We’re back with a new addition to our series “Building a Free Internet of the Future”, in which we publish a monthly interview highlighting the experiences and perspectives of individuals and communities supported by NGI Zero (NGI0) grants, as well as members and stakeholders of the NGI0 consortium. Funded by the European Commission, NGI0 supports projects related to free software, open data, open hardware and open standards.
DrFed is a web-based platform for developing and debugging implementations of ActivityPub – an open standard and protocol for decentralised networking. It enables users to inspect objects and activities, while tracing the root causes of “federation fails”. A team of four people from East Asia is developing DrFed. We interviewed one of them, Hong Minhee, regarding openness, tech, technical specs, their NGI0 Commons grant, the Fediverse’s strengths and limits, language barriers to decentralisation, and far more.
Fediverse is a neologism made up of two words, "federation" and "universe", and refers to the ecosystem of free/libre and open-source software, the people who contribute to it, as well as the people who use it. Federation enables users to interact across different services; for instance, someone with a Mastodon social media account can comment on a Pixelfed photo or a PeerTube video. “If we want distributed governance, we have to make the technology possible for more than the already powerful,” underlined our interviewee.
This interview has been edited for clarity and length.
Why did you decide to work on “reducing the complexity” of the code for federated server applications?
My first attempt to federate my own software with the Fediverse was not easy. I started with the ActivityPub specification, then draft-cavage-http-signatures-12, then WebFinger. I read the Mastodon codebase. I read the Misskey codebase. By the time I started writing code, I was already tired, and the real work had barely begun. Getting one message from my server to appear on both Mastodon and Misskey took weeks.
People often say the Fediverse runs on interoperability. That is true, but it is also where the trouble starts. Reading the spec carefully is not enough. Mastodon makes one choice. Misskey makes another. Sometimes the spec leaves room for both. Sometimes the spec is silent, and each project has already built its own answer. You find those answers one failed request at a time.
I love the Fediverse. That is what made the experience so frustrating. I wanted more software to join it, but the barrier felt high enough that most developers would simply give up before getting anywhere. That is when I became convinced that the complexity needed to come down.
What's the Fediverse for you? Why do you love it?
I used to love Twitter. In the 2010s, Korean Twitter, at least the part I lived in, was full of leftists and counterculture people. I met a lot of people there. I learned a lot there. But it was never quite safe.
Cyberbullying was everywhere. Some people could shrug it off. Others really could not. Around 2018, I started feeling that more clearly.
I went looking for somewhere else, and like a lot of people, I found Mastodon. At first I just liked the quiet. The Korean Fediverse was very small then, almost silent. I liked the calm, but I also got bored, so I started spending more time in the broader, mostly English-speaking Fediverse. I began posting in English more than Korean.
After a while I understood the place better. It had the people I had missed from old Korean Twitter: anarchists, socialists, queer people, environmentalists, feminists, furries, otaku, and many others who did not fit well in corporate social media. The language barrier was real, but I loved being there.
Then Elon Musk bought Twitter and renamed it X. That was when I started thinking seriously about Korea again. I wanted some of that old Korean Twitter feeling to exist in the Fediverse too. I wanted to see Korean-language services grow there. Eventually, I started working toward that myself.
I do not stay in the Fediverse because of the technology. Federation matters because no single company owns the room. That changes who shows up, and how they behave. The people are the reason I am still here.
When we think, when we design, when we code and develop, we do so under the influence of various factors, which are then woven into our creations (along with other constraints). An example of this can be seen in the current state of ActivityPub. Is this something you address as part of the DrFed project?
ActivityPub was shaped by the people who happened to be building around it at the time. The problem is that building ActivityPub software is hard. The cost is high enough that many developers either give up or copy Mastodon closely. That is understandable, but it also keeps the ecosystem narrow.
DrFed is partly my response to that. One goal is to help new implementations try the ideas being proposed through FEPs, the Fediverse Enhancement Proposals. We are building a tool that checks whether an implementation supports a given FEP correctly. That saves developers from verifying everything by hand.
I want more people to be able to experiment with ActivityPub without first becoming experts in every edge case of Mastodon behaviour. If more implementations can afford to try different choices, Mastodon's choices stop looking like the only possible shape of ActivityPub.
I get the impression that you’re really keen on opening things up, on enabling more people to use and get involved with this technology. Let’s leave the technical aspects aside for now. What is the philosophy underpinning the political perspective behind this project?
People call the Fediverse decentralised. I do too, with a caveat. Decentralisation is not a moral property of a network. I care about it when it moves power: when more people can govern the spaces they use, instead of waiting for a small privileged group to decide what counts as speech, abuse, safety or legitimacy.
X, Instagram, LinkedIn – most large social media platforms are run by US-based private companies. That fact shapes the whole environment. Ranking systems decide who becomes visible. Moderation teams decide which reports are believed. Capital has its own politics, usually close to whoever already holds power.
In Korea, anti-feminists have run coordinated reporting campaigns against feminist voices on X and Instagram. The platforms have often treated those reports as valid. On YouTube, content promoting microfascism keeps moving through recommendation systems as if nothing is wrong. These are not edge cases for me. They are examples of what corporate governance looks like from the outside.
The word I keep coming back to is de-corporatisation. A protocol can be distributed while power remains concentrated. If ActivityPub is technically distributed, but one private company ends up deciding how the Fediverse works in practice, then the politics have failed. If no company dominates, but Mastodon's interpretation becomes the only practical one, that also limits what the Fediverse can become.
The web shows the danger. Browser technology has become so complex that an independent implementer can barely imagine starting from zero. The major browser engines are maintained by large organisations, mostly corporations. They can follow the constant growth of standards because they have money, staff, lawyers, infrastructure. Mozilla, one of the few relatively independent voices left, struggles just to keep pace.
Technical complexity decides who gets to participate. Past a certain point, only capital can remain in the room.
The Fediverse could follow the same path. Mastodon can probably keep up as the stack grows. Meta can certainly keep up with Threads. Independent implementers will have a harder time entering, experimenting, disagreeing. A network can look decentralised from the outside while becoming very narrow in practice.
That is the future I fear most.
When I argue for simpler ActivityPub tooling, I am making a political argument. More people should be able to build Fediverse software without first becoming protocol specialists.
Libraries, test suites, documentation and good abstractions are not small conveniences. They decide who gets to implement, who gets to fork, who gets to govern.
If we want distributed governance, we have to make the technology possible for more than the already powerful.
You have been awarded an NGI0 grant to carry out this technical/software engineering work. But what working conditions would enable your team to reach this “horizon” you mentioned?
Receiving the NGI0 grant has been a genuine source of joy. What pleases me most is not the amount. It is where the money comes from.
Public funding and venture capital differ in quality, not just origin. For us, this is much higher quality money. We are being supported to serve the public good, not the private interests of any particular group. That difference shapes everything about how we work.
What frustrates me is how absent this model is in East Asia. Free and open source software is still not widely treated as a public resource in Korea or Japan, at least not in the way it seems to be in parts of Europe. There is no comparable public funding infrastructure here. That is how a team based in Seoul ended up applying to a European fund. I am grateful for it. I also think it is absurd. Public investment in free software should not be a European specialty.
Language is the part I keep tripping over.
Most technical discussion around the Fediverse happens in English, as if this were just nature.
For me, reading English is manageable. Writing it is much harder. Speaking it is worse. I was honestly relieved when this interview turned out to be written. My spoken English still has a way of making me sound less competent than I am, which is a very efficient form of humiliation.
syuilo, who created Misskey, once said that because he speaks no English, he cannot participate in Fediverse standardisation discussions at all, even when he has something to contribute. I understood that immediately. Misskey is one of the most widely used ActivityPub implementations. Its creator cannot really take part in decisions about how the protocol develops. That should embarrass us.
I do not think major technical spaces should default to English without even admitting what that choice costs.
English speakers may worry that multilingual discussion would be slow or inconvenient. Yes. That is exactly the point. Non-native speakers already live with that inconvenience every day, in every thread, every call, every standards document. The proposal is not to create friction. The friction is already here. Right now it is just distributed very unfairly.
That is why hub languages matter. They would not fix everything. They would at least stop pretending that every community reaches English from the same distance.
But how can we use technology to change the desires, habits and practices of people who are “stuck” in a Western-centric mindset regarding language, technology and geography (which is not a bad thing in itself, but one that must be tackled head-on)? Above all, what practical benefits can we derive in our daily lives from using a social network based on DrFed?
Translation is still treated as a convenience. In social networks, I think that is a political choice.
There are sharp critiques of English-centred software culture written in Korean, Japanese, Mandarin, and many other languages. English speakers often never see them. The people most implicated in a critique may be the least likely to encounter it.
If translation is built into the platform, the burden moves from the reader to the system. A Korean post can enter an English-speaking argument without first asking a bilingual person to donate hours of labour. People still disagree. They just have fewer excuses for missing what was said outside English.
Most social media rewards sameness. It shows people what already fits their language, politics, habits and tone. That is a design decision. Social media is software, and software can be changed.
I feel an obligation to test alternatives in the Fediverse, because the Fediverse still leaves room for this kind of work. Its governance is not sealed inside one private company. That matters to me as a developer, because design choices about language are also choices about power.
This is part of what I am trying to address in Hackers' Pub, a social network for software developers that I am building on top of Fedify. Translation is part of the product design from the beginning. I am also experimenting with algorithmic approaches that make unfamiliar perspectives easier to encounter, instead of letting the feed collapse into sameness.
Fedify and DrFed work toward the same goal from different angles. Because Fedify handles most of the federation implementation in Hackers' Pub, I have room left to think about social design rather than spending it all on protocol details. DrFed gives that same breathing room to any developer whose implementation needs to talk to the Fediverse. Together, they try to lower the cost of entry enough that more people can afford to build something new.
That is what I hoped both projects would make room for: software that can ask what a social network owes to people who do not share a language, a country, or a default political imagination.
What are the East Asian communities in the Fediverse like?
The East Asian Fediverse I know does not feel like one public. It feels like two overlapping worlds.
Some people arrived through the English-speaking Fediverse. In Korea this group is smaller, more Mastodon-shaped, more used to political arguments that came through English-language spaces. Many are left-leaning, or at least comfortable with liberal and left vocabulary from the West. They often think internationally. Sometimes that helps. Sometimes it makes them miss what is happening nearby.
The larger world is different. It is much more Misskey-shaped. Its conversations come from local news, from X, from fandom, from people already connected across Japan, Korea, Taiwan and elsewhere. In that world, Japan sets much of the tone.
I would not explain this mainly through national prestige. In practice, the common language is otaku culture. The Fediverse in East Asia is still used heavily by technically inclined people, and that overlaps with people who are deep into anime, games, Linux, and adjacent fandoms. Many Korean and Taiwanese otaku know Japanese, or have spent years surrounded by Japanese media. Japan is where much of that culture comes from. The history between Japan and its neighbours is real, but inside these spaces it is often not the first frame people reach for.
I am personally closer to the first world. But in Korea, and probably across East Asia, the second one is larger and more ordinary.
It is also harder to enter if you are not already part of that culture. The references come quickly. A joke from an anime. A game character. A Linux habit. A Misskey feature people assume everyone understands. Some people never find a foothold there. Some do not want to.
That is one reason I have been trying to make the Korean-language Fediverse wider than otaku culture. I want spaces where the entry point is not a shared fandom.
I also worry about the politics of the culture around it. Since the 2000s, the mainstream tone of otaku subculture has moved toward misogyny and the right. Not everywhere, not in every community, but enough that it shapes who feels welcome. I see this as part of a wider rightward shift among younger people in Japan and across East Asia. If the Fediverse stays enclosed there, it will not become a home for many of the people I want to see in it.
Last but not least, do you have a message close to your heart that you’d like to share?
Start with the online spaces you actually use. Who has power there? Who can remove you, silence you, or decide which voices get carried further?
If power sits with a few people, do not treat that as normal. Push them to give it away. If they refuse, leave. Find another place with your friends. Build it if you have to.
The Fediverse could be one place to start. It is not perfect. I do not want to sell it as a refuge. But people can still argue about power there, and sometimes change where it sits.
That is why I keep working on it.
Xavier Coadic is a consultant for the NGI0 consortium, and a free/libre open source software activist with 15 years of experience in free open source cultures and communities (software, data hardware, wetware, policy makers and political groups, research and development).