Back to Thoughts ·· LIBRE.TOWN

Where the Fediverse falls flat

Hey there!

In this (pretty long but important) post, I would like to expand upon the observations I have already hinted at in my previous blog post about quitting social media.

Specifically, I will talk about the problems I have with the Fediverse particularly as a way to find friends and socialize online.

First of all, I want to stress that I think the idea and technology behind the Fediverse and ActivityPub are great. I am using some federated services myself, actually, and being part of the Fediverse is definitely a better place to be than being on commercial, proprietary, centralized social media. This is also absolutely not going to be yet another opinion piece about how the Fediverse is oh-so-complicated to get into or needs to be much more "user-friendly", either.

I just also feel that some of the most popular applications within the Fediverse are inadequate to make and sustain friendships and often even reproduce the same issues that centralized, commercial social media sites have. They do not differentiate themselves enough from others.

Keep in mind these are my personal feelings and your mileage may absolutely vary, mmkay?

PNG: Spherical Mastodon logo pixel art

1. Familiar Patterns

Many services built on the Fediverse seem to have been conceptualized as "Twitter but federated", "Facebook but federated", or "Reddit but federated". Unfortunately, this sort of design strategy often leads to uncritical adoption of design decisions that might seem innocuous on the outside but are actually part of what made the original "service" so undesirable and toxic in the first place.

Some Fediverse teams attempt to mitigate these problems: for instance, Mastodon does not and will never have an equivalent to quote tweets, since they were famously used on Twitter to harass and send an army of followers after people without any way to properly respond if the follower count disparity was huge enough. As for another example, many projects also criticize "the algorithm" harshly, referring to the practice of curating a personalized front page using a content recommendation mechanic. Some limit the unhealthy practice of endless scrolling by designing a "front page" (if there even is such a thing) as an assortment of pages, not as an endless feed of content.

Unfortunately at the same time some other design staples aren't examined enough before blindly implemented. The goals of a non-commercial, community-run social media service is fundamentally different from a commercial one: the latter seeks to maximize profits, engagement and user count, while the former seeks to facilitate authentic, deep social interaction between people and the making of friendships. And this difference in goals should lead to a difference in design attitudes.

Twitter's whole concept, and by extension Mastodon's, does not work towards the latter. Tweets are short-form messages broadcast into a big pool of other messages without context, making actual conversation difficult to impossible. The fact that tweets are mainly presented as standalone posts means that replying to someone is a social challenge to overcome since any short comment is at the same time a whole Tweet in its own right. Any conversation or interaction needs to justify itself as "good enough" to be tweeted.

Other users feel inapproachable since you might have been "connected" with them for years but never interacted beyond likes, retweets or the occasional comment. Making friends is difficult like this and usually requires third-party avenues like a messenger to really stick.

The character limit combines with all of these other traits to the result that the only thing you can really do on Twitter is post updates or news about your life (which is not really social and could be achieved much more simply with something like a mailing list or an RSS feed), or just blast provocative opinions, identity snippets, callouts or quippy one-liners that fit into the character limit into the wild and hope they get a lot of validation from others. It is discouraged to post about niche topics, your hobbies or personal things since most people who might see it will not share those interests or an interest in you as a person; and therefore those posts will get much less traction than, say, a provocative political attack. This creates the toxicity that Twitter is known for.

This is exactly what Twitter wants: scrolling through endless short posts means a lot of addiction potential, plenty of space for additionally injected content (sponsored tweets), and therefore ad revenue. Less time spent interacting means more time spent reading and taking in content. The fact that other Twitter users are less approachable than they should be creates a sense of awe and envy when it comes to people with a lot of followers, engaging a mental loop of wanting to be as "cool and popular" as they are. Most Tweets being about outrage, attacks or one-upping others also brings a lot of (negative) engagement to the site, encouraging people to get into lengthy flame wars and therefore spending more time on Twitter than if they had had a good, chill time.

Mastodon has copied much of Twitter's design in this regard. The stereotype that Mastodon users mostly talk about Mastodon itself is a byproduct of this; there simply is nothing to talk about, since the only thing you have in common with those who see your toots are that they also use Mastodon (I will get back to this aspect later), and interaction is discouraged for the reasons stated earlier. When I was on Mastodon, most toots I would see were, as predicted, provocative one-liner political opinions, jokes, uninteresting personal updates á la "had a coffee this morning", and call-outs/PSAs. Nothing that you can form friendships over or that encourages real bonding between people.

This pattern of copying the structure of the "original" social media concept is detrimental to a working indie social media world.

We see this especially poignantly in inverse when looking at systems that were not designed as profitable social media at all, like forums: forums encourage long-form posting, ensure "old" content stays relevant for years with the thread system, and encourages specific forums for niche topics that encourage people with similar interests and personalities to interact with each other. The information density is incredibly high compared to something like Twitter, which makes them attractive and long-living plazas of social interaction. They focus on content over identity and function over form. Forums are also almost impossible to monetize and have not been designed with that in mind, so they are radically different from most other social media concepts.

If the Fediverse seeks to be better than commercial social media projects, it needs to design new systems and paradigms for social interaction online from the ground up, not simply imitate the design of the very commercial software they want to replace.

2. Designed for centralization

Another (albeit related) point particularly comes to mind when thinking about Lemmy and other federated Reddit replacements, especially in the wake of the recent troubles with the Reddit blackouts.

Reddit at its core "works" only due to its centralized design. Its appeal is that whatever niche topic you are into, whether it's a specific video game, a hobby, a music genre, or even just a type of meme, there is a subreddit that ideally holds tens of thousands of people who are just like you, on Reddit dot com. They can help you with issues you face, involve you in the community, show you things you have never heard of before, give great advice and generally act like an invaluable resource. Many people, as commonly reported, append "Reddit" to their web search queries, simply because amateur enthusiast communities on Reddit and their archives are way more trustworthy and searchable than dubious blog spam with unclear motives and authorship.

A huge reason for this is that Reddit has only one major subreddit per topic. Once you find a community about "your" topic, you can be reasonably sure that almost everyone in the community who uses Reddit at all is on there - and that is usually a number in the high tens, if not hundreds of thousands.

On federated Reddit replacements however, many different servers will host communities of the type you are looking for. If you want funny pictures, there is no way to get all funny pictures from every Lemmy instance, only ones from those communities you follow and curate. If you want the best of the best in mushroom enthusiasm and identification, you will have to post your question on the mycology communities of all servers that might hold the people who know your answer. And it is likely that they might be somewhere your server does not even federate with (like a mushroom-specific instance), necessitating the use of third-party methods to find that community.

Reddit is an aggregator of communities that each aggregate links and content. Lemmy can only be an aggregator of such aggregators, which can work only if each Lemmy instance also caters to one topic with several sub-communities (like startrek.website), but not with general use Lemmy instances like lemmy.world, lemmy.ml or feddit.de. Those only lead to redundancy.

3. Users

The best social media system in the world cannot work without its users. And while some Fediverse software have cracked critical mass to be self-sustaining (such as Mastodon or Lemmy), some others are almost entirely devoid of a real or worthwhile user-base, no matter how much I like the systems themselves.

I enjoy Diaspora, Friendica, Pixelfed, GNU social...; but without people on there, much less people I care for, these systems are entirely unusable. I could have the most beautiful profile but with nobody to click on it, it does not do me any good.

The situation gets worse when the accounts that do exist on the big instances tend to be alt-right boneheads, bots reposting articles from other sources, spam, inactive users and conspiracy theorists. Diaspora especially had an issue with it when I still used it actively, and so does Peertube depending on the instance.

The nature of federation helps a great deal here; as long as any of those services are compatible with ActivityPub, then I can interact with the rest of the Fediverse. However, crudely said, if I wanted to talk to Mastodon people, I would have a Mastodon account in the first place. It is nice not having to register to a Peertube instance just to comment on a video, but generally, people in the Fediverse underestimate how much the platform they are on shapes the way they interact. If I am on Friendica, I want to interact with people in a long-form, personal way; I want to meet people who talk about their personal lives at length, post pictures and excerpts of their day-to-day, I want it to be like old Facebook used to be; but if I choose to interact with Mastodon users due to a lack of Friendica users, then they will see my content on Mastodon and I will see theirs on Friendica. We will each have a different experience and different ways of interacting with each other, even different features built on top of ActivityPub that we cannot use on each other.

People who choose Mastodon over Friendica will simply tend to be a different kind of people than I want to meet on Friendica, so federation is not a solution to user droughts.

4. Conclusions and where to go from here?

There is one recurring theme in most of these observations, but it is not quite obvious at first:

In my opinion, it is massively, vastly harder to make friends and interact with people on sites that are designed for a general audience, than on "niche" sites for a specific hobby, interest or piece of media.

Specialization beats size.

All friendships and relationships I have ever forged online came out of something we had in common that we both cared about and interacted in the same spaces of. My first and longest romantic relationship came from a Minecraft forum; most friendships I have built online originated in unlikely places where we happened to talk about the same topic at the same time, all the time, and just got closer doing what we did anyway.

This is also why almost all successful social media sites tend to group their users in specialized communities by the use of a recommendation algorithm, personalized for-you pages, and so on. They realize that people only really interact with or want to see content of people they have something in common with. This is why things like "BookTok", "Black People Twitter" or "Fanblr" exist. The more you interact somewhere, the more closely the content you see resembles the content you seemingly want to see.

It is also why Reddit is so useful and engaging: its "sub-reddit" structure makes the structure even more explicit, since it effectively means that Reddit as a whole houses many separate topic-specific communities without the requirement to interact with people who are so different from you that you don't want to talk to them at all.

On the other hand, on many such and other mainstream pages I felt like (despite these attempts at curation) I had no way of making friends and no reason to interact with anyone.

I recently had an account on SpaceHey - an independent social media site meant to replicate golden age MySpace - and had a blast customizing my profile, filling out interests and likes and dislikes, writing a bio and picking out a wonderfully emo edited picture of myself. It was super fun!

But after that - nothing. There was no organic interaction since nobody had anything in common with the average next person. The SpaceHey forums were thematically sorted and would be my first place to look, but they were so general and non-specific that it was practically impossible to find people who happened to be into the same thing you were. Besides, moderation seemed to be almost non-existent, so the "web and computers" interest category on the forum, which I expected to be full of webmasters, questions and discussions about technology, the internet, software development and so on, was instead full of questions about how to customize your SpaceHey profile (despite there being a dedicated section to that and these threads being entirely off-topic for the forum they were posted in). Again, the only thing people had in common with the next person was that they both used SpaceHey, so that was what they talked about.

Of course it would have been possible for me to wade through the thousands upon thousands of user profiles, groups, forums and so on to find someone who was remotely close to me interest-wise, but that would have been very difficult and for me not worth the while; especially since there was no guarantee they would like me back, interact with me more than once or even be active anymore. It is not impossible to find friends there, but very difficult. I ended up deleting my SpaceHey account soon after.

The Fediverse has one theoretical big advantage over these sites: it does not need the crutch of being one big general use community with sub-groups tacked on; there can be many different instances catering to many different interests, niches and hobbies. A for-profit, commercial social media corporation has things to worry about like brand dilution, spreading of resources and many other issues. On the other hand, the Fediverse can consist of many different topic-specific instances hosted by regular fans of the niche they cater to for so much less cost. And through federation, they are not isolated from one another, so the expected small user base can expand without the need for separate accounts.

But in practice, this massive potential is barely used! Lemmy users for example largely congregate on Lemmy.World, Lemmy.ml and some other general use instances with thousands upon thousands of users. The only major popular counter-example is startrek.website, which was built by and for Star Trek fans, containing several Trek-related communities.

A big reason why early-2000s German social media sites meinVZ (working adults), studiVZ (college students) and schülerVZ (school students) were so successful was that they were catering to each of these (still broad) social groups separately, and therefore it was generally easier to make friends on all of them. It is also why they failed - it was difficult for a corporation to run three massive social media sites at once.

If I could end this article on one big takeaway, this would be it: both the Fediverse and the independent social web in general need to stop building and supporting large, loose spaces catering to the general public. The best online spaces to socialize are medium-sized special interest communities, ideally many of them.

The Fediverse is uniquely suited for such a task due to the self-hosted model.

Addendum: A Note on Safety

TW: This section contains references to harassment, transphobia and malicious mishandling of user data. Read at your own risk.

Your average federated social media platform has both big advantages and big disadvantages compared to their more popular, mainstream cousins when it comes to your own security as a user.

On the one hand, the free or at least open source nature of the vast majority of the Fediverse makes the software itself infinitely more auditable. Due to the lack of a direct profit motive for most projects in combination with the just mentioned openness, there is very little incentive to implement hostile, illegal or unethical "features" such as privacy invasions, dark patterns, or spyware. When in doubt, anyone could fork the project and remove the issue. This is a huge advantage.

Additionally, you do not have to trust a giant for-profit with your personal data - and plenty of data we are talking about on a social media site. You can choose someone you trust: for example, your local educational institution that hosts a Mastodon instance, a good friend, an organization you trust, or even yourself.

On the other hand, not everyone has the knowledge, connections or time to either find an instance they trust or host one themselves. Most people will simply choose a big or random one that might align with their interests or aesthetics without auditing who hosts it. This puts personal user data into the hands of someone who may or may not snap at some point, be responsible with user data in general, host personal data under their kitchen sink, honor GDPR requests at all, ... While corporations are not exactly saints about data protection, you can at least be reasonably certain that they misuse the data only for profit or to assist law enforcement; there is no corporate profit in something like blackmail, satisfying personal vendettas, harassment or data leaks. This is not a given with private hosters who might as well be anyone, including malicious or unstable individuals who care even less about the law than corporations.

Another security risk comes from the concept of federation itself: it simply is not really possible to delete anything off the Fediverse because as soon as something is published, it spreads throughout every other associated server like a virus. Deleting something only really deletes it off the instance you use (if they are using default software that is) and practically only sends a polite deletion request to other servers, who may or may not honor that request. Either way, your content is cached on countless servers already, and it really ever being deleted from everywhere at once is a pipe dream. Some servers in the Fediverse, albeit still a small minority, are hosted by malicious actors such as alt-right communities, who have a vested interest in bullying and harassment, and sometimes do not honor deletion requests by use of modified server software.

Now now, you will say, the internet itself is already permanent and anything you post online is already public forever. That much is true; given someone dedicated, anything can be found on a web archive or in cached versions, and many commercial websites are more than lax with whether they actually delete or only hide content meant to be removed. But generally, having one centralized place to host data in and if necessary delete it again has less of an attack surface for malicious actors than being connected to many independently run servers who all receive and mirror that same data.

This is not even to mention people who abuse modified Fediverse software in order to bypass the "rules"; I am still receiving threatening "xyz followed you" exploit messages from neo-nazi instances more than a year after deleting my Mastodon account on Toki.Social, the toki pona instance.

Especially for minorities like trans people, queer people in general, activists and so on, this type of data handling may be particularly dangerous.

Anecdotally, I knew a trans woman who used her Mastodon account to talk about her personal life and give status updates every once in a while; she had her regular followers and friends comment and generally lived a small, quiet life in the Fediverse posting about her morning coffees, pets, her job, family and daily routine. That was all well and good until she noticed that certain transphobic alt-right instances used modified Mastodon software to boost, comment and interact with her pictures and posts, hidden from her own instance but visible to anyone else. Anyone who clicked on her profile from the outside saw at least ten comments that she had never seen under each of her personal photos and updates, threatening her, ridiculing, insulting, laughing at, sexualizing and even doxxing her, boosting her to transphobic audiences. After she deleted her account, horrified, the posts prevailed, since the other instances did not honor the deletion request and cached and re-posted the content themselves automatically using bots.

I am not aware of this being a popular exploit or whether it has been fixed in the past years, but it generally turned me off of the Fediverse for a long time. With the potential for modified instances and shared hosting comes a great deal of freedom, but also a major risk.