Where do you find good online discussion these days?
-
mariusreplied to Doc Edward Morbius ⭕ on last edited by
@dredmorbius I'm about to get some free time to work on my Fediverse link aggregator and discussion platform[1].
I'd welcome to have you involved if you want to do some product management (or any other type of involvement, really) for a platform that you'd like, and eventually bootstrap a community on top of it.
Do you think you can articulate the elements that you think would make a platform appealing to qualitative discussion[2]?
[1] https://git.sr.ht/~mariusor/brutalinks
[2] This of course applies to anyone down thread of this comment. -
Doc Edward Morbius ⭕replied to marius on last edited by
@mariusor Off the top of my head:
The basic problem is being the kind of forum where people who want to and can make significant contributions will. Which means not pissing them off / annoying them / frustrating them, and making it easy for them to shine. Without scaring off absolutely everyone else, or becoming yet another Internet cesspit (which would of course piss off, annoy / frustrate your target cohort).
Search. Finding relevant content, and especially finding MY OWN EARLIER CONTENT is an absolute must have. I'm aware that search can also generate some unpleasant effects (brigading, harassment, etc.), but on balance I think it's necessary. HN's Algolia search (and some recent suggestions for enhancements above that) is a key strength of that platform. It's also a key reason I've abandoned Diaspora*: over a decade on it still lacks this. And yeah, I've gone to insane lengths to make up for this (that's a manually-curated set of significant posts).
Exposure to General Web Search. I'm ... less enamoured of this than I was (No I Do Not Want To Feed Your AI Dragon), but discoverability remains key. Another death-knell for Diaspora* is that the site is nearly/entirely invisible to general Web search (Google, Bing, DDG, etc.). Winking in the dark...
Markdown. Formatting content is a huge win, and of the various options (raw HTML, Markdown, other LWMLs), Markdown is both generally sufficient and most widely adopted. It makes for the difference between primitive wall-o-text posts and well-structured content. There are a few additions to the basic feature set I'm fond of, most especially endnotes. Diaspora* does well here, see for example https://diaspora.glasswings.com/posts/60cf1410e72f013a68ef448a5b29e257. (This toot itself makes heavy use of Markdown, part of GlitchSoc's featureset. View it from toot.cat for the full effect.)
Editing. I'm not perfect, and make all kinds of errors in posts, toots, comments, etc. Let me edit my content. Yeah, there's abuse issues, so note edits and maybe show edit histories (for non-critical disclosure bits at any rate.)
Abuse defences. Past a certain critical mass, discussion sites live or die on how well they can address abuse. This includes harassment, brigading, shilling, spam, propaganda, bots, various vice content (CSAM, drugs/guns/contraband, criminal and gang activity, village and/or Internet idiots, etc.) Sites ultimately require multiple levels of defence, you cannot presume that system-level/centralised tools will suffice, or that putting everything in the hands of end-users works. Both are needed. Oh, and the anti-abuse system itself, if sufficient to combat abuse, becomes a mechanism for transacting abuse itself (censorship, content flagging, account banning/loss, etc.). There's a fine balance between too little, just right, and too much moderation, as well as how it's performed.
Content moderation. Necessary, but difficult to do well, and It's Complicated!!! Non-brief thoughts: https://web.archive.org/web/20200629055317/https://www.reddit.com/r/dredmorbius/comments/28jfk4/content_rating_moderation_and_ranking_systems/#. I'm increasingly convinced that the ability to block individuals en masse is also essential, though with varying forms. (Mastodon ... actually does this fairly well.) Some form of moderation delegation may also be required, as well as the ability to lock (manually or automatically / timed) threads to prevent late fly-by spamming / trolling. That is, I could name several close contacts as approved moderators on my posts. Doing that also requires a moderation log for things to remain sane.
Groups. I think that these are useful, though complicated. HN is interesting in insisting on no groups, everything is posted to one flat page. Lobste.rs is interesting in that it provides tags, similar for Tildes.net. Back on G+, I found groups mostly useless, save for some smaller private ones I'd created. What I long wanted there was the ability to use the intersection of a Topic and some sort of member list ("Circle") to filter through discussions. Ping me for more thoughts on this. One key concept is that there are different kinds of groups, and those which are really aimed at conversation DO NOT SCALE. Somewhere between, say, 5--50 people is really the sweet spot. You might be able to go somewhat above that (more registereds, but a small fraction active). Somewhere in the 100--1,000 member space shit goes all to hell.
Sane discussion presentation. Again, Diaspora* does nicely here, see again https://web.archive.org/web/20200629055317/https://www.reddit.com/r/dredmorbius/comments/28jfk4/content_rating_moderation_and_ranking_systems/# and note the side-by-side placement of article and comments.
Notifications. I don't have a current example of this,b ut at various points in their evolution, Google+, Ello, and Imzy (remember that?) all had a really useful Notifications pane which made it possible to interact with comments on posts / replies to threads / mentions, without having to leave the notifications pane itself. Thinking about it, Mastodon's own presentation in this particular regard is relatively good, though Mastodon lacks the notion of a post/comments thread hierarchy. Another key element was that all participants in a given thread (up to pretty generous limits) were notified of any additional activity, unless they muted the thread. This meant that individual discussions could continue for days, weeks, months, and even years, though this required robust moderation by the discussion host.
The early/late dichotomy of Problems With Online Forums. Early on, you're simply Trying To Let It Not Die. Growth or even simply sustaining activity is the challenge. 99.9999967% of all new forum attempts die in this stage, it's simply the consequence of winner-take-all dynamics. Once a forum gains scale the problem switches often more or less instantly to not getting drowned in shit. The transition can be intensely rapid, but at the same time trying to solve problems of scale before you have scale is pretty fatal. This was, I think, one of Imzy's problems.
Addressing a range of discussion levels/interests. I prefer finding discussion that's illuminating and insightful. Many people just want to be entertained, troll the normies, or whatevs. You've either got to define your use-case / target audience, or design a system which addresses multiple needs. Both are challenging.
Viable founding cohort. The initial class of forum members sets tone, culture, norms, and a whole lot else, and is absolutely critical to wider success. Usenet and Facebook both started with highly-selective university students, and that was a key to both their initial successes. Both stumbled badly as they grew to general population platforms, for multiple reasons. I've written on Why Usenet Died, see: https://web.archive.org/web/20240000000000*/https://old.reddit.com/r/dredmorbius/comments/3c3xyu/why_usenet_died/. (I may be wrong on some or all of that, though I think there are at least some valid points, it recapitulates some of the themes in this toot.)
-
Doc Edward Morbius ⭕replied to Doc Edward Morbius ⭕ on last edited by
-
mariusreplied to Doc Edward Morbius ⭕ on last edited by
@dredmorbius thank you for the reply, I'll try to put your words into actionable items for me to work on, but some of your suggestions are paradigmatically opposite to what I am trying to achieve with BrutaLinks. I think the stronger two limitations of my design are: the service needs to be quite simple to install and operate, (so it fosters many small communities), and the interactions between users and service need to be possible using the vanilla ActivityPub vocabulary. So, some of your suggestions might require additional tooling on top of a simple main service (think something like automod), or be impossible to implement all together.
-
Doc Edward Morbius ⭕replied to marius on last edited by
@mariusor So ... yeah, I was thinking in terms of a centralised platform. Distributed is a whole 'nother ball of wax tadpoles....
I'd still stick with search, Markdown, and classification tools.
With distributed systems, discovery becomes an even bigger challenge. Note too that the idea that the average person, or even organisation (company, academic, NGO, governmental office) will install / run their own software is increasingly foreign. For better or worse, SaaS has eaten the world, and for most (people, orgs, etc.) spinning up a FB page, Subreddit, Instagram, Discord, etc., is the path of least resistance for both the org and its public.
(I'm not saying I agree with this or think it's a Good Thing. I AM saying that It Is A Thing.)
The idea behind Diaspora* was that people would run their own pods. Few have, though yes, it is distributed. I'll note that Diaspora* absolutely fails at discovery, even on platform, but most especially for the 99.999875% of all people who don't have a Diaspora* account.
I'd thought of asking you what you had in mind before writing my previous rant. If it's not too late, I'll ask now. The project page is vague in that regard.
-
mariusreplied to Doc Edward Morbius ⭕ on last edited by
@dredmorbius the idea behind brutalinks was less as a multi-purpose discussion platform, but a modern recreation of a technical discussion group or mailing list. Sadly I don't have an elevator pitch better than the README of the github repository, or the slightly expanded wiki page on sourcehut: https://man.sr.ht/~mariusor/go-activitypub/brutalinks/index.md
Until now nobody was interested enough to ask questions, so I don't have my thoughts ordered in a better way. If you go through this, I'd appreciate any questions or feedback.
-
@dredmorbius
I think one thing that doesn't appear in the wiki or readme, is the fact that ideally the individual brutalinks instances should operate without the need for users to create local accounts. This is still e nebulous technical issue in my mind, so I don't have a good solution at hand.Basically you would be able to "participate" in the discussions on a brutalinks instance with your existing ActivityPub account (be it mastodon or something else).
The instance would be the convenience gathering point for the discussion, but it would exist in your regular inbox too.
-
@dredmorbius
I'm not sure how feasible this dream is in the context of the attitude of mastodon's devs towards other ActivityPub implementations, or even making their own spec compatible, but maybe some day. -
Woozle Hypertwinreplied to Doc Edward Morbius ⭕ on last edited by
I have thoughts also, but no time left to formulate them this evening
-
@woozle feel free to drop a reply any time you organize them.
-
Some relatively-quick thoughts, to avoid the "get everything in there" mental wall -- more addressed to @mariusor than to @dredmorbius, I think (he's already aware of most/all of this).
Link-collection is something in which I have a keen interest. I kluged together a system for my MediaWiki-based political site (issuepedia.org), but it's ugly and slow and writing my own data-driven CMS is now on the project-pile. I'll be interested to see what you come up with, especially if its data can be integrated into other projects somehow.
I've been thinking a lot about what software can do towards maximizing reasonability in political discussions and minimizing bad-faith participation, and... well, basically this and this are what I've come up with so far. (Note that most of the writing there is from 2016 or so.)
Also, fwiw, I've been running/moderating this here fedi instance since 2017, so I've got some experience in finding what works (and Opinions about what doesn't).
Oh, and I should probably also mention these argumentation guidelines that I've come up with for Issuepedia over the years (mainly 2010-2022ish).
-
@woozle thank you. The thoughts about process more than technology are much appreciated.