Twitterlike is a Bad Shape

[osteophage]

Twitter and its imitators have adopted a structural design that is fundamentally bad for people. This isn't just a matter of who's in charge; it's a problem with the thing itself. Forcing users to adhere to a tight character limit, discouraging link culture, preventing people from editing their own posts, steering people into sharing things they hate, incentivizing rage bait with trending feeds, subjecting people to decontextualized encounters, encouraging conflict by discouraging tags, and leaving users powerless to clean up the resulting mess—all of this is bad shape.

Crossposted to Pillowfort and Neocities. For off-site linking, I recommend using the version on Neocities.

Bad shape is a metaphor I'm borrowing from Erin Kissane, who has previously used it to describe unmanageably-huge sites like Facebook via analogy to the game of Go:

In the game of Go, bad shape is the term for configurations of stones on the game board that are inefficient in achieving their offensive goal (territory capture) and unlikely to achieve their defensive goal (the state of "life"). You can extend a bad shape in a fruitless attempt to make it better, but you'll generally be wasting your time.

The idea I keep coming back to is that the big platforms, like Marley, were dead to begin with, and are now something particularly bad, which is dead on their feet. Not because they’re been abandoned by users (yet) but because they’re structurally incapable of governing the systems they made, and most of the things they try to do about it introduce more and weirder problems.

Erin Kissane uses the metaphor of bad shape to describe how these megacorps end up structurally incapable of good governance. The aims of continuous growth, market dominance, and profit extraction ends up conflicting with the needs and wellbeing of the users, and this kind of problem is not fixable by swapping out who's in charge. The structure itself is unworkable. It is a bad shape.

The same can be said about the distinguishing features of Twitter—many of which, unfortunately, have been copied by imitators like Bluesky and Mastodon.

 

Tight Character Limits

Tight character limits can end up at odds with the things that people want to say, resulting in posts that are worse to write and to read. When users run up against these limits, they face a choice: cut it down or spill over, each of which has its downsides.

• Cutting it down can risk changing the tone or the meaning for the worse. Whether that involves ditching clarifications, deleting phrases used for politeness, eliminating words used as hedging or qualifiers, or just generally snipping out context and nuance, the trimming process can easily result in wording that is less clear, less polite, and more absolutist. This is a recipe for generating unnecessary conflict.
   
• Alternatively, users can choose to spill over the character limit, risking future headaches by generating clunky segmented "threads." "Threads" feel miserable to wade through, like trying to watch a video that keeps buffering every two seconds. The more posts per thread, the more opportunities for one singular segment to be taken out of context via the share button, potentially changing the meaning or making it easier to miss that something was meant as a joke. Each individual segment has its own discrete comment section, too, making it impossible to get an immediate bird's eye view of what's already been said, which is great if you want to generate excuses for people who fire off a snippy retort before checking to see if a point has already been addressed.

I'm with Kev Quirk on this one—just publish the thing someplace else and then post a link. Or that's what I would do, anyway, since I'm a big believer in the value of links, but unfortunately…

 

Anti-Link Culture

Imitating Twitter is bad for developing a site culture of creating and clicking on links, which has consequences for the spread of misinformation. People might occasionally post individual links (at a rate of one-per-post), but more complex linking styles are inhibited by the constraints of the format. So folks don't link very much, and they don't click much either. Together these factors can normalize a situation where people fire off claims without evidence and then other folks just accept what's in front of them without doing the bare minimum to verify.

Structurally, the twitterlike format does not lend itself to link-rich text. Links can be inserted only as raw URLs (the whole webpage location fully written out), with no option to insert links directly into the user's own words, and so links end up hogging a lot of space in a context where space is already scarce. That scarcity and inefficiency creates a disincentive to bother. The resulting disinclination toward links may or may not be a factor in why...

Twitter users generally don't click on links, even for posts they decide to share. That lack of clicking makes it a lot easier for misinformation to spread unchecked, since people don't stop and investigate before they share. Taking the time to click through is just not a priority. So if folks want other people to actually read what they write, they end up functionally pressured to cram and chop up their posts into a more punishing format.

And speaking of which, God help you if you make a significant typo in the process—like accidentally leaving out the word "not"—because…

 

No Edit Button

By withholding an edit feature, Twitter chains people to their mistakes, and cemented immutable mistakes have the potential to cause problems. For instance, in 2013, the formatting of the hashtag Now Thatcher is Dead allowed it to be misinterpreted as Now That Cher is Dead, provoking a premature outpouring of grief from her fans. Typos and autocorrect can easily transform a friendly message to into a rude one and generate misunderstanings. During an emergency, inaccurate information can propagate farther than subsequent posts issuing corrections, potentially even putting people in danger when health and safety are on the line. These are all scenarios where people deserve to be able to edit their own posts once they realize the problem.

In many cases, just deleting the post wouldn't necessarily be helpful or desirable. That might stop the spread of the original post itself, but by that point other people have taken the mistake and run with it by making their own mistaken posts—posts that are harder to correct if the separate corrections are harder to find. Given the character limit, the problem post might even be somewhere in the middle of a longer thread, which a user might not want to mangle with a mid-thread deletion. This creates an inhibition against purging mistakes and points of confusion, and nobody should have to weigh that kind of tradeoff when a problem could have been addressed with the addition of a simple feature: let people edit their own posts.

Mastodon at least gets this right, but Twitter was wrong for doing this to its users and Bluesky was wrong to follow its lead.

 

Share Additions = Bad Shares

A share addition feature (or "quote posts") can poison site culture by giving people a reason to share bad posts for no other reason than just to announce they disagree. Sharing-to-disagree means that the more widely something is recognized as bad, the more it gets passed around, which means the site is basically designed to elevate its very worst. This is such a common part of site culture on Twitter that it prominently features in parodies like this one by Tumblr user James utopians:

A retort this shallow does not justify spreading a bad post around to more people via the share feature. It's a bad thing for platforms to incorporate a feature that facilitates bad shares, and it's a bad thing that Mastodon is adding it too.

What's worse, when people share something to disagree with it, those shares can end up contributing to the prominence of that post on...

 

Trending Feeds

Anything that ranks posts by raw numbers and shoves them in your face is a feature that necessarily amplifies your exposure to emotionally provocative material. For an explanation of how this happens, you can check out CGP Grey's video on how people get sucked into perpetual fights. That video is based on "What Makes Online Content Viral?" by Berger and Milkman (2009), which found that the emotion that's the most conducive to going viral is anger. So if anger provokes people to interact more, and if the raw number of interactions is what boosts the odds of ending up on the trending feed, what you've got are the ingredients for a vicious spiral.

A continuous slew of provocations like this creates an atmosphere where people are quick to anger. That's why people represent their experience of the Twitter trending topics list with parodies like this one—listing genuine atrocities next to things that are merely annoying or uninteresting, like "new trailer for movie you hate" and "some celebrity you don't care about." And that is what it feels like, isn't it? That juxtaposition of the horrific and the mundane, petty annoyances amplified by catastrophic stress, all blending into each other and priming everyone to get fed up and snap—even if it means unleashing their ire at the subjects that deserve it far less.

 

Decontextualized Encounters

Decontextualized encounters created by mass aggregate feeds lend themselves to unwanted interactions and needless conflict. This isn't just a matter of "trending" feeds—this is a problem with general feeds of unrelated posts getting displayed en masse, stripping away the kind of contextual factors that would inhibit drive-by rudeness. In order to make sense of why, we can contrast those experiences against contextualized encounters.

Contextualized encounters involve some kind of social context that mediates people's reactions. For example, if a post gets shared by an account that's followed by someone else, that person can interpret that post in line with their relationship to the sharer, which may temper a negative reaction and incline the reader to be more charitable. If a person adds a tag to a post and another person runs a search for that tag, both parties involved have made a specific choice to enable encounters on that topic. In cases like these, encounters happen with a baseline degree of context—and that's exactly the kind of context that gets stripped away by mass feeds like "federated feeds" and "local feeds" that simply expose whatever to whoever.

A mass aggregate feed is a roulette wheel. It provides little frame of reference for what to expect, without any particular context or connections to make sense of the result. Under those circumstances, people have less of a reason to hold back on the hostility when they encounter something that provokes a negative reaction or strikes them as worthless. So from the perspective of the original post author, they wind up with strangers showing up in their replies who 1) do not directly share any particular personal ties with them, and 2) do not share their interest or investment in the subject, a situation that generally lends itself to people acting dismissive and standoffish. This would explain why the users of Mastodon sites end up with such a Twitteresque disdain for "randos," to the point of even proposing a "before you reply" feature.

Plus, the hazard those encounters pose is amplified all the more in light of two other problems: inline tags and no host veto.

 

No Host Veto

Denying users the ability to delete comments from beneath their own posts leaves users vulnerable to just about anything that other users decide to throw out there. No real means to keep a discussion on topic, no way to keep things from getting out of hand, no option to take out the trash. Even for statements that are against the site-level rules, we can expect a delay in between a user reporting it and a moderator taking it down, with users left powerless to do anything more in the meantime. So even if they personally decide to block someone over it, those replies are still left hanging around, potentially impacting other visitors for the worse or even baiting them into a fight. For more on this issue with Mastodon, see Designing for The Guest/Host Relationship As Part of A More Prosocial Web.

In response to this problem, Bluesky has introduced a novel half-measure that frankly isn't good enough. Blocking another person covers up their replies on your posts so that they're hidden to other people, but those hidden replies can still be accessed, and the block-hide outcome end up combining effects for scenarios of wildly different severity. For instance, there are times when someone might want to delete a comment for being off-topic without having to block a person's entire account, but on Bluesky it's all-or-nothing. That means there's no preemptive way to tell if someone was just blocked over a petty annoyance or if their hidden comments are genuinely repugnant, and so if curious onlookers get the wrong idea and decide to take a peek, then the block-hide effect has failed to be as effective as an actual delete button.

 

Inline Tags

Hashtags that appear directly within the main post, Twitter-style, set the stage for unnecessary conflicts over blacklisting. Mastodon blacklist filters rely on text in order to identify which posts should be removed from a feed, and that's something they can do more effectively if descriptive tagging is the norm, which can be encouraged more easily if there is a separate, visually de-emphasized text field for tags. This is completely opposite to the twitterlike approach of shoving hashtags right into the post, making them look like excessive clutter. So although tags do still get used on twitterlike sites, in effect they become associated with loud and assertive self-promotion instead of a quiet default practice for every post.

The resulting paucity of tagging habits may help to explain Mastodon's infamous conflicts over the "content warning" feature. A content warning refers to a specific feature that collapses a post under a short description of what's inside. The framing of a description as a "warning" necessarily inclines people to think of the feature as a dramatic negative judgment, thereby sparking endless disputes about how to use it. There is comparatively less baggage around tags, which could be a great solution for blacklisting purposes, but tags are functionally discouraged by the twitterlike format. Consequently, the majority of posts go untagged entirely, and so people end up at each other's throats over "content warnings" instead—a situation made all the worse, again, by decontextualized encounters.

 

What It Adds Up To

Twitterlike is a bad shape because these structural decisions contribute to needless strife, stress, and hostility. Tight character limits force people to make their posts either short and snippy or segmented and vulnerable to decontextualization. The format discourages link culture, the lack of an edit button chains people to their mistakes, and the share additions encourage sharing-to-disagree, which results in bad shares that amplify the worst posts on the site, contributing to the negative effects of trending feeds that skew toward the emotionally provocative and distressing. Mass aggregate feeds in general create decontextualized encounters that easily lend themselves to pointless spats, especially when inline tags contribute to conflicts over blacklisting, and when people do go overboard with nasty replies, the original post author isn't even able to get rid of them.

The result is a pervading atmosphere of hair-trigger tempers and suspicion. That's how Twitter became famous for the whole new sentence and the kind of noxious, vitriolic atmosphere where people get attacked for things as innocuous as chatting over coffee. Pointing out these problems is no invitation for Mastodon users to get smug because Mastodon's reputation is much the same: elitist, pedantic, snidely dismissive, and an opportunity to get yelled at. Compared to Mastodon and Twitter, Bluesky is relatively new, but already its site culture is shaping up to go down the same road.

It doesn't need to be like this. People can design, use, and advocate for the kind of sites that aren't so twitterlike—sites that are better for writing, better for reading, better for linking, and better for conversation. We can have a more in-depth discussion about what that means, but first we need to get on the same page that the fundamental shape of Twitter is, in fact, bad.

Note: guest comments are screened! This means that if you submit a comment without an account, it will not be published until I manually approve it.

Comments

[claudeb]

Pretty much all of this. Mastodon mitigates some of these issues through larger (often much larger) character limits, no punishing links (no algorithm at all) and sharing the moderation burden. On most servers you can even enable Markdown and label your links properly. But it's not enough. The formula is fundamentally flawed.

[osteophage]

Glad to hear you agree. I appreciate Mastodon's deviations from the formula, and it could be so much better if it would deviate even more.

[grayestofghosts]

Weirdly the microblogging site that solves *most* of these problems (though not all, and it has issues of its own) is... tumblr. But it seems like both tumblr and tumblr-like sites have had serious issues with profitability/funding. It kind of seems like the major line for keeping a site free to use and not made into a hellscape by financial motive is still, after all these years, unlimited image hosting. Maybe we need fediverse tumblr -- surely they have that already!

By the way, does IndieNews have an RSS feed? I was trying to find one and would be surprised if they didn't...

[osteophage]

Structurally, Tumblr's got its own slew of issues that I wrote about back in 2019 (so parts of that post are kind of outdated). The financial issues are something we don't have a lot of exact information about, but yeah, I'd imagine a big part of it comes from media hosting, since part of how Tumblr made a name for itself from the beginning was its animated gifs. A blogging site looking to keep costs under control would need to impose some serious limits on media hosting, like Dreamwidth and Pillowfort do... which newer users end up chafing at because they're used to a way bigger budget.

Good news though, IndieNews does have RSS!

[grayestofghosts]

Media hosting costs is definitely a huge issue. I keep not writing things because I expect what I've written already exists, but I probably should write an article or something about how evil the "pivot to video" manipulation was. While images are expensive, you can still get enough free/cheap hosting to host a significant amount as an individual. By faking numbers to make video the only thing that gets attention online, it not only forces creators to rely on big web hosts but also makes it unlikely that people put in the time to create less-resource-intensive resources to things like how-to videos.

[osteophage]

If you do write up that post, I'd be interested!

[axolotls]

There is goblin.band, made by a former member of tumblr staff.

[kradeelav]

found this originally on the webdiscussions community on pillowfort, much appreciate you crossposting here. :) this articulates *so* many issues i have with twitter and its clones compared to DW or even tumblr. (ironically very similar to my latest post about context collapse).

i wish this wasn't an issue but it's gratifying to see others see it.

[osteophage]

Oh, huh, I didn't realize anyone was lurking WebDiscussions from the outside. Thanks for stopping by! And glad we could validate each other's grievances.

Is this the post you mean? Context flourishing... that's an interesting way to put it.

[kradeelav]

Oh yeah, I know at least one other person that lurks on WebDiscussions, probably more - never got on pillowfort since they're not kind to certian kinds of artwork, but the discussions on that thread are excellent.:)

and indeed that was the post! glad you like.

[axolotls]

(Oh, there was some discussion about what art is allowed recently, actually, here's a link if you (or anyone passing by) is interested (the second comment is one where I link some threads from rahaeli that you might be interested in, too, if you haven't seen them already; I hadn't known a good chunk of what was said there!)

[kradeelav]

interesting! I was familiar with Rah's thread but not the other link. appreciate the food for thought.

[osteophage]

Well thanks for visiting all the same.

Note the matter of what's prohibited on PF isn't a subject I want to play host to on this particular post -- not saying that either of you have done anything wrong or crossed a line, just letting you know the line is nearby.

[kradeelav]

noted, and appreciate the stated boundary!

[axolotls]

I'd expected as such, no worries, just wanted to share the information without getting into it further. o7

[shuttersnek]

Another WebDiscussions lurker here; this is a really good writeup. It effectively communicates a lot of the issues I have with social media.

One thing I’d like to point out re: “a mass aggregate feed is a roulette wheel” is that, imo, it ends up turning social media into a bit of an addictive Skinner box. By presenting a bunch of bland or negative posts in between stuff that gets you happy or emotionally charged, it turns the website into a sort of slot machine where you have to keep pulling for the good stuff. Combine this with reverse-chronological feeds making it likely you’ll miss posts (especially if you follow too many accounts to reasonably keep up with) and it can quickly spiral out of control into an obsession. (And stick some ads in the mix and presto - a business model!)

[axolotls]

My experience has been that reverse-chronological feeds mean I don't miss stuff, whereas as far as I've hears algorithmic feeds don't show you all of the posts from the people you follow?

[osteophage]

Thanks! I appreciate you saying so.

Hmm, I think I see what you're saying there. Even if people end up disproportionately encountering things they don't like, the element of unpredictability there can potentially feed into a gambler's fallacy.