Half-baked proposals for architectural changes to Tildes groups and tags
This is a place to post your ideas about what to do about Tildes groups and tags. I'm going to write about some problems (as I see them) and save my ideas about solutions for the comments.
The taxonomy problem
We have tags and groups and they are somewhat arbitrary. A tag could be a group someday. A group can be downgraded to a tag if it's not used much.
Topics can have multiple tags, but they can be in only one group (and its ancestor groups).
It's hard to pick the right group. An example: a post about animals could go in ~enviro (for wildlife), ~hobbies (for pets), or ~science (for a scientific study). So where do you put news article about a scientific study of the effects of house cats on wildlife?
Adding ~animals seems like it would be a good thing because now you have an obvious place to find all the posts about animals. Animal lovers rejoice! But from a taxonomy point of view, it makes things worse, because now you have another place where you could logically put an article and another place to go looking for it. More groups means more edges and more edge cases. It's enough to make you wish for crossposts.
The competition problem
Tags are better for taxonomy, so why not just have tags? Because classifying topics isn't the only thing we want to do. As Deimos wrote about, eventually we'd like to have somewhat more independent communities, closer to subreddits but hopefully without their downsides. It would be nice if subreddits that wanted to migrate to Tildes could actually do it. We also want to have a good mix of topics on the front page, while allowing some groups to have a lot more posts than others.
I'll start with an analogy: if a school has only one sport that matters, the people who are good at that sport win socially, and other people don't have as much of a chance. But if you have multiple sports and clubs that people care about, there are more ways to win at something. I don't believe pretending everyone is a winner works all that well, but more ways to win promotes diversity and creates useful social ambiguity.
The front page of Tildes is the most visible and has limited space. That creates an all-against-all competition between topics. We also have groups with their own leaderboards, but they are lesser competitions and it's unclear if they matter yet. (I'm using them more, though.) Meanwhile, each topic has an independent leaderboard for its comments that doesn't conflict with any other game. (Maybe that's why I like megathreads?)
I haven't been thinking of Tildes in terms of leaderboards, but maybe it can explain why old-timers are often reluctant to post topics? We aren't really trying to win, but we have ideas about fair play. When there's only one game anyone cares about, we don't want to drown out other worthy topics by entering too many contestants. We're also a bit reluctant to enter anything that's too specialized into the competition, because it doesn't "deserve" the attention. It's not a worthy contestant and it's just gonna lose.
Also, sometimes this isn't a game you want to win. Entering a controversial topic into a competition can get you unwanted attention, and that's often no prize at all. When a game isn't one you wanted to enter, getting attention is more like losing than winning.
For the front page, I expect this problem will get worse with more people. Entering the competition brings more attention than before.
Note: thinking of a topic listing as a leaderboard for a game is only an analogy and I don't mean to promote competitiveness. They weren't designed to be leaderboards and I think we'd like to see design changes that reduce competitiveness. There are known downsides to competition that we don't want, like "cheating" to win with "unworthy" strategies and the rules-lawyering and jealousy that come with that.
Ideas?
Some rules for this "game": Please post one proposal per comment. If you have multiple independent ideas, you can post them separately, but post them together if they're interrelated.
Let me give a
most ludicrousapparently fairly reasonable suggestion:Every tag is a group. When someone posts something, it appears in every tag/group that was selected. In every tag/group anyone can vote for it to be a subgroup of any other group. (Barring some circular chains, maybe?) Whatever the current consensus is, goes. When the overall consensus changes, the waterfall of topics changes. Thus, the majority of people who care about groups and the specific sub-group's categorization enough to vote will always be happy.
And everything was solved forever, right?
So. I worked in a taxonomy system at an employer a few years ago with graph databases.
This is literally the avenue we took. It centralised, whilst governed the 32500 tags that we ended up with in a really fantastic way.
It also allowed for an engine to pick up on usability data (I.e. What a user goes towards) and then recommend them some additional tags to enjoy.
Also, what might be quite interesting? A geotag for each post. Then you could say "Give me all the PCgaming tags, but only for emea/uk/iran" as a query.
One of the downfalls of other deddit clones was people going in and making all kinds of sub communities which then had little to no action. You ended up with a huge stream of mod requests to take over the subs. I like the idea you have here because it allows people in a general topic to explore deeper themes and these smaller niche topics can attract people without having to constantly advertise.
I really hope this site remains about discussions and not just reposting twitter news. E.g. When the golden Knights won the Stanley Cup, I don't need a post about it. If I wanted that news, I have places to find it. It doesn't need to be here.
I don't mind specific news on topics being here. But not just screenshot of Tweets.
News should be about the event, then a transcription of said article in the comments for those who may not be able to access the account.
News should be shared for the discussion. And anything with lots of news at the same time is best as a mega thread. I agree with you.
Typo, or intentional? Either way I'm in favor.
Intentional. I figured why make it easy for them to find it
I'm a certified crazy person, and I've had thoughts in this vein before.
In essence, kill the idea of a X-post. One topic, one discussion.
Seems I’m not alone, wrote up something similar in 2018 https://tildes.net/~tildes/mz/tildes_subgroup_idea
I think there are good uses for separate threads in different places on the same topic, like when one group has different levels of knowledge about something than another. Say there's another big cybersecurity vulnerability like Log4J. ~news might have a big general conversation on the topic and its impacts. ~tech might have analysis of the technical details and action items for people effected.
I think that can still work with comment threads on one post. What I’ve noticed with cross posts/separate threads is that there’s a lot of overlap, which makes wading through to find the unique commentary the group brought to the table difficult, rereading a bunch similar comments in the process. If it was all in one place, the overlap isn’t duplicated. Plus, people might learn something outside of their expertise.
Would this essentially be the same as merging repeat topics/threads?
Vord hasn’t don’t said anything explicitly along that line AFAIK, but personally I think it’d be best not to do this for reposts unless they’re sufficiently close together time-wise. Sometimes an old topic is due for another go, and if you keep reviving the same topic, you end up with massive many-paged behemoths like the forum days of yore.
Sorry, I was unclear. Merging was mentioned earlier in the week as a solution/alternative for when the same topic is posted and being discussed in two different places.
However, after going through and rereading Vord’s post I understand it better. I think pictures would help.
I feel like I need a graph illustration of this. All tags exist in a DAG of sub-main relationships, with the edges being decided by vote. Ok. Is it a forest/tree, ore a complete DAG? I.e. can there be more than one more general term for a term? Could I put eSports into both sports and games?
Groups and tags are folded into one concept I believe?
I presume for the user who white/blacklists Tags they're interested in, it'd be annoying if the structure changes and certain topics fall off their subscription list or get joined in, due to others changing the graph and not that user's actions.
Oh I can do half-baked.
What if alongside groups we created user-created equivalents, which are just views for sets of tags? Let’s call them views. This is powerful because they can be highly individualized ways to browse tildes without exploding the number groups. Imagine ‘games’ + ‘long reads’ becomes ~serious_games in your sidebar. This is functionally identical to a group, but you can create it yourself without requiring it to be part of everyone’s Tildes experience.
You could even make these similar to Twitter lists and allow people to curate and share them with other users, building organic communities on top of the tagging system. A viewable tag cloud could be a helpful way to browse tag popularity and association when creating new views.
This is something that could replace groups in the long run if it got popular enough, but it could also live alongside them similar to how tags and groups already work today.
Now I'm envisioning a competition for finding "views" with the best selection of pre-existing tags that produce no hits, with a scavenger hunt section to find an appropriate discussion piece to bring back for everyone to talk about. Who can find the best Null Set?
Here's a sketch that entirely ignores feasibility and makes some radical changes. An evolutionary approach seems like a better idea than a "big bang" redesign, but I'll describe a possible end state.
Suppose we did taxonomy with tags only, and leaned into the metaphor of groups as leaderboards? Groups would no longer be a subject taxonomy. Instead, users can create leaderboards with quirky criteria for entry. There are probably still some generic groups with obvious names, though, and a distinction between official groups and user-created groups.
Topics can be in any number of groups, including zero. There is a "new topics" page where you can create a topic that could be added to any group. Each group has curators who can add any topics they think make sense into their group. Alternatively, a user can create a topic in a specific group, and then it can't be cross-posted without their permission.
For the front page, I would replace the list of topics with a list of groups. For each group, it shows the top two or three topics, and you can "enter" it to see more. Groups on the front page can be sorted by which ones have the most recent activity, or you can get a list of which ones have the newest topics. (We might also want to make sure the same topic doesn't show up twice if it's cross-posted.)
Upvoting a topic happens within a group, because each group has independent scores. But you can still do it from the front page because topics are only displayed within their groups.
That gets rid of the all-against-all competition between topics on the front page. It replaces it with a competition between groups, but I don't think people will care much about that?
You can ignore groups if you're not interested in them, or sticky them so they always appear at the top of the list.
How do we make groups more diverse? One way might be to have some groups with invite-only judge's panels, who are the only ones who can vote for topics in that group. The limiting case is a user-created group with one curator/judge, which is basically a playlist.
My thought on this is that Tildes is the community, and the community can have subinterests and little networks within the greater community, so empowering groups is trying to impose interests top down. Let the community decide how they want to categorize their interests and which networks they want to form, push those to the top, and as long as nothing burns down, roll with that. Tag Anarchy! (I said the thing!)
Maybe in the future we could try other instances of Tildes, some method of aggregation and cross-signin that isn't quite federation but allows for some different site norms or for the Tildes architecture to live outside of our BDFL, in the Deimos can't be everyone to everybody way, not the frozen peaches business. Maybe that's something that is never in the cards, but if the invite system and the code of conduct is already screening for people who want to land claim, the site trying to determine what the conversations are about is redundant, IMO.
It seems like the question is whether you have to allow every Tildes member to help curate every group, which imposes a sort of majority view whenever a group gets a lot of attention, versus having custom views that are only managed by their owners, and if you don’t like it you can make another group.
This isn’t really top-down versus bottom up, it’s majority versus minority.
A limiting case: on YouTube and Spotify, the playlists are curated by one person and I think that works well? Anyone can have their own jazz playlist or whatever. That’s a different kind of anarchy. Or maybe it would be better thought of as a property right?
I think this is different from managing a common taxonomy. The way we do tags now works pretty well.
Personalized views are neat, but kind of in the weeds compared to my viewpoints. "Tag Anarchy," as I envision it, is a counterpoint to "Group Hierarchy," where all modifiers are equal, and there are no privileged categories to decide where conversation is supposed to go, similar to how Tumblr works. We don't need to have the talk about "do we need this" or "how do we best present this" everytime someone proposes an ~identity top level, or if it's a demotion if we put a current top-level under a proposed catchall, and articles and discussions that involve intersectionality between top or sublevels don't have to decide what goes where. Everything is, there is more discourse and less meta-discussion about sorting and bias. The meta-discussion we would have is about trends, which is a little more actionable and more informative on a personal level. No groups, no groups within groups, all tags.
This was @mimicsquid ‘s very thorough explanation of why groups are vital for some conversations:
I think we're correlating a lack of groups implies that threads become echo chambers, which I'm not sure would become the case. Let alone the possibility that someone can start a new topic on their take of the situation or the specific event outside the context of the greater whole.
Anyway, maybe the construct of groups is necessary to provide structure for diverse and inclusive opinions, in which case I concede the point that we should have something on the front page for that, but I'm not sure hiarchial groups as we have them is the best way to catalog communities. I'm probably too engrained to flip on this particular point, but I've long accepted I'm fighting a for a lost cause.
How would you categorize Reddit’s subs? Did you like that better?
My issue with Tag Anarchy really isn’t against the idea itself, but that Hierarchical Groups is the way tildes is already set up. That’s the plan that’s in the docs, wouldn’t it be a lot more work to completely rework what’s already here than to build onto it?
Hierarchies are also very intuitive to many because that’s often how we’re taught things in school. We learned to identify colors (red, blue, purple, and green) and shapes (square, circle, rectangle, and trapezoid). In Science we learned about biology, chemistry, physics, and astronomy. We have hierarchical classifications for animals (domain, kingdom, phylum, class, order, family, genus, and species). You’ve probably heard that argument before tho.
Things get more difficult when there is more than one relevant hierarchy. For example, a discussion about an animal could be classified by both its species and its geographical location.
Fortunately, Tildes has hierarchical tags so we can do both. Tags are quite nice for taxonomy.
That's a good case for having multiple discussions about the same event, with distinct audiences. But it remains to be explained why there should be multiple groups? It's possible in theory that if there were topics with different titles about the same event, people would pick the one that appeals to them and ignore the others.
It doesn't seem too likely though. Something more is needed. How do we divide up a fairly unified community into separate communities? It's kind of an odd goal.
It's a common assumption that groups will naturally create distinct audiences, like happened with subreddits. Each audience sees the topics in its own group. It doesn't seem to be happening automatically on Tildes, though.
It can’t happen automatically in Tildes because there’s no function to enable the creation of new groups, beyond a tag reaching a certain level of popularity. Even Deimos has admitted this system has stopped him from creating gaming related posts he may otherwise have made if there was a specific space for it, because he didn’t want to flood the main page with gaming news.
A petitioning system for creating groups sounds nice in theory. It has a nice parallel to the invite-only accounts, it would stop repeat groups from being created, it could keep the number of groups down, and the topics from becoming too niche (unless you show an established interest). My worry would be that it would too much for the site admins/Deimos to handle.
Creators of groups could have mod-like powers for a limited period of time (30-90 days?) to establish group specific rules, automated posts, and whatever else and then the site user-moderation model would take over. Obviously all groups would have to remain discussion based and follow all Tildes guideline, regardless of any group rules anyone may choose to implement.
Is this a bad idea? Is it too much like reddit?
I'm ignoring practicality for this discussion and I'm more interested in philosophical issues, like what are groups for, and how well do they serve that purpose?
The details about how to manage groups seem downstream from that.
Ah. That’s what I didn’t understand about this post. Why is there a discussion about reorganizing, tho? Groups haven’t worked/ really become a thing on tildes because the community was too small to support being fractured (from what I’ve read). With only approximately 100 regular posters any division would have had a huge impact.
Groups are a place for users with similar interests, hobbies, beliefs, and/or experiences to gather to discuss those interests, hobbies, beliefs and/or experiences. They keep upsetting discussions out of view of anyone who might not be able handle them (certain support groups can openly discuss things that may be uncomfortable, harmful or triggering in other situations). They help keep minority voices from being drowned out or otherwise downplayed. They help users curate their content to what best suits them.
I feel like because tildes hasn’t really used groups yet, in a meaningful way, it’s impossible to tell how well they serve their purpose. We can’t judge based on Reddit because power hungry mods and sub drama were so prevalent. I really think the best thing to do is come up with a game plan for creating groups on tildes and see how they work, what problems arise.
I'm thinking about it more conceptually, but it's fair to say that groups haven't really been tried.
I don't believe groups actually do much to separate Tildes into separate communities, due to how the front page works. I'm using "ignore topic" a lot more these days, but I still like to see what's out there.
But you just agreed that groups haven’t been tried on tildes, so we can’t say they wouldn’t separate it out into communities.
There was actually a really interesting idea over on the ~tildes.official post from the other day.
That’s the gist of it, but if you want to read more go here and read under the second parent comment. Amarok is the poster.
Well, groups aren't entirely untried since we have them and we've seen what happened so far. But we haven't seen what happens with more scale; that part is imaginative.
This whole discussion is an imaginative exercise. That's inevitable, since you need to use your imagination when thinking about what might happen in Tilde's future. Nobody's suggested changes have been tried yet.
It's true that doing nothing is less work, so there's a status-quo bias.
Copy the lobste.rs tagging system and UI. Maybe it's just the background-color that's specific to each tag, but the tagging system in lobste.rs is a joy to use and very straightforward, whereas I'm still not clear how tildes tags interface with subgroups.
(I wouldn't have posted this but you specifically mentioned "half-baked".)
On Tildes, tags are specific to a top level group. So if you're searching the "metal" tag in ~music, you won't get stuff tagged with "metal" from ~science
Also it's more likely to be
metalsin ~science.That's true of searches when you click on a tag while you're in a group. (You can see in the URL that the group name and tag are included in the search.) But when you click on a tag on the front page, it shows topics with that tag from any group.
I really like the nested system, though havent dug into it, I'm brand new. I think it would be really good if the system used the sidebar to split out the separate niches. For an example I was very into HFY on reddit a pretty niche subreddit and would love to continue something similar here my idea for this is is something like literature~writing~fiction~hfy
i.e there is a top level literature group, on its sidebar it shows the sub groups one of which is writing, on its sidebar more sub groups, and so on.
This would let you cut up the overall subject as niche as you could like but still retain an easy way to find each group. I would also nest the 'all' pages so that say the top three current posts in each subgroup are shown in the next layer up and the top three from there in the next and so on and so on. This would ensure that more niche subgroups still get a chance to appear in the next layer up bigger group. The reddit way of doing it based solely on upvotes means the same crap gets recycled again and again but having a forced contribution gives new stuff a chance to be seen.
I also like opt out rather than opt in since it gives you a chance to see something is available that you might not think to search for. HFY needed to be deliberately sought out and I think it suffered from that a little.
It sounds like you're a fan of hierarchical groups and want there to be more of them? Along with some changes to how a parent group displays topics from its children.
I suspect you may be right that flattening all the descendant topics into a single list ordered by activity might have some problems with getting a variety of topics in a parent group. However, we don't have a whole lot of experience with hierarchical groups yet.
Yeah I guess thats the right way to say that, I suppose its just the way that folders in windows work if I really think about it.