Bluesky Report – #110 Video app Skylight is available for public release, with funding from Mark Cuban and 55k users in the first 24 hours. Spark is building their own entire video platform on ATProto, and just launched in beta. Skylight and Spark The main news this week comes from two video apps for Bluesky and ATProto, Skylight and Spark. Skylight is a video client for Bluesky, that has launched to the public this week. In the first 24 hours since launch, the app has gotten 55k users, and hit #2 in the entertainment category on the Apple App Store in the US. Skylight also announced that they’ve gotten a pre-seed funding round from Mark Cuban, as well as another venture fund. The total amount of funding is unknown. Cuban said he’d fund a TikTok alternative on ATProto in January this year, just before TikTok got banned in the US for a day. In 2 days TikTok is set to be banned again in the US. If that were to happen again, Skylight is now ready to welcome people looking for an alternative. The official Bluesky app has also made significant improvements to video in the meantime as well, potentially making the entire network more attractive as a TikTok alternative. Spark is another video app for ATProto, and this week they launched in beta. Spark takes a different approach than Skylight, and is building their own entire platform that does not depend on Bluesky. They are building out their own infrastructure, including their own relay, appview, CDN and more. Spark also uses their own lexicon, allowing them to build their own network on ATProto. On Lexicons Spark and Skylight are taking two different approaches on building a video platform on ATProto, and the core difference is in how they approach ATProto lexicons. This difference is something we will likely be seeing more of on ATProto, so I think it’s worth diving deeper into the choices that both apps are making, and the tradeoffs that the choices entail. A short and simplified explainer on lexicons: lexicons are a type of file structure for social data, and it says how data should be formatted. Every lexicon has an owner, usually the organisation who is building the app who uses the lexicon. It serves two purposes: it defines what an app can and cannot do, and it allows for interoperability between different apps. For example, the lexicon for a Bluesky post defines that a post should have a maximum length of 300 characters. Because this file structure is openly available, anyone else can also publish a Bluesky post. A lexicon says that if that post that is not made by Bluesky, but does follow the correct Bluesky lexicon, the post will show up in the Bluesky app. If it does not follow the lexicon correctly, it will not be visible. Bluesky CTO Paul Frazee wrote some guidelines on lexicons this week, which says that you should use existing lexicons when you intend to interoperate with other apps, and if you don’t intend to interoperate you should create your own lexicon. Skylight takes the first approach: the goal is to interoperate with Bluesky, and thus the app uses Bluesky lexicons. Effectively, Skylight is a client for Bluesky that focuses on only a portion of Bluesky posts, namely videos. This has some advantages for Skylight: Skylight does not have to do moderation. Every video posted via Skylight goes through Bluesky’s usual moderation infrastructure. Skylight inherits the 33M+ user base of Bluesky. When you open up Skylight, there is immediately a ton of videos to see. Skylight is not limited to only using Bluesky’s lexicon, and by integrating multiple types of lexicons Skylight can create an app that offers something different than just using the Bluesky app for video. Skylight CEO Tori White says that they are working together with stream.place to bring live streaming to the app. Stream.place is a live streaming platform with ATProto integration. There are also downsides though: Skylight is fully dependent on Bluesky, for both moderation as well as operation. Skylight cannot change the limitations that Bluesky sets. Bluesky has set the length over videos at 3 minutes. This used to be one minute. Skylight has no control over these settings or when they are changed. Bluesky PBC could set the video duration limit to 10 minutes if they wanted to, without Skylight having any say in the matter. It is harder for Skylight to define its own culture. For example, US politics is one of the most dominant parts of the conversation on Bluesky. If someone is currently using Bluesky to mainly talk about US politics, they might be less interested in also having some silly meme videos get posted to their Bluesky profile. Skylight is already tweaking their algorithm to get more variation in their feeds due to the prominence of US politics videos. Spark uses its own lexicon and infrastructure, and its upsides and downsides are mirrored to that of Skylight: Spark will have to do their own moderation for videos uploaded to Spark. Spark’s ecosystem for videos posted with the Spark lexicon will have to start from scratch. There are also upsides: Spark can add features to videos that Bluesky does not have, Spark has talked about adding support to easily add music and other audio to your videos. Spark’s videos can also be of a higher quality (300mb vs 100mb) compared to Bluesky. Spark gets to create their own community and culture that is distinct from Bluesky’s culture. Spark is fully independent from Bluesky. Spark’s independence from Bluesky’s moderation is due to their choice in using their own lexicon, while the company also has made the choice to be fully independent in infrastructure: running their own relay, PDS, CDN and more. The reason for spelling out the difference between these two approaches in detail is because Spark and Skylight will likely be the first of many more organisations that will have to make this choice between using other organisation’s lexicons and creating their own lexicons. There is no singular answer on which choice is better, and each comes with tradeoffs. Image-focused apps are another case where developers will make similar considerations about which lexicons they will be using. Flashes is a Bluesky client with an Instagram-like interface, that went for the first option, using Bluesky lexicons for posts. There has not been a major push yet towards an Instagram-like app on ATProto that uses its own lexicons and stays more separate from Bluesky. When people are considering building an image-focused app on ATProto, understanding the differences between Spark and Skylight might help with the tradeoffs and impact that come with each choice.In Other News Altmetric has posted a blog with more details on their finding that Bluesky now has more posts on academic research published in 2025 than X has. What stands out to me is that Bluesky has a considerably larger ratio of original posts discussing research, compared to X. In contrast, X’s posts on research trend much more towards reposts. Altmetric also reports a finding I have not seen anywhere else yet, namely that Bluesky tends to be more active on weekdays, whereas X tends to be busier during weekends. Bluesky PBC is growing; the company currently has around 20 employees and is hiring for 6 new positions. The company recently hired a product designer, and is now looking for a Head of Product as well. Custom feed builder Graze has released a tool, Contrails, to turn their custom feeds into a developer-friendly structure that allows the feed to be used for other purposes than reading the posts via the Bluesky app. For example, when building a bot that listens to certain keywords and then take an action, bots developers usually listen to the firehose of the entire network. With Contrails, Graze can do the filtering, and developers can use this filtered stream of output to build their own toolings, such as bots, dashboards or moderation systems. A sprawling blog post by Blacksky founder Rudy Fraser, about building an internet of many autonomous communities. There are many ideas in the post, and I can recommend reading it in full. I want to highlight a small section, where Fraser lists some things that are needed for building communities: Treat “communities” as first class citizens (like “users” or “customers”). This is like how labelers are a special kind of account. Allow people to build things specifically for community account types. Trust that individuals will show up to shoulder the burden of forming and maintaining these communities (incl. but not limited to performing moderation responsibilities – some of us are weird like that). This means as new features are designed, you don’t need to ask “but who will do that”? Probably me tbqh. At best provide a non-extractive/non-taxing monetization path (DID2DID payments?) and at least don’t get in the way of communities forming reciprocal or gift economies. Treating communities as a specific type of entity within the ATProto network makes sense to me. Both Blacksky and Northsky are creating spaces within ATProto and Bluesky. Blacksky is already successful at creating a vibrant community, and Northsky looks to do the same as well. Other communities are likely to follow suit to build their own places on ATProto. This might just expand beyond Bluesky-type communities, see for example how video apps like Spark have the potential to build a video-first community that is apart from Bluesky and microblogging, while also being connected to it. One of the core ideas of the fediverse is that make up of an interconnected network of many independent communities. While the vision of interconnected communities is one of the most appealing part of the fediverse to me, it struggles to make the switch from ‘connected servers’ to ‘connected communities’. The Links In the media: New Paradigms in Trust and Safety: Navigating Defederation on Decentralized Social Media Platforms – Carnegie Endowment for International Peace. Can Bluesky’s AT Protocol build the decentralized social media ecosystem the Fediverse aspires to? – Joshua Benton/NiemanLab Blue Skies Ahead: Social Media’s Quiet User Revolution – Damion Taylor/Forbes Beyond Bluesky: These are the apps building social experiences on the AT Protocol – Sarah Perez/TechCrunch And some more links: Hose race is a silly race game that allows you race words against each other directly from the Bluesky firehose (accompanying blog post here). A writeup of the recent ATmosphere Conference in Seattle by one of the organisers. That’s all for this week, thanks for reading! If you want more analysis, you can subscribe to my newsletter. Every week you get an update with all this week’s articles, as well as extra analysis not published anywhere else. You can subscribe below, and follow this blog @fediversereport.com and my personal account @laurenshof.online on Bluesky. #bluesky image
Fediverse Report – #110 A vulnerability in Pixelfed caused private posts from other platforms to leak, a post-mortem on the CSAM scanner from IFTAS, and Fediforum has been cancelled.Pixelfed vulnerability impacts private posts across most of the fediverse The fediverse suffered from a significant breach for private accounts, that affects the large majority of fediverse servers, due to a vulnerability in the Pixelfed software. What is notable about the situation is that the software vulnerability is in Pixelfed, but the affected accounts are not exclusive to Pixelfed: accounts on Mastodon and other fediverse software with a form of private accounts are also vulnerable. The vulnerability was found by the independent developer Fiona, who wrote a blog post about the vulnerability and the disclosure process. To understand the situation, a short explanation of two features of Mastodon and some other fediverse microblogging software, locked accounts, and follower-only posts. Together these two features make it possible to have a form of private accounts. Locked accounts means that you cannot automatically follow that account, it has to be approved instead. Follower-only posts means that the post will only be displayed to your followers. When a locked account approves a follower, follower-only posts now get send to the server that this follower is on. Because the receiving server now has this follower-only post in their database, they need to correctly handle whom they show this post to and whom they do not. If another account on the other server also tries to follow the locked account, but the locked account does not approve, this third account should not be able to see the messages. This is where Pixelfed’s vulnerability comes in: Pixelfed was not waiting for a confirmation if a follow request was approved, it assumed that it was automatically approved. That is how any private posts made on (almost) any fediverse server could be leaked: if a Pixelfed server already had the private post (because of someone of Pixelfed followed the locked account with approval), it would show it to anyone else who also tried to follow the locked account, even if the locked account rejected the follow request. Pixelfed’s vulnerability points to deeper issues with the fediverse, activitypub and private posts. If all it takes to leak private messages is another server to be misconfigured, than it indicates the huge security risk inherent in private posts via ActivityPub. Even more so considering that the network incentivises and encourages people to build their own software implementations, which increases the risk of security vulnerability and other misconfigurations significantly. For simplicity I’ll focus here on Mastodon, although it also goes for other microblogging fediverse software that offers a combination of follower-only posts and locked accounts. At its core, private posts via ActivityPub requires to trust other servers. This is how ActivityPub works: your server sends posts to another server. There is no way to enforce that this other server respects your preference on how they should handle this post. If you do not trust another server to handle your data properly, the only way to deal with that is by not sending your post to that server. When you make a follower-only post on Mastodon, the UI prompt warns you that followers-only posts without setting your account to locked allows anyone to view your posts by simply following you. The documentation for Mastodon also reinforces this, saying: “To effectively publish private (followers-only) posts, you must lock your account–otherwise, anyone could follow you to view older posts.” The documentation makes it clear that Mastodon views the combination of follower-only posts with a locked account as private posts. But nowhere is it made clear that these posts being private depends on other servers being good actors and not having an error in their code. So using private posts on Mastodon comes with the risk of the private posts being leaked due to flaws in other software, without people being aware of this risk. Once a leak like this one happens, it is unclear who is responsible for communications with affected users. It was a flaw in Pixelfed that caused the vulnerability, but it is other people on other fediverse servers that are affected. Pixelfed developer Daniel Supernault has only made minimal announcements, urging Pixelfed admins to upgrade, without further explanation to the people who are actually affected by the vulnerability. Personally I think Supernault should have handled communications significantly better. But it is the thousands of fediverse server admins who provide the actual social networking to people on their server. They are the ones who are offering a social networking site with a variety of features, including the ability to make private posts (as advertised by the Mastodon software), and are the ones who are responsible for handling the data of their users. I could only find one example of a server admin that has informed their users of the situation, even though it is the data of their users that is affected. I’m unclear if this is because the admins are not aware of what’s going on, or the admins view it as the responsibility of someone else to inform people that data they thought was private might potentially have been leaked. Overall, it means that there actually three separate problems going on at the same time: The first problem is that Pixelfed had a vulnerability which leaked private data from people on other platforms. The second problem is that software like Mastodon and others promise private posts, without explaining what the risks are of using private posts, and that this depends on other servers behaving correctly. The Pixelfed vulnerability shows that these concerns are not theoretical or minor, but can happen to one of the biggest fediverse software/server. The third problem is that when private data gets leaked, most fediverse server admins do not inform the people on their platform that they might have been affected by this. It is still unclear to what the direct impact is of the Pixelfed vulnerability, and how many people’s private post have been accessed by others, and it’s unsure if that will ever be answered. But it is the indirect impact of the situation that I’m most interested in: will this change how people perceive private posts, and will it fediverse server admins take a clear position on when they should inform their users, and when the should not?IFTAS’s post-mortem on their Content Classification System IFTAS, the Independent Federated Trust And Safety organisation, has released a post mortem on their content classification system (CCS). The CCS project was a pilot project to detect and report CSAM for a small group of Mastodon servers, and lasted for half a year. The pilot was shut down after IFTAS did not manage to find the funding they were looking for, and the organisation had to shut down most of their projects this month. CCS operated on 8 servers, which combined have around 30k monthly active users, and IFTAS found a total of 80 matches, averaging 4.29 matches per 100,000 media files. IFTAS writes: “4.29 matches per 100,000 may not sound like a large number. However, to be clear, this is a higher number than many services would expect to see, and it includes a broad range of media, from “barely legal” minors posted publicly, to intimate imagery shared without consent, to the very, very worst media imaginable. In some cases, it was apparent that users were creating accounts on host services to transact or pre-sale media before moving to an encrypted platform, under the belief that Mastodon would not be able to detect the activity.” The results show that there is a clear need for proper CSAM scanning and reporting services for the fediverse, and that IFTAS does not have the funding to provide such a service is a significant loss to the network. On a note related to IFTAS’s funding: Erin Kissane gave a talk at the AT Protocol conference recently, in which she talked about ‘vernacular institutions’. She described vernacular institutions as emergent and local organisations, which solve practical needs on the ground. Kissane describes vernacular institutions as ‘more useful than legible’. She then mentions IFTAS as a clear example; it provides a need for local communities (as illustrated by the CCS project), but its illegibility made it hard for funding organisations to understand what IFTAS was doing and provide them with the funding they need.Fediforum has been cancelled Fediforum has been cancelled, to be rescheduled at a later date. The unconference about the fediverse and the open social web was scheduled for today and tomorrow, April 1-2. This was supposed to be the 5th edition of Fediforum, which consists of speed demos and sessions that anyone can run on any topic. Fediforum is organised by Johannes Erst, with Kaliya ‘IdentityWoman’ Young as the co-organiser. Transphobic tweets by Young had surfaced in the days leading up to the event, and various prominent community members announced that they were either withdrawing themselves from the event, or said that they personally would not want to go to the event. Ernst then announced on his personal account that Young would be “transitioning out of Fediforum”. A day later (March 31), the official Fediforum account confirmed that Young would no longer be involved. At this point, community trust in Ernst was damaged and the discourse had reached a harmful stage, and Ernst decided to cancel the unconference and reschedule it to a later date. WeDistribute has a more extensive writeup of the situation here. An unconference like Fediforum depends to a large extent on community trust and good intentions, and it was clear that the vibe was not great for constructive conversation at the point that Ernst decided to postpone the event altogether. Still, Fediforum provided a great place for fediverse projects to do some promotion with the speed demos, and Fediforum said that they even had a waiting list for this edition. There is a clear demand for an (un)conference like Fediforum, but the fediverse has not managed to create other community events that allow people to showcase their fediverse project in the last few years, besides Fediforum itself.The Links Mastodon is hiring a Senior Product Designer. Independent fediverse developer Emelia Smith wrote two articles this week, one on the ‘Open-source tools needed for the future of decentralized moderation’, as well as on how ‘Federation on the fediverse doesn’t have to be a binary choice between allowing everything or needing to pre-approve your entire network. The Newsmast Foundation is taking over the administration of the indieweb.social server Funkwhale has released a first alpha version of Funkwhale 2.0. This week’s fediverse software updates. PeerTube: the Fediverse’s decentralized video platform (part 2: creator edition) – Elena Rossini The Lemmy developers held an AMA this week. I didn’t get into covering their responses this week, that will happen next week. The entire AMA can be found here. That’s all for this week, thanks for reading! You can subscribe to my newsletter to get all my weekly updates via email, which gets you some interesting extra analysis as a bonus, that is not posted here on the website. You can subscribe below: #fediverse image
ATmosphere Report – #109 The ATmosphereConf was last weekend, independent relays are starting to appear, and more.Conference This weekend was the first ATProto conference, the ATmosphereConf, in Seattle. Over two days there were a large number of speakers and sessions, with over 150 people in attendance, and a significant number watching the live streams as well. I could not make it to the US, so for a full overview of the event, I recommend this extensive article by TechCrunch’ Sarah Perez, who was present at the event. The entire event was livestreamed, and all talks can be viewed via this YouTube playlist. Some assortment of thoughts I had while watching the livestream and VODs over the last few days: Bluesky CEO Jay Graber gave a short speech, about her background as a digital rights activist, and how she is now “holding the door open, so people can see another world is possible”. Graber is clearly aware of her position, where she is seen as a figurehead of the network, while also wanting to build a decentralised network where there is place for competing platforms. Being a figurehead of a network, without becoming the de facto leader of the network, while also holding the leadership position of by far the largest organisation in the network, is a challenging position to balance. Bluesky CTO Paul Frazee talked about where Bluesky came from and where it is going. One of the things he talked about is the consideration of why Bluesky decided on their own protocol and not ActivityPub. His answer focuses on practical considerations, especially how ActivityPub handles identity and account migration. Watching the ATProto Ethos talk by Bluesky protocol engineer Daniel Holmgren it struck me that the question could also be framed as a matter of lineage. Holmgren talks about how ATProto takes inspiration from the Web, Peer to Peer systems as well as Distributed Systems. Placing it in such a context makes it clear that ATProto has quite a different background and other ways of thinking than ActivityPub has. Ændra Rhinisland talked about how community projects can become load-bearing for the network, without adequate support structures for the people who run such projects. She also runs the popular news feeds using Graze. Graze has been adding support for advertisements, and Ændra is one of the first to take advantage. In her talk she walked through how at current usage rates, the feeds could generate over $20k per month in ad revenue. She plans to use this revenue to support the queer communities building on ATProto, and showed early plans for a self-sustaining fund powered by Graze’s feed revenue, to support initiatives such as Northsky. The talk by Ms Boba is a great indication of how much under-explored design space there is on Bluesky and ATProto. Her talk focuses on labelers and fandom communities, and has some great examples of how they can be used outside of moderation. Blacksky founder Rudy Fraser gave an excellent talk, describing Bluesky as a skeuomorphism, meaning that it imitates the design of the product it’s replacing. This phase is a part of the adoption cycle for new technologies, but Fraser does not to stop at imitation but instead explore the new ways that communities can be build online. Fraser is specifically interested in building platforms that can serve mid-sized communities, ranging from hundreds of thousands to a few million people. The Blacksky community is an example of this, and Fraser hopes that Blacksky can inspire other communities to do the same. His framing of content moderation as community care and not a cost of business also resonated with me. Erin Kissane’s talk goes into detail about vernacular institutions, local and grassroots organisations and practices that are often illegible to outsiders but deeply embedded in local communities. This allows them to be close to the needs of their community members, but makes them hard to see and understand from the outside. This outside illegibility is a double-edged sword: IFTAS served a crucial role for trust and safety in the fediverse ecosystem, but had to shut down to a lack of funding as a result of being illegible to financiers. Some more articles on the events: The under-the-radar tech revolution that could change how the internet works – Marcelo Calbucci/GeekWire Things That Caught My Attention – Dan Hon What’s next for ATProto, the protocol powering Bluesky and other apps – Sarah Perez/TechCrunchOn relays Bluesky PBC has been working on a new version of the relay that makes it easier and cheaper to host, under the Sync 1.1 proposal. This new version is now starting to roll out, showing a significant drop in resource usage. Bluesky engineer Bryan Newbold shared some statistics here. Independent ATProto developer @futur.blue set up his own relay as a speedrun. He shows that a full network relay can be run on a 50USD Raspberry Pi, with an easy-to-follow tutorial here. That full network ATProto relays are cheap to run has been known for a while within the ATProto developer community, but that knowledge has not spread much yet. One reason for this is that independent developers have set up relays primarily for their own use, sharing access with a few friends, but no other publicly accessible full-network relays exist yet1. Upcoming short-form video platform Spark is building their own complete infrastructure. Spark’s relay will publicly accessible, and hosted in Brazil. Having ATProto infrastructure outside of US jurisdiction is a conversation that has come up regularly, and often followed by the assumption that the alternative is to have infrastructure like a relay hosted in Europe. Spark is bringing in a slightly unexpected twist here, by having the first publicly accessible relay that is not owned by Bluesky PBC being hosted in Brazil instead. Having other relays that are not owned by Bluesky PBC has been the subject of a lot of conversation, and the Free Our Feeds campaign was founded on the idea that a significant financial investment is needed to do so. Furthermore, it assumes that such a relay is not only expensive, but that it requires an extensive governance infrastructure to manage it. The current developments regarding relays call both of these assumptions into serious question: relays are cheap, not expensive. Furthermore it seems that there is enough incentive that organisations that are serious about building their own ATProto platforms are willing to run their own relays.In Other News Bluesky PBC has published a proposal on how they want to handle OAuth Scopes. OAuth Scopes is one of the main projects on the roadmap for the first half of this year. Currently, logging into an ATProto app via OAuth requires you to give that app permission to access all the data for your account. OAuth Scopes allows an app to only ask for the permissions that are necessary, and not the entire account. There are two problems that need to solve: the technical part of making it work, as well as the handling the UX to communicate clearly to people what data an app wants to access. The challenging part of the UX is how to handle the translation from the technical description of the data that is requested (stylised like ‘app.bsky.feed.getFeed’, for example), into a way that is understandable for the everyday user. The second challenge is that apps require permission not for one, but for many types of this lexicon data. A third-party Bluesky client that is restricted to only Bluesky data will still have to request a dozen of these Lexicons. A long list of technical lexicon names makes it impossible for regular people to have an informed opinion on what data is and is not being accessed. Bluesky PBC’s proposal is to group different lexicons into bundles, and create new lexicons that reference these bundles. Scoped OAuth can then request access to a bundle of lexicons, with a description that is legible for regular people. Git repository platform Tangled is working on news ideas how a GitHub alternative might do things differently, and one of their first proposals is defining two types of pull requests. For another look at Tangled, this blog post experiment with what the platform allows. One of the talks at the ATmosphereConf was by independent developer Rashid Aziz, who is the co-founder of basic.tech. Basic is a protocol for user-owned data, and seems to be fairly comparable to the PDS part of ATProto, with the major difference that Basic allows for private data on their version of a PDS. Aziz used the combination of these two protocols to create private bookmarks for Bluesky. The new Record Collector labeler automatically displays if someone has been using other apps in the ATmosphere outside of Bluesky. Rocksky is a new music scrobbler service on ATProto, that is currently in closed beta testing. It allows people to connect their Spotify account and automatically ‘scrobble’ (track) the music they are listening to.The Links Some tech-focused links for ATProto: ATCryptography is a package with cryptographic utilities for the ATProto, written in Swift. A blog post by independent ATProto developer Mary about ‘AT Protocol, OAuth, and well, decentralization’. Mary also published a repository car file explorer this week. A Python API library for Bluesky. Independent ATProto developer Kuba Suder writes about database optimalisation and speeding up the process of firehose processing. ‘this website is hosted on bluesky (for real this time)’ is a blog post about hosting your website data on an ATProto PDS, with a ‘thin AppView layer’ for easier to parse URLs. That’s all for this week, thanks for reading! If you want more analysis, you can subscribe to my newsletter. Every week you get an update with all this week’s articles, as well as extra analysis not published anywhere else. You can subscribe below, and follow this blog @fediversereport.com and my personal account @laurenshof.online on Bluesky.Cerulea.blue is a publicly accessible relay, using a custom implementation, but it is limited to non-Bluesky PDSes. ↩︎ #bluesky image
Fediverse Report – #109 An essay on user preferences, and how the fediverse’s interconnected network of communities can play into that, as well as some other news.User Intents The Bluesky Company (Bluesky PBC) recently announced a proposal to add User Intents to the AT Protocol (ATProto). The proposal allows people to set account-wide preferences how their data should be handled outside the network. It gives people the ability to opt in or opt out their account from a few different things, such as bridging to other protocols or not wanting any of their data being used in generative AI datasets. The proposal is similar to how robots.txt works, meaning that it is a machine-readable format which good actors are supposed to abide by, but is not legally enforceable. I cover both the fediverse and Bluesky (including ATProto) under Fediverse Report because these networks are deeply interconnected and influence each other. Decisions on one network, like Bluesky’s User Intents proposal, can influence how the fediverse develops and builds their own features. My goal is to help readers understand the fediverse more deeply. By observing how Bluesky’s approaches default user preferences, the fediverse can build their own systems that use its strength of having many diverse and connected communities. The proposal by Bluesky PBC is as follows: People are able to set their preferences for four different categories: generative AI protocol bridging bulk datasets public archiving and preservation These preferences are account wide. They are valid not only for Bluesky, but for every app build on ATProto. The default value is ‘undefined’, not opt-in or opt-out. Projects which are intending to use the public data should decide for themselves whether data reuse when the intents are classified as “undefined” is acceptable or not. the current proposal is set to lead the way for more granular user preferences, allowing people to specify on an app-level or post-level what their preference is. Also, some concepts of ATProto that are relevant, which makes the protocol different from ActivityPub: On ATProto, a user has only 1 account, and can use that account to log into any service. This is in contrast with ActivityPub, where you need a new account for every service. Data on ATProto is public by default, and designed to be accessible. Everyone has full and free access to the data of the entire network. One thing about user preference settings in social apps is that they are a bit of red herring. The majority of people never change the default settings. Giving people choice is a good thing, but it is impossible force people to choose: the majority of people will just not choose anything. This makes it so that the default value for any preference is hugely important, as it is the de-facto value that the majority of people will experience. Bluesky PBC tries to avoid this issue by introducing a default “undefined” value. The advantage of using a default value of “undefined” is that Bluesky PBC will not overstep their boundaries and determine the preference of everyone on the network, including people who are not using Bluesky but are using other platforms on the network. The downside is that Bluesky PBC effectively makes no decision at all for the majority of people. Bluesky PBC leaves it to the organisations who use the data to determine how data can be handled if the preference is set to “undefined”. These organisations are likely to value their own interests more than the interest of people whose data they intent to process. Bluesky PBC has three options here, that all have a downside: If Bluesky PBC sets default values for how ATProto account data can be handled it reinforces its centralising role in the network. If Bluesky PBC does not set a default value, no decision is made for the majority of people, and it is left to organisations whose goals do not align with those of the people whose data they process. If Bluesky PBC sets User Intent not on an ATProto-account wide level, but only on an per-app basis, choices quickly become overwhelming if users must set preferences for every app. So far I’ve only been talking about Bluesky and ATProto. But the fediverse has a long history of debates, conversations and drama on how to deal with data processing that happens outside of the network. Some high-profile cases include the blowup around Bridgy Fed considering making the bridge between the fediverse and Bluesky opt-out, or the backlash against Searchtodon, which saved user’s timeline locally for searching. These debates are around data scraping, consent, things being opt-in or opt-out. But one of the struggles that the fediverse has had is to build structural solutions. A significant portion of the fediverse does not consent to have their data handled outside of the network. A persistent problem is that this preference is not expressed in a machine-readable way. This leads to an endless cycle of new developers coming in that are not familiar with the culture who then cross lines of consent and it all blows up in drama again. Moreover, the fediverse and ActivityPub have a significant advantage on how to deal with the dilemma of setting default values over ATProto. The fediverse is a network that is build up of many different communities connecting with each other. A variety of communities allows for diverse preferences, which can also be expressed in setting default values. And it is a shame that the fediverse is not capitalising on this advantage. There are communities from whom discoverability is important. Just as there are communities for whom not being easily publicly discoverable is important. These preference can differ within an individual as well: people treat personal photos shared with friends differently from blog articles. The fediverse can sidestep the question of default account values because people have many accounts on the fediverse, for different use cases. This gives the option to set a different default value for different services. A Pixelfed platform for close friends should set stricter default data-handling preferences. A Mastodon server for blogging platform Medium that has the goal of giving more visibility and reach to its writers could consider setting default values to be more open. The power of the fediverse is in that there does not have to be a single default at all. Instead, communities and servers should be able to set default values for themselves. This can help shape the tone of the community, and makes it clear what the identity of a community is about. What’s even more powerful is that this only concerns the default value, giving people the ability to set their preferences as they desire. The state of the open social web is such that there are now two protocols in competition with each other. That gives the ability for the fediverse to take ideas from other networks, and improve on them in a way that plays up to the unique strengths that the fediverse has.The News Reminder: next week will be FediForum, on April 1-2, and you can register here. FediverseSharing: A Novel Dataset on Cross-Platform Interaction Dynamics between Threads and Mastodon Users is a new academic paper (currently under review and up on arXiv) that explores the interaction between Threads users and Mastodon users. It takes a dataset of 20k Threads users that have fediverse sharing enabled and compares it to 20k Mastodon users that have interacted with these Threads users. The main goal of the research is to build up this dataset and share it with the community for further research. How sharing a dataset of aggregated user interactions relates to the above essay on user preferences for being included in bulk datasets is left as an exercise to the reader. PeerTube has done a major redesign for their v7 of the software that came out a few months ago. The organisation now shared the design and development reports that shaped the update. IFTAS recently had to shut down most of their larger projects due to a lack of funding. One of their projects, FediCheck is now available as open source for someone else to continue with. FediCheck is a deny list management tool that allows server admins to subscribe to external deny lists. The Lemmy developers will hold an AMA on Wednesday March 26th. Last week, Ghost made their ActivityPub integration available in public beta for Ghost Pro subscribers. Their weekly update says that now over 250 sites already use the integration. WeDistribute has a hands on with the new features that Ghost offers. Note: Last week I wrote about the new fediverse platform Forte, and said that the repository did not include an install guide. This is incorrect, the guide can be found here.The Links Website League and the Rise of Island Networks – Sean Tilley/WeDistribute. The fediverse has a long tradition of building silly clients for Mastodon. This article has an overview of some of them. Two new video tutorials by FediHost, for setting up a GoToSocial instance and configuring a PeerTube instance. An update on how search works in music sharing platform Bandwagon. A development update for Letterbook, an upcoming fediverse microblogging platform. This weeks’ fediverse software updates. Fediverse Events hackaton project. That’s all for this week, thanks for reading! You can subscribe to my newsletter to get all my weekly updates via email, which gets you some interesting extra analysis as a bonus, that is not posted here on the website. You can subscribe below: #fediverse image
Bluesky Report – #108 Northsky is a new cooperative that is building their own space for the trans and queer community on Bluesky/ATProto, multiple apps are starting to work towards financial sustainability, and more.The NewsNorthsky announced Northsky Social is a newly announced cooperative that is “working to build a digital space designed around active moderation and user safety for 2SLGBTQIA+ communities.” The Canadian cooperative features a group of people from Bluesky’s 2SLGBTQIA+ community that have decided to build their own place on ATProto, serving the trans and queer community. The first part of the Northsky project focuses on data hosting. Northsky is providing their users with managed PDS hosting. Users can transfer their data from a Bluesky-managed PDS to a Northsky-managed PDS. Northsky focuses on all their infrastructure being hosted in Canada, outside of the US. The anti-trans climate in the US is given as the main reason why Northsky is first going with PDS hosting. Even though it does not contribute to Northsky’s goals towards better moderation, it is something that Northsky can help their community with right now. Northsky is building out their own custom tooling to help people with this transfer. The current tools for transferring an account to a different PDS require technical know-how, and Northsky is building tools to make it accessible for the layperson. Northsky is also making their customers aware of, and helping them, with registering recovery keys. Setting and storing a recovery key allows people to always gain control of their account using that key. This is one of the features of ATProto that has been hidden so far, and also currently requires technical knowledge to set up. One of the challenges for Northsky is that Bluesky PBC currently does not allow accounts to migrate back to their PDSes. This means that if an account starts using Northsky, they’ll permanently have to either use Northsky or find another PDS provider service. Bluesky PBC seems open to changing this however. In their announcement, Northsky talks about building safer digital places, and addressing “the gaps we see in moderation by implementing tooling and processes focused on keeping our users safe.” These plans cannot be met with just PDS hosting, and need moderation systems, and Northsky’s own appview as well. Northsky says that this will be phase 2 of their project, with more details to be shared later. The social dynamics of moderation are much more difficult than providing hosting services, so the real challenge for Northsky to build safer places will come later.In Other News This weekend will be the ATmosphere conference in Seattle. The event is sold out, but there will be remote livestreaming available. The conference has a busy schedule with two full days of talks. For Europeans there is another ATProto conference upcoming: AHOY! This single-day event will be held on April 24th in Hamburg, Germany, and tickets are now available. The first two announced speakers are the developers of Skyfeed and Tangled, and more speakers are already scheduled to be announced soon as well. As a note: I’m involved with helping organise the AHOY! conference, and hope to see many of you there! Multiple apps in the Bluesky/ATProto ecosystem are taking the next step towards professionalising and are working towards financial sustainability. Spark is an upcoming ATProto short-form video platform, and they launched a fundraiser this week. The organisation of four people incorporated into a Public Benefit Company (PBC) recently, and are now looking to raise 100K USD via a crowdfunding campaign. Bluesky video client Skylight is also now incorporated as a PBC. The creator of Bluesky client deck.blue announced they’re working full-time on deck.blue, and can financially sustain themselves on the Patreon subscriptions. Bluesky video client Flashes is launching a fundraiser for bringing the app to Android, and will include an optional $1.99/month subscription to keep the apps financially sustainable. The most recent update to the official Bluesky app contains a separate inbox for DMs and 3-minute videos. Bluesky has long had an issue with DM spam, although the team has been working to reduce this issue. The new DM inbox contains chat requests from unknown accounts where they can be accepted or rejected. This makes spam DMs less intrusive. The other update is that longer videos are now allowed on Bluesky, from 1 minute to 3 minutes, and from 50mb per video to 100mb per video. Notably, the option for longer videos put Bluesky even more in direct competition with Spark, an upcoming video platform on ATProto: Spark advertises its higher limits of 3 minute videos of 300mb as a unique feature, and of the reasons why they went with its own lexicon instead of using Bluesky. Altmetric is a platform that tracks online engagement on scientific research across social networks. They observe that Bluesky has become a major platform for sharing scientific publications. Publications from 2025 are shared more on Bluesky than on X, and that overall volume of conversation about research is getting close to equal in absolute terms to X as well, even though Bluesky has an order of magnitude less active users than X does. Bluesky PBC is pivoting into a t-shirt company. Last week CEO Jay Graber wore a tshirt to the SXSW conference with the text “Mundus sine caesaribus” meaning “A world without Caesars”. The Bluesky community demanded the shirt as merch, and the first batch sold out in 30 minutes. A few days later the company brought the shirts back into stock, and COO Rose Wang said that the company had more money coming in from the shirts in a single day as they did from two years of selling custom domains. Bluesky PBC has set up an EU representative office in Belgium, as part of mandatory compliance with the EU’s Digital Services Act (DSA). Bluesky PBC made the news last November when a spokesperson for the European Commission called out Bluesky PBC for not following some of the compliance requirements for the DSA. The DSA mandates that all platforms that are active in the EU register their EU headquarters, and now Bluesky PBC has a point of contact in Belgium for the specific purpose of contact with EU Member states and agencies. Fast Company ranked Bluesky PBC at number 17 on its list of the World’s 50 Most Innovative Companies for 2025. Alongside the ranking, they featured an interview with Bluesky CEO Jay Graber. Timeline app Surf is an upcoming app to build and browse custom feeds across different protocols and platforms. It allows you to browse content from platforms such as the fediverse, Bluesky, Threads, YouTube and more, customised into whatever topic or feed you want. Surf’s latest updates allow for significantly better integration with Bluesky: you can now log into the app with your Bluesky account, have a unified home timeline for both your Bluesky and Mastodon account, and engage and post from the Surf app onto Bluesky. The app is build by Flipboard, and is currently in private beta. And some Bluesky/ATProto software that caught my eyes: Wamellow is a third-party tool for Discord that allows for a variety of social media services to be further integrated into a Discord server. Their latest update to the Discord integration allows people to like a Bluesky post that has been posted on Discord by reacting with the 🩵 emoji within Discord. Bluesky video client Skylight now has an in-app video editor. One of the challenges facing Bluesky media client apps is in how they’ll distinguish themselves from the official Bluesky apps. Having an in-app video editor is such a feature, where Skylight now can provide a meaningful value add over using the Bluesky app to watch videos. ATProto.im is a client for Bluesky DMs, built in the style of the old AIM chatting platform. Dragonfly is a new Bluesky client for MacOS desktop and the iPad.The Links Statistics for the different languages used on Bluesky. A sticker pack for Signal with logos and images of various ATProto projects. Bluesky By The Second is a dashboard for showing realtime statistics of the firehose. Feed generator platform Graze did interviews with the creators of the Medsky feed and the Art Feed Collection feed in their newsletter. A Bluesky For Dummies book will come out soon. That’s all for this week, thanks for reading! If you want more analysis, you can subscribe to my newsletter. Every week you get an update with all this week’s articles, as well as extra analysis not published anywhere else. You can subscribe below, and follow this blog @fediversereport.com and my personal account @laurenshof.online on Bluesky. #bluesky image
Fediverse Report #108 Newsletter publisher Ghost is now connecting to the fediverse in public beta, updates about the bridge that connects the fediverse with Bluesky, and more.The News The public beta for connecting Ghost to the fediverse is here, and the ActivityPub integration is now available for Ghost Pro subscribers. Ghost is a publishing platform for sending out blogs via email. With this latest update, Ghost now has another method of distribution, namely via the fediverse. Ghost’s integration with the fediverse consists of two parts: sending out long-form articles published on Ghost into the fediverse, and a reader app to the fediverse from Ghost. Publishing Ghost articles on ActivityPub makes them accessible to the rest of the fediverse, similar to how WordPress with the ActivityPub plugin works. For users of Ghost this is an easy sales pitch, it is simply another free and automatic distribution channel for their blog. The second part of Ghost’s integration with the social web is a reader app. This app allows Ghost users to browse and read posts on the fediverse. It is split up into two parts: an inbox for reading other long-form posts from Ghost or WordPress, and a feed for all other types of posts. This allows accounts on Ghost not only to send out posts via the ActivityPub integration, but also to connect, respond and follow their audience. It even allows you to post short-form microblogs (notes), just like you’d use on Mastodon, that do not show up on the Ghost website. This makes the Ghost integration a full fediverse experience. A New Social is the non-profit organisation that builds and manages cross-protocol tools for the open social web. The organisation currently manages Bridgy Fed, the connector that allows accounts to ‘bridge’ between both ActivityPub, ATProto, Nostr and more, and is currently in the process of setting up and launching the organisation. In their first update they shared this week, A New Social shared that they have a board of directors, consisting of Erin Kissane, Ben Werdmuller and Susan Mernit. Bridgy Fed Config is the first upcoming launch that they announced, scheduled for early April. To bridge their account, Bridgy Fed currently requires people to follow the Bridgy Fed account on their platform, which can be confusing and opaque for people as to what is actually happening and if it is working. The upcoming Config settings page allows people to log in with their social web account (Bluesky, Mastodon, Pixelfed) and turn the bridging on with a simple switch. A New Social also mentions supporting Threads with the new Bridgy Fed Config update, which is currently not supported by Bridgy Fed. Forte is a new fediverse platform, that comes from the lineage of Hubzilla and Streams, created by the same developer Mike Macgirvin. Forte’s major feature is that it has Nomadic Identity over ActivityPub. Nomadic Identity means that you can port your entire account, including your posts, settings, social connections, etc. It is slightly different than the account migration that Mastodon has, which transfers your social graph to a new account. With Nomadic Identity, you create a single identity that can be connected to multiple different servers, so when one server becomes unavailable, all your personal data can be transferred and accessed from another server linked to your account. Forte, as well as Hubzilla and Streams, remain on the bleeding edge on what’s possible with ActivityPub. However, Forte also suffers from the same issue that its predecessors have, namely that getting to use the software is surprisingly difficult. By design there is no way to see a list of Forte servers. Forte mainly targets people with technical know-how, as the code repository does not include guide on how to setup your own Forte server. It leads to the funny situation where I would like to give Forte a try because I’m interesting in trying out the new features, but I legitimately do not know how. Myo is a new image-focused client for the open social web, and allows you to connect your Mastodon, Bluesky and Nostr accounts into a single timeline. Combining multiple accounts into a single timeline is similar to OpenVibe, but Myo instead focuses media, in a design that is more reminiscent of Instagram than Twitter. Myo is made by the same developer as SoraSNS, which is also a multi-protocol app that focuses on microblogging instead. Myo and SoraSNS are both available for iOS. ActivityPub badges is a new project that is currently in development to build a badges/credential system similar to Credly on ActivityPub. The project is currently at the proof-of-concept phase, where badges can be created and send over ActivityPub. IFTAS, the non-profit for collaborative work on trust & safety on the fediverse, recently had to shut down various of their services due to a lack of funding. In their latest update, the organisation talks about how they are rescoping and moving forward, as the organisation itself is not shutting down. IFTAS will continue with various community support projects, such as their community platform IFTAS Connect. They will also continue providing insight into commonly blocked domains, in a scaled down version of the shut-down FediCheck program. A new form of spam/scam has recently emerged on the fediverse, and it involves private messages from an account that identifies itself as ‘Nicole the fediverse chick’. So many people have gotten a variation of this message that it is quickly becoming a meme on the fediverse. It is unclear what the exact purpose of this spam is, with either a doxing ex or an elaborate 4chan troll as likely explainers. Keeping Watch Over the Fediverse: Mass Surveillance in Non-Centralized Social Media – Eric Fassbender This article by Fassbender examines how state surveillance treats federated and decentralised social networks, focusing on the BlueLeaks dataset, which contains a large amount of internal documentation of state surveillance organisations. Fassbender writes: “[…] surveillance actors are less interested in understanding decentralization within platforms, but rather look at organizations first, then take an interest in all platforms that they spread to. This means that any platform (or in the case of the fediverse, grouping of platforms that share a method for interconnecting) can become suspect.”The Links The fediverse promises social media without Big Tech – if it can avoid familiar pitfalls – Aram Sinnreich and Robert W. Gehl/The Conversation. PeerTube’s latest update revamps the about page and brings better podcast support. The Fireside Fedi livestream interviewed Laurin, the developer of ActivityPods, as well as PieFed developer Rimu. Piefed makes community discovery easier by integrating with the Lemmyverse community dataset. Sneak peek: Mastodon’s upcoming update will finally include the ability to show all replies on a post. FEP Search Tool is a small web tool to search all the FEP’s. Forum software Flarum got funding by NLnet in 2023 to implement ActivityPub, but recently decided that this effort would be postponed for the foreseeable future. Notes on migrating an account from Mastodon to GoToSocial. A Manyfold 3D viewer directly in a Mastodon timeline. This week’s fediverse software updates. Setting Up A New Mastodon Instance – a PeerTube tutorial by FediHost. Mastofuse is a Mastodon file system client. Roboherd is a tool to build automated Fediverse actors. That’s all for this week, thanks for reading! You can subscribe to my newsletter to get all my weekly updates via email, which gets you some interesting extra analysis as a bonus, that is not posted here on the website. You can subscribe below: #fediverse image
Fediverse Report #107 Pixelfed raises 138k Canadian dollars for their project, and a new way to connect researchers to the fediverse with an upcoming ORCID bridge.The News The Pixelfed Kickstarter campaign has concluded, and the project has raised 138k Canadian dollar (88k EUR/95k USD). The campaign raised money from over 2100 backers, and reached far past it’s original goal of 50k CAD. The campaign has grown significantly in scope, and indicates that the Pixelfed campaign is much more than just about the image-sharing platform Pixelfed. Pixelfed itself has also grown, and there are now reportedly 8 people joining the team. With the money, the team is working on the following: Further development of Pixelfed, as well as supporting the pixelfed.social and pixelfed.art servers Development of Loops, and getting it to a state where it can be made available as open source. In the most recent update Pixelfed says that this will be “once it is ready in 2025”. Building a dedicated server environment around the world, that can handle “the 1000s of TBs of video traffic (plus storage requirements)”. Building Fedi-CDN to host and serve Loops videos, as well as offering “excess compute/bandwidth to other fediverse platforms as a collaborative shared service.” Building an E2EE messaging platform Sup, with the near future focused on development planning. The latest update of the Kickstarter also notes that Pixelfed has started another side project, FediThreat, for fediverse admins to share information about lower-risk harmful actors such as spam accounts. This project is currently in the proof-of-concept stage. Launching a Pixelfed Foundation. Setting up a foundation was originally put behind to a 200k CAD stretch goal, but it seems like this will still happen, even though the goal is not met. The latest Kickstarter update notes that a Pixelfed Foundation is currently being worked on, as a non-profit under the government of Alberta, Canada. The amount of money that Pixelfed has raised is significant, especially by fediverse standards. At the same time, this is a lot of different types of projects that the team is undertaking. Pixelfed has a history of overpromising and underdeliving, for example the Groups feature has been announced to be released “soon” for over 2 years now, and this is a feature that they have gotten an NLnet grant for. The new projects that Pixelfed is working on, such as a shared CDN are definitely valuable for the fediverse. But with the attention of the Pixelfed team being pulled in so many different directions, and a lack of clarity on which projects will get focus, it is unclear on which timeline Pixelfed can deliver the planned features. Encyclia is a newly announced project to make ORCID records available on the fediverse. ORCID, Open Researcher and Contributor ID, is a unique identifier for researchers and scientists. Every researcher can have their own unique ORCID, and with it, every publication become records connected to that ORCID. With Encyclia, all these ORCIDS can be followed from your ActivityPub account, meaning that you can always keep up to date with research, even when the researcher does not have a fediverse account. Encyclia is currently still in pre-alpha, and not yet available for use by the public. This weekend was the SXSW festival, and Flipboard hosted the Fediverse House, with quite some well-known names within the fediverse community, as well as representatives from Bluesky and Threads, as well. There does not seem to be recordings available, but Jeff Sikes was there and had a good live blog if you want to also experience some FOMO. In my recent updates on Bluesky and ATProto I talk about how Bluesky is increasingly becoming a political actor, due to the presence of various high-profile people who are actively speaking out against the Trump/Musk regime. This impact so far is less visible on the fediverse, as there are no politicians like Alexandria Ocasio-Cortez using the platform to speak out. But resistance does not only come from high-profile individuals, it comes from people on the ground that organise themselves. To that end, Jon Pincus wrote two articles on organising on the fediverse: If not now, when? Mutual aid and organizing in the fediverses, the ATmosphere, and whatever comes next has an overview of the current state of the networks in relation to organising. Notes (and thoughts) on organizing in the fediverses and the ATmosphere has a lot more practical details, examening various softwares that can be used in practice. Both articles are great sources of information to get more practical details for people who are considering using decentralised social networks.The Links Decentralized Social Networks & WordPress with Alex Kirk. The Open Web Conversations has a new Fediverse series, hosted by WordPress ActivityPub plugin creator Matthias Pfefferle. They discuss talk about how a WordPress blog can be build into a full decentralised social networking node with the Friends plugin by Kirk and the ActivityPub plugin by Pfefferle. Standards War? – Robert W. Gehl. Gehl compares IFTAS’ funding struggles with the Free Our Feeds campaign, who are raising money to build alternative ATProto infrastructure, and describes it as an illustration of the emerging standards war between ActivityPub and ATProto. A Long-Shot Bet to Bypass the Middlemen of Social Media – John Markoff/New York Times. The NYT interviews Flipboard’s CEO Mike McCue to talk about how the company is using building a new decentralised social web with Flipboard and timeline app Surf. The Software Sessions podcast did an interview with Hong Minhee. Hong is the developer for ActivityPub framework Fedify, as well as Hollo, a single-user microblogging platform. The Fediverse Isn’t the Future. It’s the Present We’ve Been Denied – Joan Westenberg/The Index PeerTube: the Fediverse’s decentralized video platform (part 1: first impressions) – Elena Rossini Ghost’s weekly update on their ActivityPub implementation. This week’s fediverse software updates. Some notable app updates: Ivory now supports grouped notifications, and OpenVibe can save your position in a timeline. Keeping PeerTube Storage Under Control – a tutorial by FediHost. That’s all for this week, thanks for reading! You can subscribe to my newsletter to get all my weekly updates via email, which gets you some interesting extra analysis as a bonus, that is not posted here on the website. You can subscribe below: #fediverse image
Bluesky Report – 2025mar.a Frequency is a new project for a decentralised relay using a blockchain. The Bluesky clients ecosystem keeps developing.The NewsRelays, a blockchain and Free Our Feeds The Bluesky relay is often seen as a centralising aspect in the ATProto ecosystem, and there are various projects underway to combat this. Bluesky PBC is working on significantly lowering the technical requirements for hosting a relay. Free Our Feeds is working on getting a publicly-accessible relay on European infrastructure. Blockchain company Frequency has announced that they are building their own custom relay implementation. Frequency’s plan is to split the relay into two separate parts: an archive and a firehose. The archive is non-blockchain database that stores and archives all the data coming through the network (unless people delete or opt-out), while the firehose is the actual event stream, which continuously broadcasts all the events on the network as they happen. All data that lives on ATProto already has a unique identifier, a CID (Content Identifier). Every time an event happens on the network, the CID of that event gets stored on a blockchain, while the actual data gets archived in the regular database. This blockchain is effectively a list of CIDs with timestamps of when these events happen. This is the comparable with the firehose output of Bluesky’s relay, it is a list of events as they currently happen. Frequency’s blockchain-style relay has some differences, and some similarities, with how Bluesky’s relay functions. Both Bluesky’s relays output a continuous stream of events as they happen on the network. Users can tap into this firehose to get continues updates on all events that happen on the network, and use this to power their apps. Bluesky’s relay is realtime, while Frequency updates their blockchain every six seconds. This means that using Frequency’s relay will have a slight delay. Frequency also has an advantage over using Bluesky’s relay, it allows you to ‘replay’ the event stream from a past date, allowing you to see what the firehose looked like at an arbitrary date in the past. I’m not clear on what the actual use case in practice for such a feature is, but at least it is a feature that is new. The main reason why people became aware of Frequency’s announcement is for the ties it has with Free Our Feeds. The press release ties Frequency’s project explicitly as a project between Project Liberty and Free Our Feeds. Robin Berjon, one of the driving forces of Free Our Feeds however stresses that there is actually little connection between Free Our Feeds and Frequency, stating that there is no money or anything exchanged. Instead he describes it as two separate projects that both are working on the same goal: making Bluesky and ATProto more decentralised by providing more publicly-accessible relays. The whole situation honestly has me quite confused: if there is indeed no meaningful connection between Free Our Feeds and Frequency’s blockchain project, why is Free Our Feeds mentioned so prominently in the press release? Furthermore, it is clear that the culture of Bluesky is one of a deep distrust of anything that reeks of blockchains and crypto. What is the value of tying the Free Our Feeds name to a project that was always going to be distrusted and ridiculed by the community, simply because it uses a blockchain? That said, one common criticism aimed at Bluesky and ATProto is that the relay is a centralising force. While there are few independent developers that run their own relay, there is no other relay that is publicly accessible to anyone, besides the relay run by Bluesky PBC. Frequency, for all the hate and ridicule that blockchains and crypto companies rightfully deserve, is an organisation that is actually working on further decentralising the ATProto network by building a different type of relay. But for Frequency to actually get any real credits, I’ll first want to see it working in practice, because I do not feel that the crypto/blockchain ecosystem is one where organisations deserve the benefit of the doubt, and a large amount of “Ill believe it when I see it” is warranted.Bluesky clients update Tapbots, the makers of popular apps like Ivory for Mastodon and Tweetdeck for Twitter have announced they are making an app for Bluesky called Phoenix. Tapbots said that they see Mastodon as their home on the social web, and plan to continue developing Ivory. They also said that they need to keep up with Bluesky, as that is where the growth is, and the company cannot sustain itself on subscription income from Ivory alone. Some app developers, like Openvibe, are working to combine the multiple networks such as fediverse and Bluesky and Nostr into a single app. However, this is not the direction that Tapbots is taking, saying they believe the user experience is better if the networks keep being separate. The plan is to release Phoenix somewhere this summer. Image-focused Bluesky client Flashes has officially launched this week, and the client has racked up over 50k downloads so far. Flashes is a Bluesky client, and at its core it is a way to view images posted on Bluesky. One of the challenges of building an Instagram-like image-sharing app as a Bluesky client is that people use microblogging platforms in a different way than image-sharing platforms like Instagram. Flashes includes features to help people share photos in a way that’s more Instagram than Bluesky. For example, the Portfolio feature allows people to their best photos as their portfolio’, while filtering out less serious content such as a quick meme. Flashes also prominently features two custom feeds for only posts that are made with Flashes, to get timelines that are more catered towards photo-sharing. ThreadSky is a Bluesky client, that is a mix between Reddit and Bluesky. ThreadSky takes Bluesky posts, and displays them in a Reddit-style format. This makes ThreadSky different from other ATProto-based link-aggregator platforms like Frontpage. Frontpage is largely separate from Bluesky, only your account is shared between Frontpage and Bluesky. ThreadSky is a client for Bluesky, and displays Bluesky data in a visually distinct manner that is more suited for reading comments sections. The app is still in development, with quite some features not working yet. However, I do find that a Bluesky client that focusing on the comment section of Bluesky posts to be an interesting direction, and it does make reading comment sections on popular posts quite a bit easier.Bluesky is political Two weeks ago I wrote about how the America’s rapid decline into authoritarianism is resulting in Bluesky becoming a political actor in itself. Authoritarianism does not tend to go well with places that allow people of the opposition to freely gather, talk and organise. In that light, some more illustrations on how Bluesky is an explicitly political place: The TeslaTakedown protest movement against Elon Musk has its roots on Bluesky in a conversation between actor Alex Winter and sociologist and professor Joan Donovan. Winter describes Bluesky as his major tool for signal-boosting and driving momentum for the protests, with writers like Ed Niedermeyer also using the network to drive the momentum on the protest movement forward. A simple comparison in engagement on Threads and Bluesky for the phrase Slava Ukraini. Alexandria Ocasio-Cortez not going to Trump’s Joint Address, instead commenting and answering questions on Bluesky and Instagram Live. Bluesky as a place that shapes politics is also visible in Russian disinformation network Matryoshka being active, this time to spread false rumours about Germany’s election being rigged.Some updates on Trust & Safety on Bluesky: Last week, someone hacked television screens inside an US government office building to display an AI-generated video of Donald Trump sucking on Elon Musk’s toes. Videos from inside the office that also showed the AI-generated video appeared on Bluesky, first posted by reporter Marisa Kabas, where it went viral. This post then got taken down by Bluesky for violation the rules on “non-consensual explicit material”. 404 Media published an entire article on the situation. Bluesky reversed course quickly and restored the post, saying “This was a case of our moderators applying the policy for non-consensual AI content strictly. After re-evaluating the newsworthy context, the moderation team is reinstating those posts.” Another Bluesky employee clarified that “the mods are *very* touchy on any AI generated sexual videos. We have updated policy to make exceptions for notable political/newsworthy cases like this.” Bluesky also announced they are partners with the Internet Watch Foundation. Bluesky will use several of IWF’s services, which are related to shared lists of known CSAM material to help Bluesky find and take down such content more easily. Bluesky also worked on cutting down on DM spam in recent weeks.In other news Blacksky’s Rudy Fraser talked about Blacksky in a podcast interview Registering a recovery key for your ATProto account. Bluesky ads. cred.blue is a library of ATProto resources. Bluesky will be at SXSW, with CEO Jay Graber giving a keynote as well as a session with The Onion. PMSky is a project to build peer moderation on Bluesky.Next week’s sneak peek Next week’s edition will be focused on ATProto and the wider ATMosphere network again. Some news that I’ll be covering, with a sneak peek if you’re interested in digging in yourself: Threaded is a Git collaboration platform on ATProto. The ATProto Community fund announced a project for location data on ATProto. Popsky is a review platform app for iOS. OpenMeet is an open source alternative to MeetUp build on ATProto. That’s all for this week, thanks for reading! If you want more analysis, you can subscribe to my newsletter. Every week you get an update with all the articles of this week, as well as extra analysis not published anywhere else. You can subscribe below, and follow this blog @fediversereport.com and my personal account @laurenshof.online on Bluesky.
Fediverse Report #106 IFTAS is shutting down most of their services following a lack of funding, and Tumblr-like platform Wafrn now has its own apps, and a Bluesky integration to boot.The News The fediverse trust and safety organisation IFTAS has announced it is shutting down most of its services, following a lack of funding. Last month the organisation said that they would soon run out of funding, and that they’d do a final effort at getting structural funds for the organisation. This has not happened, and now IFTAS will shut down most of their services. The biggest project to be shut down is IFTAS’ Content Classification Service, a service which handled CSAM scanning and reporting for fediverse servers. When fediverse server admins encounter CSAM, most countries have mandatory reporting requirements that admins are obliged to follow. Another project that is shutting down is FediCheck, which provides shared deny lists that server could use to build their own deny lists for their servers. IFTAS shutting down their services is a double blow to the fediverse. The obvious one is that functions like IFTAS’ Content Classification Service were aiming to provide a service that filled an crucial gap in the operations of many fediverse servers. Scanning for CSAM, and handling the legal requirements on reporting to the relevant agencies is a challenging task for server admins to execute, and many fediverse servers do not have good procedures in place to handle this delicate process. IFTAS’ CCS would have provided a way for smaller fediverse server to handle the legal obligations they have regarding handling CSAM. The second blow to the fediverse is in that IFTAS fills an important role in building a collaborative structure for moderation across fediverse servers. The fediverse is a network of independent places (servers), and while they are interconnected on a technical level via a protocol, building connections between servers for collaborations is proving to be much harder. Over the years there have been many suggestions and ideas on how fediverse servers could work together, for example regarding on sharing information on which servers to block. These conversations currently take place mainly via admin backchannels or via the #fediblock hashtag, and a more structural interface could help streamline this process. For such a process to work trust is needed between fediverse server admins to participate with such infrastructure. IFTAS, as a grassroots fediverse organisation, is one of the best-placed organisations to have build trust and provide a nexus around which such infrastructure could be build. IFTAS got pretty far with their rollout of FediCheck, which was building such a place for collaboration between server admins. Now that IFTAS will not be the center around which shared moderation infrastructure can be build, will there be another organisation in the future to do so? Especially when IFTAS found out that getting funding for such a project is so difficult? Fediverse platform Wafrn has announced they now have apps for Android and iOS available in testing. I have not talked about Wafrn much, but it is one of the more interesting fediverse platforms that is currently being worked on. Wafrn is a Tumblr-inspired platform that clearly does not take itself too seriously: the name stands for “We Allow Female Representing Nipples“. It is a reference to a decision by Tumblr to ban adult content, and they used the phrase “Female-presenting Nipples” in their community guidelines which became a target of ridicule. Wafrn has a variety of unique features, such as a place to ask and answer questions for the Wafrn community. The most standout feature of Wafrn however is a native integration of both ActivityPub and ATProto. A Wafrn account allows you to have a full connection with the fediverse, as well as with Bluesky. On the fediverse, your account is visible as @name@app.wafrn.net, while on Bluesky your account is visible as @name.at.wafrn.net. Because this is not a bridge, and instead a native integration, a Wafrn account can interact with any Bluesky and fediverse account, other accounts are not required to opt-in in order to connect. One criticism that often gets put at Bluesky from people within the fediverse is that it has not federated yet. I do not think that ‘federation’ is a helpful term to help understand ATProto actually works, see this article for more context. However, the interoperability between Bluesky and Wafrn does involve interoperability between servers over ATProto, making that Bluesky and Wafrn are federated in the way people on the fediverse understand the term federation. The fact that an app called “We Allow Female Representing Nipples” is what makes Bluesky federated is honestly extremely funny to me. Link aggregator platform PieFed has added support for feeds. Feeds on PieFed are similar to how multi-reddits work on Reddit: it allows you to create a custom feed that displays posts from multiple communities. Feeds can also be shared, allowing people to follow a feed that others have created. Feeds on PieFed are somewhat similar to their Topics feature. Topics are also a collection of multiple fediverse communities around a certain theme. The main difference between topics and feeds is that topics are created by the server owner, and set for the entire server. With feeds, anyone can create and share one, and you can also follow feeds from other PieFed servers.The Links Timeline app Tapestry has gotten an investment by Tumblr. WeDistribute writes about Funkwhale and their decision to filter out far-right music. Ghost‘s weekly update on their ActivityPub implementation Xenon is a new fediverse client app for iOS Fireside Fedi is a interview series on PeerTube, and this week they’re talking with one of the people behind ActivityPods. This week’s fediverse software updates. That’s all for this week, thanks for reading! You can subscribe to my newsletter to get all my weekly updates via email, which gets you some interesting extra analysis as a bonus, that is not posted here on the website. You can subscribe below: #fediblock #fediverse image
Fediverse Report’s deep research on Deep Research’s fediverse report ChatGPT recently released a new feature, called Deep Research, that allows ChatGPT to “use reasoning”, and process large amount of online information. In his newsletter Platformer, Casey Newton reported on the new ChatGPT’s new feature. To judge the quality and functionality of Deep Research, Newton prompted ChatGPT to output a report of 5k words using Deep Research, and compare this to a similar report made by Google’s Gemini. Notably for this blog, Newton asks ChatGPT for a report about how the fediverse could benefit publishers. A Fediverse Report, you could call it. Newton does not spend a lot of time analysing the results, saying that the output hits on the requirements of the prompt, and says that compared to Gemini, deep research “blows it out of the water”. Not all tech writers are as impressed with Deep Research, AI doomer king Ed Zitron wrote another long article, ‘The Generative AI Con’, in which Zitron pushes back against AI hype. He takes a Deep Research by reading through the same report that about the fediverse that Newton has published in Platformer. Zitron is not impressed by OpenAI’s new feature, saying that “the citations in this “deep research” are flimsy at best“, and that “this thing isn’t well-researched at all.“ A report on the fediverse is quite up the alley for a blog named Fediverse Report. So let’s do some deep research on Deep Research’s fediverse report. I’ll go over ChatGPT’s output in detail, analysing what information ChatGPT gives, and which information is missing. ChatGPT’s output gets quite a lot of information correct, and more importantly, it structures the information well. It pulls in relevant examples, and also manages to find relevant obscure information. Deep Research’s ability to show what information the output is based on gives insight in how an LLM’s output gets constructed. It also allows the sources to be analysed, and it turns out there is a lot of interesting information you can learn by looking at the sources that Deep Research uses for the output. The report is also pretty well structured, and hits on most of the relevant points that publisher who is curious about the fediverse needs to know. A common critique of LLMs is that they will give factually incorrect information, often simply called hallucinations. This problem is also visible in Deep Research’s output, it contains factual mistakes. The problem of factual (in)correctness of LLMs is well-known, and not a debate I want to rehash here. What I am interested in is the analysis and research part of Deep Research: which sources does the output use? Are those sources any good? And just as importantly: which relevant sources should have been part of the report, but aren’t? Judging if the output of an LLM is correct or incorrect is reasonably straightforward. But judging if the output of the LLM does not include information that reasonably should have been included is much harder. This goes doubly so for emerging fields like the fediverse, where there is no authoritative base of knowledge to rely upon. In ChatGPT’s output that Newton uses to get a sense of the performance of Deep Research, I find that there are three types of issues relating to data and analysis. There are issues with the quality of the sources that ChatGPT cites, and there are sources missing that I expect to have been cited. But the most intriguing part for me is when ChatGPT cites a source correctly, but the resulting output is still lacking, because the needed information to get to a good understanding is not actually available on the internet.Source quality One issue that Deep Research struggle with is with the quality of the data sources it cites. The clearest example is this specific Reddit post, which ChatGPT cites six different times as a source in the section on monetization in the fediverse. The post is titled ‘monetization’ and posted on /r/fediverse. This specific Reddit post is the second search result on Google, as well as Kagi and DuckDuckGo for the search query ‘fediverse monetization’. ChatGPT heavily focuses on WebMonetization by Interledger in this section, and how it integrates with Castopod. Interledger is a (non-crypto) payment network that allows people to send microtransactions to creators. There is indeed an WebMonetization integration with Castopod, but ChatGPT’s output gives no indication of how unrepresentative this is for the fediverse. With respect to Castopod and Interledger and what they are building, WebMonetization is in no way any meaningful part of the fediverse. In fact, both organisations presented their integration in 2025 during FOSDEM. The room consisted of the most in-the-know in-crowd of fediverse developers, and as far as I can tell this potential fediverse integration with WebMonetization was new information for most if not all of them. ChatGPT provides a reasonable summary of the position of Interledger regarding monetization in their output, but the complete lack of context of how early in the adoption stage WebMonetization makes that the output does not represent the state of the fediverse, as it is currently used by most people, at all. An article by TwipeMobile is the prime source that ChatGPT uses throughout the article. TwipeMobile is software development company that builds apps for newspapers. The company also publishes research papers about related topic, and one is called “What is the Fediverse? A guide for publishers and the uninitiated.” With a title like that it is no surprise that ChatGPT likes the article. The quality of the article is mediocre however, and it lives in the twilight zone where it is impossible to tell for sure whether the article is generated by an LLM or written by a human. The TwipeMobile article has some major issues with factuality, and as a result, ChatGPT’s output suffers as well. This is especially noticeable in ChatGPT’s comparison between ActivityPub and ATProto. The TwipeMobile article describes ActivityPub as ‘widely adopted’ and having an ‘established user base’, and ATProto as early in its growth. The article was published on 6 December 2024, and at that date the fediverse had 1.1 million monthly active users, while Bluesky had around 11 million monthly active users. That ATProto is 10x the size in terms of active users compared to ActivityPub does not seem particularly clear from the language used in the TwipeMobile article, which seems to imply that ActivityPub is more active than ATProto. As ChatGPT’s output relies so heavily on TwipeMobile’s article, this misconception is reflected in ChatGPT’s advice as well, which describes ActivityPub as having an ‘established audience’ and ATProto as being in an ‘early growth stage’.Missing sources ChatGPT does miss a few relevant sources, that would help publishers get a good understanding of the state of the fediverse and whether the network is relevant for them. One source that is missing from ChatGPT’s output is regarding monetisation and sub.club. Sub.club was a platform that let fediverse creators offer paid subscriptions and premium content, using existing fediverse infrastructure. It allowed people to set up a fediverse account, to share content with the rest of the fediverse. Creators could then set a paywall on posts if they so wanted, and sub.club provided the payment infrastructure. Sub.club shut down in December 2024, only a few months after launched, and they managed to onboard only 150 people. For publishers that are interested in monetisation on the fediverse Sub.club’s struggle to gain traction is a relevant data point. Sub.club got a fair amount of media attention (1, 2) from well-known outlets, so there was not a lack of sources for ChatGPT to cite from. Another example of sources that are not included in ChatGPT’s output is not only a matter of not linking to articles, the lines between ‘relevant information that is missing from ChatGPT’s output’ and ‘relevant information that is not covered in well-known news publications’ are thin. Some of the most relevant information that is missing in the output is also missing in articles that rank high in search engines. A notable example is the statistics that Heise editor Martin Holland regularly publishes, which compares traffic to their site from Mastodon, Bluesky, Threads and X over a longer time period. The original prompt by Newton asks for report on how the fediverse could benefit publishers, and traffic data time series is one of the best ways of showing the concrete benefits for publishers. As best I can tell this data series is not published in news media, which explains why it does not show up. ChatGPT seems to prefer to use English-speaking sources. For example, there is no information on how ZDF, one of Germany’s largest public broadcasters, has had their own Mastodon server for years. Heise being a German news outlet also likely contributes to the data series is being not being reported on in English-speaking media, and not being found by ChatGPT.On data availability One limitation that ChatGPT’s output has is that it is dependent on available information. But what if the information is not available online, nor a clear indication there is missing data at all? Let’s take a look at how ChatGPT describes Medium as an example for how publishers can use a Mastodon instance to amplify author’s reach. The output cites Medium’s announcement blog post and summarises the point of why Medium started their Mastodon server this correctly. On a surface level ChatGPT’s output is good: it found relevant information and summarised the important points. What’s missing here is the follow-up: Medium launched their Mastodon more than 2 years ago. So did their plans actually work out? That seems pretty relevant information for a potential publisher to know. In 2024, Medium barely posted about their Mastodon server on their blog. CEO Tony Stubblebine mentions Medium’s Mastodon server once in his ‘State of Medium‘ post, also saying that he finds Threads to be a better place for self-promotion. Those two additional data points suggest that Medium’s experience with launching a Mastodon server is mixed: the me.dm server has 1.7k MAU, so it clearly provides some benefit to the Medium user base. But neither are there signals that it is an overwhelming success. The issue here is that Medium (or anyone else) has not published an analysis or statement with a follow-up on Medium’s Mastodon server, to ask the question: “are the benefits of running a Mastodon server by a publishing platform good enough that it is recommended for other publishers to do so as well?” Newton’s original prompt asks for “high-level strategic analysis for a digital-native publisher”, and this is the type of analysis that is needed for a publisher to actually make a decision. For a publisher it is only knowing about the existence of a project is not the point, a publisher needs to know if that project is successful and if they should consider it as well. This type of analysis is hard for ChatGPT to properly execute: LLMs are fundamentally about processing data, and it cannot process data that doesn’t exist.Some more notes ChatGPT’s output on recommendation I find to be quite good. The first three points of advice – Establish an Authentic Presence, Integrate Your Website with ActivityPub, Engage with the Community- are quite good at a high level, and advice that I would give to publishers as well. The advice to “Stay Adaptive with Protocols” is also good advice, but the description is wrong on some pretty important parts. Then again, this is also because the quoted source (TwipeMobile again) is wrong on this, so at least ChatGPT cited a bad source correctly. ChatGPT spends another section on discoverability and engagement, writing a comparison between ActivityPub and ATProto. It correctly notes that discoverability on the fediverse happens to community sharing and hashtags, and that ATProto has space for algorithmic discovery. ChatGPT’s tendency to equate both sides is in full play here: it does correctly say the difference between the networks, but refrains from making a material conclusion about it, describing them both as equal but different. Again, the biggest fault ChatGPT here is not in what it writes, but what it does not write. There is no information on that search is opt-in on the fediverse, that around 5% of accounts have opted into being discovered, and that as a result search and discovery works significantly less well on the fediverse than it does on Bluesky. Finally, ChatGPT offers the advice to use “third-party analytics services to gauge engagement.” There are indeed tools available for analytics, but not all of them are easy to find, or to know which one to use. ChatGPT does not tell you which tools you can potentially use, instead offering very basic advice on tracking engagement. ChatGPT’s output would have been better here with some more ‘deep research’. Another general note on ChatGPT’s output: Zitron does not like Deep Research’ tone of writing, and I agree with what he writes here: “I don’t like reading it! I don’t know how else to say this — there is something deeply unpleasant about how Deep Research reads! It’s uncanny valley, if the denizens of said valley were a bit dense and lazy. It’s quintessential LLM copy — soulless and almost, but not quite, right.” Over the last years how you feel about LLMs and generative AI has quickly become an identity marker for many people, with opinions ranging from LLMs being the new way to build machine god to a torment nexus that is designed to strip workers of their power. For this article I am not aiming to argue for a specific position in the debate about the values and impacts of LLMs, and instead I’m aiming for a smaller goal. OpenAI released a new mode that says it can do deep research, and a prominent tech writer used a singular prompt to get a sense of how good this new mode this. This prompt happened to be on a subject field I do know something about. I wanted to know how much ‘deep research’ was in ChatGPT’s output, so to that end, I simply did some deep research of my own.