hide any future comments from that commenter.
There, problem solved with no need for fancy machine learning algorithms or other institutions to play big brother for the internet commenting culture. Voting based systems generally seem to do a pretty decent job most of the time, especially when the community can be segregated by topics/interests like on Reddit.
There, problem solved with no need for fancy machine learning algorithms or other institutions to play big brother for the internet commenting culture. Voting based systems generally seem to do a pretty decent job most of the time, especially when the community can be segregated by topics/interests like on Reddit.
I love Mozilla's efforts and I think this is an important area to improve.
But, as petty and selfish as it sounds, I wish it were built in my development language of choice. Then I could modify and use it myself, or contribute, without that irrational dirty feeling.
I know, node is very popular, and no disrespect to you if you use it. It's a personal failing. I just can't put my personal time into it any more. When I see a great project I get excited, and search out the source to see what it's written in. Then I often feel disappointed. Maybe I need to gain some flexibility.
But, as petty and selfish as it sounds, I wish it were built in my development language of choice. Then I could modify and use it myself, or contribute, without that irrational dirty feeling.
I know, node is very popular, and no disrespect to you if you use it. It's a personal failing. I just can't put my personal time into it any more. When I see a great project I get excited, and search out the source to see what it's written in. Then I often feel disappointed. Maybe I need to gain some flexibility.
Are there any notable new techniques used here, whether social or technical?
Comments just aren't a very good approach if the point isn't to just pass time. Annotations are much better. Systems like hypotheth.is are slowly becoming more pervasive.
Back in the days I thought that what made the Huffington Post popular, politics aside, was its gamified commenting system which I though was excellent. Since then of course, it degraded UX wise until it was replaced by Facebook comments(horrible UX) and something completely useless today. At the end of the day what makes the difference between a good community and a bad one is tight moderation without censorship and focus.
I think it would be interesting to see a filter that generates something along the lines of Amazon.com's Statistically Improbable Phrases, or "SIPs", the most distinctive phrases in the text of books, but for articles, and then tracks how closely the comments line up for that particular article .. might be a good indicator for authority ?
I believe it's better to have discussion happen elsewhere, as in HN, Reddit, Usenet, &c., or in blogs and columns. This is cheaper for the publisher, and often those communities generate better discussion as they are comprised of more interested individuals with more genuine participation.
Off topic, but this is the first time I see two slightly different sandwich menus on the same page
The top of HN for an article which says nothing much at all, really?
> It’s fast. Talk is small — about 300kb
Hopefully that's not the client size
Hopefully that's not the client size
I know I'm super biased, but I still find Reddit comments the easiest to read and follow (HN is a pretty close second, especially after they added comment collapsing).
Is it really just bad marketing it on Reddit's part that keeps others from using their embeddable comments? As a site owner, you can fully moderate the conversation by only embedding comments from a subreddit you fully control. You may not get quite as much data as Mozilla's solution, but in exchange you get the benefit of Reddit's 12 years of spam data that train its filters.
I just don't understand why more sites don't embed the reddit comments.