The "No Clear Winner" Era of Federated Microblogging

[cart]

This is a followup to my post suggesting that Bevy consolidate its microblogging presence to Bluesky, which is currently spread out across X, Mastodon, and Bluesky. I also strongly encouraged the community to do the same, for both practical and ideological reasons. If you haven't already, go read that post (and the responses).

Our decision is to stay on both Bluesky and Mastodon for now, while dropping X support. I still strongly believe that further consolidation should be our goal, but I concede that the jury is still out on Bluesky vs Mastodon ... that particular decision needs a bit more bake time. I concede that the realities of the Bluesky implementation are currently suboptimal. We have good reason to believe they will get there, but I understand and respect peoples' desire to wait and see.

The response was largely as expected. Many people are on board, but a variety of concerns were expressed, which I will do my best to address (or at least cover) now. The concerns coming from people varied by platform:

Concerns from X users

This move is too political

I received a number of comments (exclusively on, or in defense of X) accusing me of being "overly political". First, I think it is impossible to run a project while also being completely apolitical. Interacting with people (and running a huge project that consists of and markets itself to people), is politics. If you don't believe that, I encourage you to ponder that question for a bit (ask yourself why I might think that) and if you are still dubious, we can argue about it.

That being said, I believe the people making these accusations are accusing me of making this decision as a response to the specific politics of the moment (X is increasingly right leaning, Bluesky is increasingly left leaning). I will admit to being left-leaning personally, but the argument I provided for moving is intentionally decoupled from that, and I agree that a project like Bevy should generally not explicitly bias against specific political groups (especially something as large and nebulous as "the right"). I assert that my argument should appeal to people regardless of their political agenda. If it does not, I encourage you to consider whether it is you that is politically compromised.

To re-summarize my argument (but please read my previous post in detail):

1. I believe strongly in "community". It is very hard to build a community when there are hard walls between everyone. I believe a large part of Bevy's success (both technically and socially) is that we have "one place" for each thing (Github for development work, Discord for real time communication, and for a long time Twitter for microblogging). I believe the "3 separate microblogging worlds" situation is actively impeding our ability to come together as a community. I also believe that is true outside of our community. I don't want to (and dont) check three apps every day. I don't think anyone else does either. If you stop and consider the consequences of this, it is a big problem, not just for projects like Bevy, but for everyone.
2. Assuming the goal is consolidation, we should pick the platform that best serves us.
3. We should pick a platform that is federated, where you have ownership and autonomy over your identity and your audience. Social media should not be own-able by a single group or person. Social media is serious business. It ties directly to human rights and business outcomes. It shouldn't be in anyone else's control but yours. This is a political opinion, but it should appeal to the left and the right equally.
4. Bluesky, when considered holistically on its technical, ethical, and "audience size" merits, "wins" from my perspective (see my previous post).

I also made an additional (admittedly minor) point that managing posts across 3 platforms introduces additional daily overhead (currently "paid" by me). This is not the primary issue, just another consideration on the pile.

None of these arguments are about "boosting or alienating a political side", and that is not my motivation. This problem transcends left/right politics, and should speak equally to everyone.

You should go where the people are, and the people are on X

I think it is pretty clear to everyone right now that the "microblogging space" is up for grabs. Everyone should be carefully evaluating what they want the future of the space to look like, and pick the platform most likely to get us there (weighing things like features, governance, and momentum).

Additionally, when considering the gamedev community, I think the "just follow the numbers" approach isn't particularly clear. For example, a week ago I did a breakdown of "posts worth reposting from official Bevy accounts" and this was our breakdown: Bluesky 18 posts, Mastodon 7 posts, X 6 posts. From a "Bevy content perspective", X is the least compelling platform.

Concerns from Mastodon users

Bluesky is run by a for-profit (American) company, which could do a "rug pull"

Bluesky is owned and operated by the Bluesky "public benefit corporation". First, I give effectively zero weight to the "public benefit corporation" title, as it provides no real checks on corporate greed. This is still a "for profit" company we are dealing with, which answers to its shareholders.

I will be the first to say that this is not an ideal situation. It means we need to rely on:

1. The realities of AT Proto and the current Bluesky implementation to protect us
2. The responsible stewardship of AT Proto (which from what I can tell is "governed" exclusively by the Bluesky company).

AT Proto (and the current Bluesky implementation) give us some reasonable protections:

1. You can own your own identity, in a way that is decoupled from the Bluesky instance (by using a domain you own as your identity, ideally in combination with owning your DID, which is actually not ideal at the moment). I consider this tangibly better than Mastodon, which requires you to tie your identity to someone else's instance (if you don't want to run your own, which is not something we can expect normal users to do). Yes Mastodon supports "instance migrations", but this only works if the instance you are migrating from still exists and is willing to facilitate your migration (by informing users of your move and redirecting requests to the new instance). In a "rug pull" situation, you cannot rely on this. And in a "your instance goes down situation", you are also out of luck (and instances disappearing should be a huge concern for the fediverse, as most of them do not make a profit and exist only as long as a few people are willing to operate them and pay the hosting fees).
2. You can easily host your own posts with a PDS (either one that you own, or someone elses')

This means that on paper, you can directly own both your content and your identity. A PDS is literally "just storage", so it is minimally cheap to set up and minimally complicated to host yourself.

The big question is "if Bluesky goes rogue or disappears, will people still see your content", and that gets more complicated:

1. Currently, pretty much everyone gets their content from (and posts their content to) the Bluesky relay, so if they choose to exclude you for whatever arbitrary reason, you disappear.
2. Most people (including those with custom domain identities) are using Bluesky did:plc, which supports migrations to other providers, but for a variety of reasons this is (currently) not bulletproof and requires trusting Bluesky to allow the migration to happen. Most people do not own the keys to their identity. People with custom domains can make them point to a new did, but that will functionally be a "new account" from the perspective of the network. So from an identity ownership perspective, if you don't own the keys to your did, it isn't yours.

When I wrote my first post, I naively thought that custom domains directly protected you against instance rug pulls by decoupling your identity from the instance. But because they (by default) tie to the Bluesky did:plc, which requires Bluesky's sign off to migrate to a different identity provider, this essentially replicates the migration issue with Mastodon (identity migrations require the original instance to still be around, and be ok with the migration). You can migrate off (by going through their sign off process), but it is one-directional, "advanced user territory", and you need to do this before Bluesky goes bad. Bluesky has stated that they want to improve this situation, but I have not seen any progress here.

Until there is more "ecosystem diversity" (more partial relays, shared governance of a large relay, more instances, etc) and/or identity is more thoroughly and universally decoupled from the main Bluesky instance, I understand why Mastodon folks are not ready to bet it all on Bluesky.

Bluesky is not currently "meaningfuly federated"

As touched on above, I agree that this is essentially true. AT Proto does support the creation of a network that shares messages and identities, but:

1. There is one dominating instance at the moment with no major alternatives in sight.
2. Migrating off of the official Bluesky instance requires the "ok" from Bluesky, unless you are an uber-power-user (that prepares ahead of time prior to a rug pull). For most people, their identity is not yet truly their own. This is also true for Mastodon, and the situation there is actually worse. AT Proto is built to support instance-agnostic identities, but the current realities of the Bluesky identity provider lock it down to something closer to the Mastodon model. Bluesky is better positioned to support true identity ownership (and technically already does for the power user), and has stated that they want to make this happen on a larger and more approachable scale (but we'll have to "wait and see").
3. The Bluesky model is built around the concept of an (essentially) global index: relays. AT Proto supports querying personal data stores directly, or having smaller-scoped relays, but in practice everyone (read: the Bluesky app) is treating the Bluesky relay as the source of truth. If you spin up a separate instance with a separate relay, you are cut off from the main Bluesky instance (unless you write app code to also query the Bluesky relay, or query PDSes directly). You can, however, feed your server's content into the Bluesky relay (and in theory they could do the same for yours), query PDSes yourself, etc.

(3) is kind of the crux of the issue on both sides. My big problem with Mastodon is that instances dictate your view of the world, it is hard to know as a user what subset of the fediverse you are seeing on your instance, and even under the best circumstances (you picked a "highly connected instance") that view is limited (and subject to the arbitrary whims of your instance admins). The fact that things like "what shows up in a hashtag query", "what comments you see", and "like / repost counts" are different for each person in the fediverse is deeply concerning to me. And users generally aren't aware of the arbitrary conditions that produced their particular filtered view. Access to the world is a user autonomy / human rights / collective action issue and I think mastodon does not do a good enough job here in its current form, and I would assert that in many ways the uncertainty is doing active harm.

I firmly believe that your view of the world should default to everything and you should filter down from there based on your individual preferences. This requires building one or more global views of the world that we can all point to and run queries on. This is expensive. Mastodon wipes its hands of this problem and instead relies on organic, lossy approaches. This is "resilient" and "decentralized", but as a user it is also terrifying, for the reasons mentioned above (or at least, it should be ... I don't think people are thinking or talking about this enough).

Bluesky's relay-forward approach re-centralizes (with all of the relevant concerns there), but in return everyone gets a full view of the world. Again, AT Proto does not dictate a single view, but culturally, organizationally, and technically (the official Bluesky app) a single view has been adopted. This allows Bluesky to provide what I consider to be a table stakes feature: a shared, queryable global view of the world. But comes at the cost of centralizing, and in the case of Bluesky, that centralization is in the hands of a for-profit American company. This is not an issue unique to American capitalism, but the world has certainly been given many reasons to distrust it, especially in the social media space.

Mastodon rough-equivalents to Bluesky relays are being proposed (and Mastodon currently has its own concept of "relays" which are slightly different in that the do not externalize the indexing problem). But "indexing and searching large views of the fediverse as public infrastructure" is an unsolved problem. Right now you just cross your fingers and hope that your instance admins handled federation in a way compatible with your desires, and that the people on your server happened to follow the right people.

This is a "pick your poison" situation. From my perspective, Mastodon has not been focused on the right things, and continues to not be focused on them. They have not yet built the "table stakes" platform that I think the world needs (large shared indexed views of the world, identity decoupled from instances, users personally selecting what slice of the world they want to see / what algorithm they want). Mastodon may get there some day, but for years now members of the community and leadership have actually embraced these flaws, implicitly (via actions) or explicitly (via words) saying that this is actually the future we should want. BUT Mastodon also has true federation today, both technically and culturally. You can spin up your own instance and participate "equally" (although doing this would make it much harder / more expensive for you to view the world "equally"). It is run by a non-profit, which buys even more trust.

Bluesky on the other hand, is building a network and protocol that is federated, but they have built an app that re-centralizes it under a single relay (which is owned by a for profit company). Note that Mastodon also treats instances as a "single relay" in a way, but users notably have many options to choose from (and they generally aren't for profit). Bluesky "supports" owning your identity, but they have set up the defaults in a way where you definitely dont (even when it appears that you do, in the context of using domain name). Once again, Bluesky (the for profit company) is the gatekeeper. BUT they have also managed to build a large, global, indexed view of the world. They have created the conditions where others can do the same (and open sourced their code). The protocol is built in a way that maximizes user autonomy (even more so than Mastodon imo), and the system as a whole acknowledges (and actually implements) the need for global views, which users get to filter down.

Mastodon is less functional today in terms of what I think the world actually needs from a communication system. But it is also more trustworthy and resilient to bad actors. It may become more functional in the future, but you're taking that on trust. And every day they fail to do this, they will lose potential users to Bluesky.

Bluesky is much more functional today in terms of what I think the world actually needs from a communication system. But it is also less trustworthy and is not currently resilient to bad actors. It may become more trustworthy and resilient in the future, but you're taking that on trust.

This is a gnarly situation and the choice is not nearly as clear as I made it out to be. All I know is every day we continue existing in this middle ground state of affairs is a day of lost potential, where we are more disconnected, and where we risk losing to worse platforms without any of the scruples we're worrying about here.

Concerns that overlap across platforms

Just bridge your posts across platforms

This would not solve the fundamental problem of "community fragmentation", as we would not be bridging everyone else's posts, nor the comment threads. The labor of posting across platforms is minor, and is not the primary reason for this switch.

You can't just tell people to move platforms (this is hubris)

I have a modest amount of reach and influence as Bevy's creator and project lead. I think my argument has (for better or for worse) already convinced a subset of people. I'm certain that our various social links on our website / github / etc have been peoples' first step into the fediverse, bluesky, or X (or encouraged people to spend more time on a given platform). If the convinced people move platforms (and the "bevy microblog converts" all go to one place instead of three), they will create content on a specific platform, which will then further increase the draw of that platform. I don't think this is hubris. I have a rough conception of how far our reach extends, and that is enough to enact some level of change.

You shouldn't tell people to move platforms (it is unethical)

I believe resolving the current "three platforms" situation is a high-priority issue for Bevy and the world at large. Removing arbitrary communication walls between us is an ethical / human rights / collective action issue. I think the "federated platform infighting" is causing active harm to people (admittedly, competition does also breed innovation, but so far that innovation has not involved connecting people across platforms).

I understand why some might find me exerting my influence in this way unpalatable, especially given that Mastodon is "more ethical" in some ways (developed by a non-profit, more federated in practice).

Conclusion

We will remain active on both Bluesky and Mastodon, and drop X, with an eye toward consolidating on one platform as soon as the remaining concerns have been addressed on one side or the other (increased federation and/or identity-portability for Bluesky, improved global visibility, algorithmic choice, and/or population on Mastodon).

I am still biased toward Bluesky, but I also understand why Mastodon folks are unwilling to bet on it at the moment. I think with time those concerns will be addressed, but until then I'm content to wait and see.

[zicklag]

This is definitely a much more complicated and sub-optimal situation than we would hope for.

I've spent most of my time in the last year, since I postponed gamedev stuff for the most part, working on building social web projects with @erlend-sh and getting a grip on the Fediverse, Mastodon, ActivityPub, Bluesky, ATProto, and tons of other "open web" protocol stuff.

It sounds like you've got a rather good grip, @cart, on what the situation is. I think I pretty much agree with everything you're saying.

Yesterday I finished a blog post, ATProto Isn't What You Think, which may be relevant for people wanting to get a little more background on ATProto. This is semi off-topic, and not meant to sway the decision in any way, just to add some extra context to the protocol side of things for those interested.

I was kind of averse to ATProto initially because of the apparent centralization, and much of the concerns still exist, especially when your primary interest in ATProto is to use Bluesky, but it's held up better than I thought in some important ways after I really got into it.

Again, it's slightly off-topic, dropping it here in case it might be useful.

[Bluefinger]

Thank you for the write up, gives plenty to think about. However, I just want to throw in some of my thoughts given my own experiences with social media and also how I feel about certain things:

a large, global, indexed view of the world

I am not convinced this is actually a good thing to have by default, especially in the context of microblogging. Context Collapse is a thing, and having the global view exacerbates this. Getting a firehose of stuff thrown at you in terms of a global view might be great at hooking people in to use your platform, but then expecting users to filter that down to just what they want assumes that the firehose of content isn't the constant and addictive dopamine hits that contemporary social media abused so much to entrench themselves.

I don't think humans as a whole are really equipped to deal with that. And seeing the mental health impact it has had on myself and loved ones makes me question why we want that default global view in the first place. This is why I appreciate the more organic approach Mastodon takes, and also making the user actively curate their view of things (yes, assuming an instance does things correctly, etc). It is overwhelming for users at first, yes, but it is at least not addictive. It also discourages bad actors because they don't get full reach and visibility from the moment they join.

Having reach and discoverability without the problems of the default global view is not something we are remotely close to solving yet.

You shouldn't tell people to move platforms

I'm very much in this camp. There's nothing stopping you from moving official channels to where you want/need them to be, but people will move or stay based on their preferences and desires. I absolutely agree with your statement of "Removing arbitrary communication walls between us is an ethical / human rights / collective action", but I vehemently disagree that the answer to that is consolidation. I don't think you can look at what happened to Twitter and think "we need to consolidate", given other options are also imperfect for various reasons.

I personally think we should be ready to adapt to a more 'fragmented' future, but discussing about which particular fragment to focus on is unproductive in the broader picture, especially given the statement of "Removing arbitrary communication walls between us is an ethical / human rights / collective action". A fragmented or decentralised microblogging ecosystem is much more resilient and robust against bad actors, especially if it is not all contained within one platform. We really should be moving the conversation on how to solve the problem of arbitrary barriers so that the whole discussion of Bluesky vs Mastodon becomes irrelevant.

It may not be Bevy's place to maybe have that discussion, but adding our voice to that collective desire hopefully will move the needle to get enough folks interested in solving that problem.

In the mean time, Bevy can go where it needs to be in terms of official channels. We should just expect that whether users follow us or not isn't something we can force. Plus, if we have time to let things play out, it might not be either Bluesky or Mastodon that emerges as the 'winner', but something new entirely.

[prime31]

I don’t really have a strong opinion either way but I do wonder why not just use software to cross-post to all socials? It seems like a simple solution that would minimize or eliminate any detractors entirely.

I should note my only semi-selfish goal would be to make sure Bevy gets as many eyes on it as possible and not excluding any social would maximize that aspect.

[laundmo]

You shouldn't tell people to move platforms (it is unethical)

There's an important difference between seeing these spaces as "data to run queries on" versus "a community of people." Bluesky seems to prioritize the former, while Mastodon fosters the latter. IRL, when participating in communities, they are inherently limited in size and scale, which allows for more meaningful, personal connections. Online spaces, however, often scale far beyond what we can naturally manage - Dunbar's number shows we can maintain around 50 to 200 meaningful social relationships. Mastodons typically smaller instances fit this better, they represent an approach more like an online town. This smaller scale helps mitigate a lot of issues, for example dogpiling, doomscrolling, or moderating for a specific community. This is undeniably a feature to many, something which I think you are disregarding when trying to convince people to move. You assert that "what the world actually needs from a communication system" is removing arbitrary communication walls, which ignores that quite a few of those walls you're trying to remove are not arbitrary at all, and in fact very important to some people. This is why I find you exerting your influence this way "unpalatable" (or rather, disappointing).

Personally, i never used Twitter/X due to its scale (and post length limit which inherently limits nuance and considered conversation, encouraging outrage and misinformation, but that's another topic), and i'm currently staying away from Bluesky for the same reason.

[Bluefinger]

I think cart's position is understandable from the perspective of doing outreach/marketing to get Bevy more and more recognised and bring in new users. For the project to grow, it kinda needs to be discoverable and Bluesky is just doing this better than Mastodon by comparison.

Fostering community, however, is indeed a different priority than just bringing in new users, and presents a whole set of challenges.

[laundmo]

Well, no matter what the goal is, he's trying to convince people to move platforms, so the effects of that should be considered.

[cart]

which ignores that quite a few of those walls you're trying to remove are not arbitrary at all, and in fact very important to some people. This is why I find you exerting your influence this way "unpalatable" (or rather, disappointing).

I concede that there should have been more nuance in my rally for people to move: I don't think small / insulated / tightly knit Mastodon communities should disappear, and I'm very cool with people sharing and discussing Bevy content in those contexts (or other communities intentionally limiting their scope).

If I was still pushing for aggressive consolidation right now, I would add some sort of qualification like "if your goal is to share your Bevy content with the world, I strongly encourage you to do it on Bluesky. If you want to share it with your insulated slice of the world, continue to do so on your platform of choice".

[zicklag]

I started by replying in a comment and then realized both @laundmo and @Bluefinger had similar questions and I wanted to reply to both of them. This was also longer than I initially intended so I deleted the comment and moved it out here.

---

I think a big part of the issue is that while the things that Mastodon and Bluesky provide are similar and may be used for the same things they are also largely different and people may choose one of the other because of that.

As was mentioned, Mastodon has more of a small feel and things like the longer post size limit allow you to have different kinds of conversations on it that are harder or less common than on Bluesky. These are important differences and, as people, we need to have the ability to use the one that we find useful or feel we need.

The problem is that choosing what we want comes with the caveat of a certain level of "disconnection" from those who have chosen the other option.

What I think is important about Bluesky, and it being built on ATProto, is that I think it gives us a better chance at actually being able to keep our identity while still being able to make that choice between apps, needs, and experiences.

ActivityPub and Mastodon have many problems when it comes to portable identities and inter-operable apps. I personally don't believe that it's on the path to solving those issues in a meaningful way for a variety of reasons.

I think ATProto is doing much more practically in that direction, though admittedly there are still big things that need to be solved and improved.

If someone wanted, for example, they could feasibly create an app for smaller-scale, more meaningfully connected communities, and it could feel very much like Mastodon, or even better, but you could actually have the same identity across Bluesky and this other community app. You could also add inter-op with Bluesky if you wanted, such as allowing you to follow Bevy's Bluesky account, even from your smaller community app. ( To be clear, feeds from other apps is something that can be accomplished in ActivityPub, too. )

The ability to consistently preserve your identity across multiple apps, which is actually important for helping to form more meaningful connections across app boundaries, is something that is actually working in ATProto today, and not in ActivityPub.

---

Related side note, me and my team are actually working on an app, roomy.chat, that is focusing exactly on allowing you to build meaningful Dunbar sized communities ( or larger if you want ), starting with group chat, on ATProto.

We are also building weird.one as a way to help build kind of a "Web Rings 2.0" where people can link to each-other to help build networks of personal websites and find other people in similar niches, and the plan is to integrate it with ATProto, Roomy, and Bluesky.

Bluesky's "big world" social networking is not necessarily what we think is the best way to generally communicate with others or build relationships. But we also think that ATproto is the closest thing to building the foundation that we need for, not just microblogging, but a more user-controlled and user-respecting internet.

So if we are talking about the effect of asking people to move to Bluesky, I think Bluesky is, surprisingly enough, actually doing more overall for promoting a useful, connected, and free internet where we get to choose our experiences, than ActivityPub and Mastodon is.

This is a controversial opinion I'm sure, and ATProto needs to catch up when it comes to the smaller community feel, but there is a lot brewing right now on ATProto some projects like Smoke Signal, which allows people to find, create, and RSVP to events, is looking at more community focused experiences. The point is, it's possible to do and people are actually working on it right now.

It might not be just about the features of Bluesky today that's important, but the effect that it's network and protocol will have in the future. I think Bluesky has risks being that it's at an early stage, but I still give it better chances than any existing alternative today.

---

PS: For what it's worth, we're not totally betting on ATProto or Bluesky either. We're building in a way so that our apps can continue to exist without it because we think the future is still uncertain and there are still things ATProto and Bluesky doesn't do, like private data and offline support, so we are building our own framework to address that. Still, the open identity system of ATProto allows us to integrate even our divergent app with it, which has the above mentioned advantages. And that is, again, a huge win.

PPS: I also acknowledge that me saying we're building an ATProto apps marks me as "biased". My defense is that we actually spent more time with Mastodon and ActivityPub while making our app, before we started to see our values being upheld better in ATProto and that is why we decided to actually build on ATProto.

Our values are not "big world" social media, and still we, as builders looking to promote the freedom and communities of the internet, are finding ATProto as a very possibly better home than ActivityPub despite some uncertainty of the future.

[laundmo]

I'm glad you moved this out of a reply to my post, as while its a interesting technical analysis, my whole point was basically a criticism of the idea of convincing people who happen to use bevy to move to platform A or B. Its basically entirely about a "Its unethical" argument Cart hasn't addressed.

I wish you the best of luck with the projects, but for personal reasons won't ever use them.