Eugen Rochko is the CEO of Mastodon — the open-source decentralized competitor to Twitter. It’s where a lot of Twitter users have gone in our post-Elon Musk era.
The idea of Mastodon is that you don’t join a single platform that one company controls. You join a server, and that server can show you content from users across the entire network. If you decide you don’t like the people who run your server or you think they’re moderating content too strictly, you can leave and take your followers and social graph with you. Think about it like email, and you’ll get it. If you don’t like Gmail, you can switch to something else, but you don’t have to quit email entirely as a concept.
Now, if you are like me, you hear the words “open source” and “decentralized” and then the word “CEO” and think, wait, why does the decentralized open standard have a CEO? The whole point is that no single person or company is in charge, right? Well, welcome to the wild world of open-source governance. It’s a riot, my friends. You’re going to hear me and Eugen say the phrase “benevolent dictator for life” in dead seriousness because that’s how a lot of these projects are run.
Of course, we also talk about money and structure. Mastodon doesn’t make a lot of money, and Eugen is figuring out how to build a structure that scales past just a handful of people. This tiny and mostly volunteer labor of love might very well be the future of social networking and, if you believe the hype about ActivityPub, might have some part in the future of the web. That’s pretty exciting, even if things seem a little messy in the moment.
The following transcript has been lightly edited for clarity.
Eugen Rochko is the founder and CEO of Mastodon. Welcome to Decoder.
It’s very nice to have you here. Mastodon is having quite a moment, but I know you’ve been working on this project for a long time. I think there’s actually quite a lot to talk about.
Let’s start at the very beginning. Mastodon is in popular conversation right now as a competitor to Twitter, but that’s not really what it is and not really where it started. What is Mastodon?
I mean, it’s not that far off from where it started. I started working on it in 2016, because I thought that something as important as Twitter should not be in the hands of a single company. I was a pretty heavy Twitter user back then — I think I started using it in 2008 or so, when I was a teenager — and it quickly became a very important part of my life for talking to friends or finding out what was happening in the world. Around 2016, I felt fed up with how Twitter was being run as a company, where it was heading, the community that was on there, the harassment, and so on. I started looking into alternatives, and after viewing the landscape, I decided to build a product of my own — and tried to make it good.
I appreciate that. What I mean specifically by, “It’s not exactly like Twitter,” is that Twitter is a company. Up until recently, it was a publicly traded company that had investor reports, had to make a profit, and had employees that controlled the entire product. Mastodon is not that in almost any way.
Indeed. Mastodon is free and open-source software that allows you to create a social media server or platform that connects to a decentralized network of similar servers, all talking the same protocol, and allowing not only different Mastodon servers to exchange information, but also other software that speaks the same protocol. It’s a very powerful ecosystem with a lot of potential.
When you say you’re the CEO of Mastodon, but the product itself is open-source software that anybody can run, how does that work? What are you actually the CEO of?
I am the CEO of the company Mastodon, which works on the software called Mastodon. It’s slightly confusing perhaps, but it’s also fairly straightforward about who is making the software. It’s us, and we have the same name.
For the longest time, it was just me. I started working on this in 2016, when I was still in uni. After graduating, I started a Patreon page for it, which was $5 per month for a while. Then it gradually increased to [making], I don’t know, like $200 per month or $600 per month. It was enough for me to live off of in my situation back then, and I started working on it full-time. I was technically a sole proprietor, which is a type of way to do business in Germany; you basically just represent yourself as a company.
That went on for a long time actually, until 2021, which is when we finally incorporated a separate legal entity for Mastodon. Over time, we got more people contributing to the code. Some were on a voluntary basis, and there were a lot of drive-by contributors on GitHub. Over 700 people have contributed lines of code to the software. There are also people we work with regularly, such as the contractors that I hired.
Right now, compared to where it started, it’s quite a sizable company. There’s me, who is a full-time employee, developer, CEO, whatever you want to call it, and another developer, Claire. We also have a CFO on a contract basis and a talent manager, and we are hiring for three more positions full-time. We have two developers working on our iOS app and one developer working on the Android app, all contractors. Then there is a user experience design agency that works with us on a contract basis. That’s about it, but it is quite a lot of people to manage.
So the company itself — and I’m just keeping a rough count here — has four full-time people, a handful of contractors, and an outside agency?
Well, currently there are two full-time people, but we’re hiring for three more. In the end it would be five people.
Is the talent manager also on a contract basis?
Okay. Decoder for me is all about structure. How do these things happen? How do we put these companies together? For Mastodon, there’s a lot of pressure on the protocol and there’s a lot of interest in it. There are a lot of new servers being spun up. Then there’s having to develop and manage the development of a large open-source project like Mastodon itself. How do you split the time between all the things people want Mastodon to do, all the features people might wish to add to it, and you actually running your company?
I’m not going to lie, it has been quite tough since November. For most of Mastodon’s lifetime, it wasn’t as difficult, because the pressure on the project was much lower, even during busy times. I mean, Mastodon didn’t just pop up out of nowhere in November 2022. We had our run-ins with global fame even going as far back as 2017. That’s when it went viral for the first time and magazines like Mashable and The Verge covered us.
In hindsight, it was much lower pressure back then. I was able to manage on my own, but now, clearly the pressure is much, much larger. It’s not enough for just me to be involved in the project. It needs to expand, there needs to be more people working in different areas, and there needs to be more delegation.
Since November, it has basically been a whole ongoing project within our company to figure out the hiring process, because we’re basically hiring for the first time. As you can tell from the story I told about the development of the company, the first step is basically going from just me to more people and more full-time employees. It’s a huge step and it’s a new process for us, so we’re spending a lot of time figuring out, “Okay, these are the key roles that we need to fill as soon as possible.”
Again, we’re also dealing with a limited budget, even though a lot more people donated through our Patreon in November. It was a huge explosion of funds — going from $7,000 per month to $30,000 per month. That is a big budget increase, which allowed us to look into hiring more people, but it’s still a constrained budget. So we had to figure out, “Okay, these are the key positions that will have maximum impact on what Mastodon is doing as a company.”
We decided we need a DevOps person, because I can’t be running the company and solving technical issues 24/7. We need another developer, so somebody else can work with Claire when I’m not around. We need a product designer. We got this far with me doing the design and figuring out usability and user experience, but I’m not actually qualified to do that.
And I want Mastodon to be the best product it can be. I want it to be polished, and I want it to be on par with any commercial alternative you might throw at it. It’s really important for me to find the right design person who would also be able to take ownership and initiative at this stage. A lot of larger companies have whole teams working on delegated tasks, but in our case, we have to find somebody who’s able to multitask a little bit and fill the role that I had to fill until now.
I want to come back to all of that, because I think it’s very interesting. Just being on this podcast, a lot of people are going to come talk to you, and I’m very curious to see how that goes. But I want to sit for one second in what I think of as the middle stage, which very few people pay attention to. I always think about this when listening to the story of a band or something. The band forms, they play one show in a garage, and then everyone skips to them playing in stadiums. You’re kind of in the, “Oh boy, we better staff up so we can play in stadiums,” part of the journey.
That middle part for you is between 2016, 2017, when we covered Mastodon as it launched, and November, which is when Elon bought Twitter and had all the attention. What made you stay convicted in Mastodon, that this was a thing? Regardless of the big spikes of growth, the challenges, and the financial limits you had, why did you stay focused on it? You were ready for the moment in November of 2022 — the product was there, the servers were there, the ecosystem and the community was there. It’s very important to be ready for the moment, but you couldn’t have possibly known it was coming. So what made you stay focused?
Well, I would argue that I couldn’t possibly have known the moment was coming then, but the writing was on the wall for me back in 2016, that something was going to happen sooner or later.
Do you mean something with the big social networks or something with Twitter specifically?
“Nobody’s talking about MySpace like it’s a relevant thing anymore... I never believed that Twitter would be completely immune to that.”
Just a shift, because social media websites come and go. They die. Even if they stay around, like MySpace did, nobody’s talking about MySpace like it’s a relevant thing anymore. There are plenty more that have literally disappeared, like App.net, Google Plus, Friendster, or whatever. It is a graveyard, and I never believed that Twitter would be completely immune to that.
You can see that for all seven years I’ve worked on this with a preparation for this sort of thing, just without knowing exactly when it would happen or what to do. It has been a long process of figuring out the right features, the right designs, and the right messaging, how to do this and how to do that, so I will not claim that I had all the details figured out from the start. It was a long process to figure stuff out.
So you’ve described the financial journey here as $600 a month, to $7,000 a month, to $30,000 a month, all on Patreon. Is there another source of revenue? Or another potential source of revenue?
No, that’s pretty much it. Over the years, we have created an additional platform for the higher-tier Patreon sponsors to give us money as a sponsorship, which allows us to give them invoices back and save on Patreon fees a little bit. That’s the only thing. For the most part, Mastodon is financed through Patreon and the platform we built for the higher tiers. That’s it.
Last year, we did receive a grant from NLnet, which is a grant that comes from the European Commission for open-source projects that work on decentralized social networks. That was a little bit of site funding, but it was a one-time thing. In the past, there was another one, a prototype fund project, but that was more like a personal grant kind of thing. Sometimes you find grants that sponsor people who work on open-source projects, and it can help continue working on this.
To answer your original question on why I stuck around, I believe in the vision. I believe that Mastodon is the better way to do social media. I’ve stuck around because, well, I’m dogfooding it. I am using it. It’s my daily driver and I have a home feed that I look at every day. I share what’s happening in my life or the music I listen to on my Mastodon profile. I most enjoy Caturday, and looking through and sharing cat pictures every Saturday. So that’s why I’ve stuck around.
Of course, there is a personal element to this as well, in that it’s kind of a cool job to have. You’re doing something unique. In a way, you’re getting paid to work for yourself at your own pace. Sometimes that means times are quiet and there’s not much to do, and unfortunately, sometimes that means that you have to work pretty much nonstop. That can be extremely stressful and extremely draining, like it has been for the past three months. That is the downside.
Usually, people go through that experience at a startup because they have equity and might go public or because they might sell the Facebook for a huge amount of return. Do you see that outcome for yourself? Will there be some kind of exit from Mastodon where you’ll reap some sort of massive financial reward? Or is it, “This is the vision. It’s an open-source project, and we need to stay focused and idealistic”?
It’s not really why I went into this, and no, I don’t see that. It is an open-source project, it is free software, there’s nothing to be reaped from it, but I think I see a good future for Mastodon growing both as a network and as a project. My wage right now is really rather low, and for the most part that’s because when new funds come into the project, there are important things that need doing more than paying myself more money. But I am hoping that there will be a break-even point where I’ll be able to have a dignified wage as well.
Fair enough. This is the classic Decoder question. Just listening to you, it sounds like you’ve had to make an escalating set of ever more complicated decisions since you started this project and it was just you. You’re now at what might be the most fun part, which is staffing up, figuring out who to hire and how to hire. How do you make decisions? What is your framework to make decisions?
A “framework to make decisions” sounds very serious indeed, and it makes me wish that I had a more elaborate answer to that. Sometimes, my decision-making is led by what the community is asking for, but of course, it all comes down to my vision for the project and what I think is right. That is my framework for considering community suggestions or community demands and figuring out what the project needs and what it should avoid.
In the future, I am really looking forward to having a more organized model for this kind of decision-making. The keyword that has been thrown around is “participatory governance.” Now I’m not going to promise anything specific, but that is something I’m talking to a couple people about.
One thing that I would want to get out of something like that is a more organized way of inferring community opinion about specific features or directions, because at the moment, it is quite chaotic. The main way that people can voice their opinion about what should or shouldn’t be in Mastodon is issues on GitHub, which is basically feature requests, bug reports on our source code repository. If somebody submits a feature request, how do you know if that’s something a lot of people wanted and would benefit from, or if that’s something that only one person would benefit from and a whole bunch of people would hate?
Right now it’s very chaotic. My wish is to find a way to have some kind of vehicle where people would feel like they’re represented. I would get a way to know they’ve made their decision, so I can then either go along with it or not. But I would know that there’s some kind of backing to it, instead of trying to figure out, “Okay, 40 people on GitHub have given a thumbs up to this feature request. What does that mean for the 2 million people using Mastodon today?”
“How you build a social network using the tools of social networking” seems like a recursive problem. We could do the entire podcast and probably an entire conference on open-source governance models. There have been a lot of them over time, and each of them has trade-offs. I’ll just use Linux as an example, because I’m confident the audience is familiar with it. At the top of Linux is Linus Torvalds, and he literally has the title of “Benevolent dictator for life” of Linux, and that is the governance model that flows down from that title. Are you thinking, “I have to pick one of these well-worn models”?
That’s how it is right now. Honestly, I subscribe to that model. I think it’s effective, and I think that it leads to a better product. A good product needs long-term vision and it needs cohesive vision. That’s something that a committee cannot give. When you have a lot of people who have pet issues and one thing that they care about, it kind of ends up being a patchwork. It loses some of its focus, and it can end up in a situation where it just stops being a good product and becomes too confusing. Sometimes, you need to make executive decisions about changing stuff in a serious way, which might not be popular with what most people in a committee would want.
“I don’t look down on the ‘benevolent dictator for life’ model. I think it has its place, in open source at least.”
So I don’t look down on the BDFL model. I think it has its place, in open source at least. I would not make any comments or claims about other areas of life, but in open source, I think it makes sense. That’s what I would prefer to stick with, but that doesn’t mean I don’t think there’s better ways to involve other people and have better communication.
One more question here, and then I want to talk about Mastodon, its community, and how it’s working as an actual social platform. What you’re describing right now is familiar in shape, if not in the specifics. It’s an open-source product that has found product-market fit, or at least the opportunity for product-market fit. It’s going to grow really fast, and you need some help scaling.
This is when venture capital firms show up at your door and say, “Look, we know how to do DevOps. We can install a bunch of lawyers to help you figure out open-source governance. Look at all these tools. Here’s a whole army of people that have done this before. Let us give you the money to help, and we’ll scale it as fast as we can.” Has that been happening to you?
We’ve definitely had a lot of people from venture capital firms reach out to us. We’ve even tried having a few conversations just to see where they were at and what they wanted. We obviously had considered ways to have a more sustainable funding model for the open-source project that would not rely on Patreon. We have explored if maybe there would be a way to find funding for that. I cannot say that anything productive has come out of it. We’ve rejected every venture capital firm that has reached out to us so far.
Now, when I say a more sustainable funding model for the project, I obviously mean software as a service, because it’s the most natural way for an open-source project to basically find a business model for itself. You have a product that you have the most familiarity with, and you can offer it to other people in a way where they don’t have to figure out how to install it, how to manage it, and so on. It is quite natural, and it has been on my mind for quite a few years as a sort of backup plan in case people stop donating to the Patreon for some reason. But there’s a lot of activity in that space right now. A lot of these hosting firms are starting to jump up. I think we could have an edge in that space, but it also doesn’t seem like a priority right now, if I’m honest.
Are you rejecting all the VC firms because the pitch is basically the same for all of them? “We’ll give you a bunch of money and resources, but we’re looking for 30X revenue growth,” or whatever?
A hosting business is not really venture capital scale. It doesn’t promise the same returns. What they seem to be interested in is, “We’ll give you money now and you don’t have to think about monetizing, but in two years, let’s figure out how we could turn your open-source project around.” That’s kind of a no-go zone for me. It’s a trap. It’s clearly against our project’s ideals.
Let’s talk about the Mastodon community, and the product. So I kind of understand the company now. I understand very clearly that you have a specific set of ideals — about how you want to run the company, about the nature of decentralized social networks, and about moderation. I do want to talk about all those things.
Right now, the basics of the community are that you download the Mastodon software from Mastodon the company. You can spin up your own server and try to get your own users for it. Then you can, in a federated way, send messages back and forth across these servers.
How many users does Mastodon have now across instances?
God. I am actually unprepared for this question because the statistics API on our official website has not been functional for a couple days. I need to change the code a little bit to make it more efficient. I have actually not kept up with it, but I would say at least 1.5 million monthly active users.
Okay. In November, when Elon took over Twitter, you saw a huge spike. What does that spike look like?
That spike was 2.5 million monthly active users.
So it has gone down since then?
Yeah, but there’s a caveat to that spike. The way that active users are counted on Mastodon in the software is when you log in, and every person who signs up logs in. It’s kind of inflated through people who sign up and then bounce, so it’s natural for it to go back down to more natural levels. It’s just an unfortunate side effect of how the statistics were implemented in it. It’s impressive because the big number highlights how many people checked out Mastodon, but it’s also normal for that number to go back down to a more level area again.
Are you still growing?
Yes, it’s still growing, and in fact it’s growing faster now than it was before because of some changes we did to our official labs, as well as the changes we did to the official joinmastodon.org website. I adjusted the sorting to highlight bigger servers first. It’s a big change from what it used to be, where we tried to highlight smaller servers. A lot of them at the top were the ones where you can’t create and cast straight away, but you have to wait for approval. That was just not very user-friendly. Now the ones with larger open registrations are at the top.
Of course, that has had a very quick effect on increasing the growth. Similar changes have been made in the official apps, as well as the onboarding improvements in general that we’ve been working on since November. They’ve finally been released on both iOS and Android. They include both a better explanation of what the hell decentralization is and what the servers are before you get thrown into the server list. Also, changes to the server list allow a bit more sorting and displaying information in a better way. It also provides a “pick for me” option, where if you don’t make a selection, it gives you a random one.
I have an account on mastodon.social, which is the one that you run. By the way, I follow Eugen. He’s not kidding with the cats thing, in case you’re listening. That was 100 percent true. But there’s all sorts of other servers.
What are the economics of running one of those servers? To me, that is one of the most difficult things to understand. I spin up a Mastodon server on Squarespace, or whatever hosting company wants to let me do this very easily. I run it, I get users, and I have to pay Squarespace. Do I, the administrator of the server, make any money? Is there anything built into the platform that lets me make any money doing this?
“When you host Mastodon for yourself, you’re in control. You own it. It is your megaphone.”
No, not really. It’s not really about making money off of hosting Mastodon. It’s that when you host Mastodon for yourself, you’re in control. You own it. It is your megaphone. Nobody can kick you out, tell you what you can or cannot say, or take your data and sell it to a third party. It’s your complete ownership over your own social media platform, without losing global reach because people can follow you from other servers.
Hosting a Mastodon server for yourself, your family, or a small, private group of people that you know is really relatively simple and not too expensive. You do need either a hosting provider, somebody who will offer Mastodon as a service, or a small virtual private server from DigitalOcean or a similar hosting company. If you have the know-how to install programs on Linux and run them, you can just install Mastodon; you just need your own domain name, and there you go. When you run the server for yourself, you don’t really need to think about things like moderation, because you’re just responding for yourself. Obviously you’ll need to block bad actors and stuff, but you’re just doing it for yourself, you don’t have to worry about community management.
Then a completely different side from this is when you want to run a server for a community or the public. It’s a whole different level of responsibility, and it obviously involves moderation burdens. You need to moderate or find people who will moderate for you. Sometimes for smaller communities, people just do it out of their pocket, because the community matters to them or it’s an offshoot of their main community website, like if they have a forum and get a Mastodon server as a bonus value add to that.
We could actually talk about Medium as an example of that, because Medium has spun up their own Mastodon server. Now, they are a publishing platform for long-form articles for writers. What do writers do? They usually announce what they’ve written on social media, in short form content. Mastodon is a perfect fit for them to basically offer it as a value-add. You have your Medium account and you can sign into Mastodon to post about what you’ve just written on Medium and drive traffic to your writings that way.
That’s one example of how it fits with something that already exists. Of course, for other cases, the Mastodon servers that aren’t monetized — if that’s the right word — are through Patreon. Basically, people are providing a service, and then some people who value that service donate back to keep it afloat.
So mastodon.social is the one that Mastodon the company runs?
That is, I would say, the most famous one, for lack of a better word. You open signups, you close signups. Why do you sometimes close signups?
In the past, it has been a huge burden, especially when I was working alone, to wrestle with the scaling problems or the technical issues of running a large-scale server at the same time as running a company and writing code. That’s just one side of it.
The other side is that, ideologically speaking, Mastodon is a decentralized social network. We don’t want to promote a single node more than other ones. The ideal system is one where there are a whole bunch of different servers, all roughly the same size, and it doesn’t matter which one you use, because they’re all interconnected. Because when you get to a situation where one single node is much, much bigger than all the others, the problem is that it gets disproportionate power to change things and enforce its own whims.
An example of that is email and the situation with Gmail. Gmail is huge. They have very good spam filters, but unfortunately, those good spam filters sometimes catch people who try to self-host email. It creates a situation where it’s so difficult to self-host email that you’re basically forced to just go use Gmail or one of the other large providers instead. That’s the kind of situation we ideally want to avoid in Mastodon and in the fediverse, which is the name for the network that Mastodon is basically a part of based on this activity protocol.
We want to avoid that, so for that reason, we have historically been trying to promote a healthy distribution of people across these different servers. However, I’ve learned over the years that there is no replacement for having a default, right? I’m sure you realize. When people are used to just going on a website and creating an account, presenting them with a choice of hundreds of different servers, all provided by different people and organizations that they don’t know, it is quite a paradigm shift. It’s quite difficult for people to grasp at first sight.
“The power of somebody like Elon in the fediverse is greatly diminished.”
Really, it is a strength of the platform that there is such a diversity of offerings. On Twitter, you just have Elon Musk. If he decides to do X, there’s nothing you can do about it. On Mastodon, you still have the potential for a mini Elon in some ways, but unlike Twitter, you have the option to just take your account and go somewhere else, or to start your own, and still participate in the network. The power of somebody like Elon in the fediverse is greatly diminished.
That’s the strength of it, but to realize that strength, you basically need to be into the idea already. For that, we need a simple conversion, a simple signup that people can go through without too much hassle or being presented with a completely different way to do things that’ll make them say, “It’s too complicated, I can’t choose.” That’s what I’ve realized over the years. The idea right now is to keep registrations open and continue having them open to the best of our ability.
The only situation I can think of where we might close signups temporarily is if there are technical issues that affect quality of service. If lots of new people are joining and the servers are melting down, the priority for us is to ensure that the people who are already using our server can continue to do so with good quality of service. If that means closing signups and directing people elsewhere, then so be it. But the way we are expanding the company, hiring a dedicated DevOps person and getting new funds, I think will allow us to keep registrations open going forward.
The Gmail comparison is really interesting here. I cannot believe I’m about to explain Gmail to the Decoder audience, but I’m going to do it just so I’ve said it out loud. Email is an open protocol that is run by standards organizations. Gmail runs on those protocols — SMTP, IMAP, and the rest — and so does Outlook or whatever. The reality for most people is that there’s a collapse between the protocol and the application on their phone. If you have Outlook, you are almost certainly not using the Gmail app. If you have Gmail, you are almost certainly not using the Outlook app. You’re going to the service and putting that app on your phone. The only real exception to this rule is the Apple Mail app on the iPhone. With everything else, there’s a collapse between the protocol and how it’s expressed to the user.
mastodon.social is the one you run. You download the official Mastodon app, and it’s going to default you into it because defaults are important. I mean, literally, the number one criticism I hear from people is, “Well, you have to pick a server and it’ll never work,” because no one wants to pick a server. You’re going to solve that problem, but aren’t you now getting closer to that collapse between the open protocol and the user experience, where people download the Mastodon app and end up on the Mastodon server?
Well, for reference right now, mastodon.social isn’t actually the default in our app. It’s just one of the top ones that shows up. However, I think that possibly going forward, we might rework the onboarding user experience into presenting a default option as well as an advanced option, where all that stuff with choosing a server would basically be hidden away from the people who get intimidated by choice.
Yes, you are correct in that it gets us closer to the Gmail situation. But it’s kind of unavoidable with the constraints of the problem I’ve described, where the choice is too complicated. You need to convince people that this is better and that this is something they should invest some time into. Then they’ll realize both how it works and what they can do with it.
That is the idea. The idea is the funnel. They get started on mastodon.social, but afterwards they can move to an account on their own server that they create or on a different server provided by a different company or person. Historically, that has been the case. A lot of people who are currently running their own servers had their first account on mastodon.social. So it is working, and I imagine it will make mastodon.social somewhat disproportionately large in the future, but that’s just part of making it work, I think.
Is there a point at which the cost of mastodon.social will be too much and you’ll have to actually monetize it directly?
It’s hard to say if the cost will be too much, and it’s equally hard to say if there will ever be a point where some kind of monetization is absolutely necessary. One thing that I’ve looked into as part of our software offering, as part of the Mastodon software somewhere down the line, is a way for admins to monetize account creation, to offer paid accounts like a premium email service would do. I think that’s a very fair way to monetize.
Right now, it’s what people manage to do with Patreon, basically an honor system, and you just codify it into the software. “Okay, I pay you this amount per month, and for that, I receive an account and good quality of service.” That makes sense to me. I can’t say for sure if we ourselves will ever make use of something like that. It’s hard to predict so long in advance, and with unknown growth as well, but it is something I want to make available for other admins so that there can be more sustainable business models in the fediverse.
I realize I keep asking about money, but one of the reasons I keep asking is that if you want to be a success on the scale you’re talking about, it actually needs to become an ecosystem. There need to be multiple kinds of business models and businesses using the protocol and using Mastodon. That’s what makes it resilient to change in a way that MySpace’s business was not. Are you open to saying, “Okay, some people are going to monetize this thing in a way that maybe I personally don’t like or I’m ideologically opposed to, but in order for the ecosystem to thrive, lots of people have to be trying lots of different things”?
Well, I don’t know about something that I would be ideologically opposed to.
You’re one of the few CEOs I’ve talked to that will state their values as clearly as you have, so that’s why I asked it that way.
I think a diversity of business models in the fediverse would be a good thing. I think different people exploring different ways to do stuff in this ecosystem would benefit everyone. The bad business models would fail and the good ones would thrive hopefully. At the end of the day, I think it’s good to be able to support these different business models. I think companies are eager to start building on top of the fediverse, because for the first time in history, it’s an ecosystem where one company cannot just pull the rug from under them by closing the APIs. It’s open source, so it’s built on open protocols. The API literally cannot be locked down in the way that Twitter is doing right now or has promised to do.
When the developers from Tapbots build their Ivory app, the premium paid iOS app for Mastodon, they know for sure that I am not going to pull the rug from under them and they lose their existence. I welcome apps like that being built on top of the Mastodon API, and I welcome other software being built on top of the ActivityPub protocol that integrates with Mastodon. I think it’s very exciting that Tumblr has announced ActivityPub support, that Flickr has expressed interest in it. I was very happy when Vivaldi, the browser, announced their own Mastodon server. Mozilla has announced the same, and Medium. It’s extremely exciting.
Those companies are announcing support for ActivityPub, which is the underlying protocol from Mastodon. We had Automattic CEO Matt Mullenweg on the show. He said, “ActivityPub is great. Fundamentally, this is all blogging. My whole company is blogging, whether it’s Tumblr, WordPress.com, or whatever. We’re going to support this and be in full support of blogging.”
We talked to Mitchell Baker from Mozilla. She said, “Mastodon is really interesting. This is where I’m pointed. First, we’re just going to set up an instance and see how it works.” You can see the turn there for her is building some ActivityPub support directly into the browser itself. Do you worry that, “Here’s a lot of attention from big companies, they’re just going to spin up their own massive instances, or that Tumblr will become a sort of strange new default ActivityPub source”? That we’ll kind of just end up back at Twitter, but with an underlying open-source protocol? That is the Gmail problem, right?
Open source doesn’t factor into the protocol bit. Twitter based on an open protocol would be a good thing, even in its current state. You would be able to follow the people you care about and be followed back by them without engaging with the actual platform. Instead of being locked into one commercial service, you would be allowed to leave and build your own platforms and do something better without fighting against the network effects of a locked-in commercial platform.
I see it as a good development, no matter what, when more platforms are announcing support. It means that the network, the fediverse, becomes more useful for everybody. If we create a future where a whole bunch of social networks are interoperable based on ActivityPub, it’s a great future. Ideally, that is also open source and everything, but even without that, it would still be a better situation for the web than right now.
One of the things about ActivityPub that is interesting is that it’s closer to format agnostic than the other…
It’s very versatile.
Right. Versatile is a good word for it. Just as Instagram is incompatible with a tweet from the jump, or a TikTok is not compatible with a Facebook post from the jump. They’re just not the same thing, and you can see how much work was done internally to retrofit something that looks like TikTok onto Facebook. ActivityPub is more versatile than that. The posts can be long, and they can contain multi-mode media, be that pictures and text, or whatever it is.
Do you think that someone might fork it and make it more specific? The social platforms that tend to win are the ones that kind of funnel everybody towards a particular format that captures the zeitgeist. I’m specifically thinking of Friendster, to MySpace, to Facebook, to Instagram, to TikTok, the funnel has been, “Just point a camera at your face and go.” You’re kind of back to, “Well, actually, writing is the ne plus ultra of the internet. We should just do a good job at it.”
I would argue that this has already happened, but not quite in the way you’ve described it. ActivityPub is the protocol for exchanging information in the programmatic way. In a programmatic way, an Instagram photo maps onto the same structures as a video on YouTube or a post on Twitter. You can map these concepts onto very generic concepts, and add all the necessary metadata as extra on top.
Now, Mastodon is a product built on top of this protocol that focuses on short micro blogs, on short posts with videos or images attached, or polls where people follow each other. But there are already other fediverse projects that focus on different aspects of the social experience, like Pixelfed, which is focused entirely on photos. It’s a photo app, so it isn’t really concerned with short-form blogging, it’s just concerned with photos.
You can follow a Pixelfed user from your Mastodon account and vice versa. Then in your home feed on Mastodon, you get all these photos that people on Pixelfed post; on Pixelfed, you get all the photos that a Mastodon user posts. It maps back onto each other — maybe not always 100 percent, and it doesn’t have to. As you say, it’s so versatile. It doesn’t always make sense. If there is a new platform that does something very, very unique, that would have no way of mapping onto what Mastodon is doing today, and they would not have to share any space. Then there might be another platform that is kind of similar to that one, and they would be able to interoperate using those same semantics.
There’s a variety of these platforms on the fediverse, and Pixelfed and Mastodon are just some of them. There’s another one called PeerTube, which is one where you have channels and you can publish long-form videos on them. Again, you can just follow a channel on PeerTube from your Mastodon account and get those videos in your home feed. You can leave comments, and then those comments appear on the videos. A reply on something like Mastodon, which could be compared to Twitter, really maps well onto a comment on a video on YouTube.
Do you think there’s any danger of one of those platforms saying, “We’ve hit the limit here. We’re just going to fork the Mastodon code and build our own version of this thing, with a separate network that might have its own network effects,” and then that becomes a challenge to Mastodon?
Of course something like that could happen with the protocol. There is an XKCD comic for that. The punchline is that you now have one more standard to care about.
It’s a very famous comic, yes. I kind of walked into this answer.
I think what speaks in favor of ActivityPub and Mastodon is that we have built up this network. We have built up some momentum and some network effects for this protocol, which means that it’s not really in anybody’s interest to start completely from scratch. Of course, it doesn’t stop everyone. There’s Jack Dorsey’s pet project, Bluesky, in which they analyzed all the different decentralized social media projects and decided to come up with their own, completely separate from everything that came before it, again. That stuff happens all the time.
They’ve succeeded in hiring one person after this project.
After two years of deliberations, they’ve published something. Now it’s quiet again.
There’s Nostr, which I think is also Dorsey’s.
Yes, there’s that. As far as I know, it’s just full of spam and people talking about cryptocurrencies. Stuff like that pops up, but ActivityPub is a very promising technology. No protocol is perfect and can never be perfect. That’s important to acknowledge, because you will never be able to make an ActivityPub 2 that will not have some issue that somebody will find and say, “Okay, now we need ActivityPub 3.”
Acknowledging that, you might come around to the conclusion that, “Okay, ActivityPub is here. A lot of people are using it. We’ll stick to it and make it better.” You can make it better, because it’s a technology that can be extended. There are these basic concepts, basic semantics, but if there’s something that doesn’t map onto those semantics, you can progressively add them on top of it for new features and new concepts. Then other people that interoperate with you can decide, “Okay, we want to display that information in our app as well, so we’ll start supporting that.”
I have wanted to ask this set of questions, because I’m very curious how it works and how you think about making it work. The technology, and how you personally think about who might use it and why, is all based on the central notion that one company should not be in charge of social networking. You should not have a Twitter, a Facebook, or even a YouTube that is in control of how we express ourselves online. I ask these questions because I never really hear people ask them. I only hear people ask about the next set of things, which is content moderation. I feel like you can’t quite understand the content moderation aspect of Mastodon without understanding the governance of the protocol — but now I have to talk about content moderation. The point of this, as far as I can tell, is that you don’t think one person should be in charge of content moderation.
So, anybody can start a server, and they can impose any rules they want. Some of those rules can be very strict, and some of them can be very permissive. There’s no central place to impose rules from your perspective.
As you come to a place where Mastodon socially becomes the default, do you feel any burden to change the content moderation rules for the default server?
We haven’t identified any issues with the rules we have right now. For the most part, our users are happy and our moderation load is bearable. We have a bunch of paid moderators working on our server, and as it grows, the moderation team will obviously expand further and further. In the sense of rules, I think it’s quite fair to keep them as they are. They are on the safe side of the gradient between permissive and safe, but not too restrictive. I think it’s a good balance we’ve stricken over the experience of seven years of running that server.
This is the challenge, though. Small communities are relatively self-sustaining and self-moderating to an extent. As they scale, things get out of control very quickly and people start doing very bad things very quickly. We have written about the challenges moderators face on skilled social platforms for years now. Is that a set of challenges you’re facing? Are people doing the worst possible things with your service? Do you have automated tools? How does that work for you?
I would say that we’re lucky enough that, due to our smaller size, we haven’t had to deal with the kind of horribleness that, let’s say, Facebook moderators have to deal with. Honestly, most of our reports are either just generic spam or people being mean to each other. It’s not the horrible stuff that Facebook moderators need therapy for. In that sense, we’ve been lucky.
To answer your question on automation, we have avoided it so far because we believe in the personal approach. What allows this personal approach is the decentralization and the different size constraints. First of all, back in regards to the rules, it is actually quite liberating in a sense to have this decentralized network. You can say, “Okay, these are the rules of the service that we offer. If you’re not okay with it, that’s fine. You can go somewhere else where what you want is allowed or you can just run it on your own.” We don’t have any obligation to allow you to post this or that if you can just do it yourself on your own money.
I just want to point out that this hypothesis has never been tested: if you give people a choice of moderation regimes, they will pick one that offers the most safety, and they won’t pick one that has the horribleness, because right now everything is so centralized. This is a large-scale test to see if the market-based version of content moderation actually succeeds in the way people have always said it would.
Yeah. Going from that point back to moderation, this decentralized system also allows sharing the load of the moderation burden between completely different systems and completely different organizations. If you count a person running a server as a moderator, then the lower bound of the number of moderators in the fediverse is something like 9,000 or 10,000, because that’s the number of servers there are.
If you count a person running a server as a moderator — the fediverse has a much higher ratio of users to moderators than any other commercial social media platform.
Of course, for the single-person servers out there, it doesn’t really matter. As I said, they’re not really moderating, they’re just running their own personal platform. But for all these communities, if you have 20 people and one moderator, that’s a much higher ratio of users to moderators than on any other commercial social media platform. This decreases the moderation burden for everybody, because for the most part, people take care of the rule-breakers on their own servers and the other moderator doesn’t have to deal with it.
This leads to challenges, because what you want is a megaphone, right? If you’re a writer on Medium, when you write the thing and post it to the Medium Mastodon instance, you’re not trying to get to 20 other Medium writers. You’re hopefully trying to address millions of people who follow you across Mastodon instances. They all might have different rules. If you write about some hot-button issue that breaks a bunch of rules, you won’t actually reach all those people. Those communities also have to moderate themselves. You kind of end up at a place where it’s harder to know what is going to happen because there are so many fragmented rules, as opposed to, “I’m just mad at YouTube all the time because YouTube is too opaque.” I’m wondering if you see a balance there that actually results in something different. Right now, I think the frustration most people have is that they don’t know what’s going to happen.
I would say that content moderation is one of the hardest problems in social media. There’s no denying that. We can add features, we can remove features, we can improve how a button looks, whatever. Content moderation is always going to be at the core and center of what makes a social media platform work. So yeah, it is complicated.
But to counter your point, if you publish something about some hot-button issue that a lot of people will disagree with, they might not see it, even if you have a centralized service where everything is allowed. Because they’ll block you, they’ll filter out the word, or they’ll use a block list that they’ve made themselves and share around. That sort of thing is not unique to the fediverse and to the decentralized system of moderation. If for whatever reason you get blocked by a server of 20 people, well, that’s part of it. Just imagine you were blocked by 20 people for writing that issue.
Now, I’m not saying that everything is perfect. For example, I think that there needs to be more transparency for users about what is happening with moderation on their servers. I don’t think it’s great that your admin can make a decision to block another server, and then suddenly you lose a bunch of followers without knowing about it. I think that we need to build some kind of notification system into it that will tell you, “Okay, your admin has made the decision to block this server, and you’ve lost this amount of followers. If you disagree with this, here’s what you can do.” As long as you have that, then people can, again, move to a better server or start their own. Nobody really suffers, right?
I agree with you. I think content moderation is the most complicated part of operating a social network. My thesis is that it is fundamentally the product. Any Mastodon instance is only distinguished from the next because it has slightly different content moderation rules or a slightly different community, and that community might self-enforce those rules.
What I’m curious about is that moderation is the cost, even beyond server costs or whatever. As your community scales, moderation and legal compliance become a cost that is almost unbearable for anyone except the largest companies in the world. Facebook has to run moderation services across the world, they have to be in legal compliance across the world. Right now, as we speak, there is a Supreme Court case about Section 230 and Google’s recommendation algorithm on YouTube. I don’t know how that’s going to go.
I can point to two specific Mastodon examples here. The Financial Times, I think, started a Mastodon instance and shut it down within days, saying, “Okay, just the legal compliance costs of running the server are too high for us.” The other one, which was much funnier than that, was a Harry Potter server that shut itself down, because there was an argument about Harry Potter and the community got out of control. There are a lot of arguments about Harry Potter. That community is just fully out of control right now.
If I start a server and it gets moderately popular, then suddenly, as an administrator, my costs in money, time, legal fees, or whatever it is, start to skyrocket in a non-linear way to my user base, my activity, or my financial return. Is there a way for you, at the protocol level or at the governance level, to bring that back in line? Or is it, “Well, someone’s going to have to figure out how to monetize this to make those costs work out”?
First, to address the Financial Times, I think they made the fundamental mistake of trying to run a public server instead of just running a server for their own journalists and people who work for them. The moderation load, even the technical investment required, is completely different for these models. I’m not going to recommend everyone start a public Mastodon server and start accepting registrations from the public and get thousands and thousands of users, because it is a huge responsibility. It is a responsibility that comes with a cost and potential liabilities. It’s not easy.
“Mastodon is a shortcut to starting your own Twitter for the public if you want, but it comes with all of the costs, problems, and risks of running a public service like that.”
Is it easy to start a social network from scratch? No, it absolutely is not. Imagine starting something like Twitter completely from scratch. Mastodon is what allows you to do that if you want to. It’s a shortcut to starting your own Twitter for the public if you want, but it comes with all of the costs, problems, and risks of running a public service like that.
Now if you just run it for yourself because you’re an author, a blogger, a publisher, or an enthusiast, and you want to consume content from other people on your own terms, it’s a completely different model. You’re not really liable for anything, because it’s just for yourself. You’re not really risking much, and it doesn’t cost a lot. You don’t need to think about moderating or hiring moderators.
For the servers that do run publicly and accept registrations from the public, yes, moderation comes at a cost. Now, of course, as I mentioned before, the cost of moderation and the burden of moderation is shared across the entire network. It is a lot lower than if a single company was doing it for millions and billions of users. If all you have to manage is a community of, let’s say, 5,000 people, unless you have a lot of churn — which means you have a lot of new people joining all the time — for the most part, the people on your server will be well-behaved, because the bad ones will already have broken the rules and be suspended.
Can I push you on this? I believe the same thing that you believe, but I’m not the CEO of Mastodon. Do you have data that says that is actually true? I feel like that is the lived experience of being on the internet. If your community is small, it will self-regulate, but when it gets big, it goes nuts. If I’m making this bet, I would want to know that that’s true. I could provide you with anecdotal counterfactuals, right? I could tell you to look at Reddit, which is a history of small communities fracturing into rival subreddits, and that will go on for infinity. So is this something that you believe? Is this something that you know? Is this something that you can measure?
Well, it’s based on my experience running mastodon.social and the moderation loads that we get. Most of the reports that come to us are usually about people who have just signed up and don’t belong here. They’re the people who don’t actually agree with our rules and break them straight away. During times when we had closed registrations, the load on our moderation team was a lot lower. For the most part, it’s quite straightforward really. The people who break rules show themselves very quickly. They get banned, and then there’s nobody left to break the rules.
I feel like people sometimes overestimate the amount of reports and the amount of moderation necessary that comes with a specific server size with a specific amount of users. Most people don’t regularly break the rules. It is still a minority of people who do that and a minority of cases where it happens. It only increases when you have lots of new people coming in, and possibly if you have a troll army that decides to make a bunch of accounts, which can also make the numbers go up. Otherwise, it has been quite manageable so far.
As you look across the spectrum of Mastodon instances, do you actually see a diversity of rules? Do you see that, “There’s a spectrum. We have the most permissive sites and the least permissive sites”?
There is a spectrum, but I think for the most part, the majority of the servers that run Mastodon today have agreed on a certain set of rules — probably because a lot of them have been inspired by mastodon.social and our rules originally. There is a lot of overlap. Some are a bit more restrictive about specific things like content warnings, the kinds of self-promotion you can do, or whether you can run an account as a company or a brand versus as a natural person. Sometimes people do get restrictive with those sorts of things.
Obviously, there is a side to the fediverse that is completely the opposite and leans towards permissive; everything goes, free speech, everything. For the most part, it is ostracized and distanced from the previous side. A lot of different approaches exist out there.
Are they growing in parallel? Is one growing faster than the other? How is that playing out?
Well, I believe that our part grows faster because they get traffic from joinmastodon.org, our official website, and our official apps. So that helps.
Do you worry that the other side, the most permissive libertarian side, is going to say, “This is unfair, you need to list our servers as a default as well”?
Well, there is no real obligation for us to link to anybody. Sometimes, yeah, I get people that are very upset with me for not giving them a free link from our official website, but we’re not really obligated to provide a link to anyone. We just do it to make the sign-up process easier for end users. We’re only going to link to a place if we think that people are going to be safe there and have a good time. If your server is allowing everything and all types of language that is completely inappropriate, then they’re not going to be safe there and we’re not going to send them there. So, there’s that.
Eugen, you’ve given us a ton of extra time. I want to say thank you. I’m very excited to see where Mastodon goes. I did sign up over the weekend. Decoder listeners know I took a break from Twitter-like networks to heal my brain after 10 years, but I signed up for mastodon.social. I’m excited to use the product. I have two feature requests for you. One, it’s slow. Can you make it faster?
Yes, I think so. That would normally not be a software problem, but an infrastructure problem, something specific to mastodon.social. Honestly, depending on when you accessed it, it could have been a DDoS attack, because that happens. Just last night, I was looking into a traffic spike seven times higher than usual that was causing a lot of errors to appear. So, if you’re complaining about slowness, it could be that. I would be happy to report that we are upgrading our infrastructure completely from what we were using before.
Our infrastructure for mastodon.social basically grew out from a single box, originally in 2016. We moved providers — I say “we”, but back then it was just me — quite a few times during these seven years. I’m trying to remember where our original box was. Probably DigitalOcean, actually. We tried using Scaleway, and then we eventually moved to Hetzner, the German hosting company, which is where we are right now. Basically, it grew out of a single box into a cluster of 15 or 20 different machines manually managed by me.
Right now, we’re upgrading to using Kubernetes, and that simplifies a lot of the scaling significantly. Where before it would take half a day to set up a new machine, now it’s like, “Okay, it’s a couple clicks and a couple of keyboard button presses,” and now you have the capacity for a lot more traffic than you had before. With that, plus having a DevOps person on the case, it will hopefully improve the speed by a lot.
My second feature request actually has me curious if you feel the pressure to look and feel more like Twitter now than you did before. Can you do quote tweets? I often want to quote tweet someone from The Verge and say, “This is great, look at this.” I know that you’re opposed to it, but are you starting to reconsider?
We could do a whole episode about quote tweets if I start talking about it now, describing the history and the positions people have on this feature. I’ll keep myself short. We have a public roadmap on our website, and quote posts are under the exploring section of that roadmap.
The most complicated section of all.
It means that we’re looking into it, and hopefully we’ll bring you quote posts.
I like it. What’s next for Mastodon? What should people be on the lookout for next?
Groups, probably. That’s a big feature that’s coming. Well, it’s what you would expect. Facebook has groups, Twitter has communities, and Mastodon will have groups as well.
Very good. Eugen, thank you so much for being on Decoder. We’ll have to have you back soon.
Thank you, and have a good day.
Decoder with Nilay Patel /
A podcast from The Verge about big ideas and other problems.