DHH: Linux Can Finally Beat the Mac, and Apple Can't Stop It

The following is a conversation with David Heinemeier Hansson, also known as DHH. You know that David has been working very hard on Omarchy over the past year or so, and recently Omarchy has been gaining lots of popularity, and there's so much happening there. You also know that I made fairly critical videos about Omarchy in the past, but I'm a man who likes to have strong opinions weakly held, and Omarchy is getting to a stage where it becomes impossible to ignore. They raised over 20 million in funding, they're setting up a security team, and I think in the history of desktop Linux something truly unique is happening here.

So I invited David because I wanted to know more about what is actually driving him to do all of this: why is he building Omarchy, and what is happening right now in the project? We also discussed security and much more. So yes, I have changed my tune on Omarchy, and in this podcast you can find out why. Enjoy.

Everything, all at once

David, welcome to the show. Every time I open X these days, I see just you and Omarchy, Omarchy, Omarchy. So there's lots of things going on. Let's start there. What's going on in Omarchy right now? What are you working on?

Thank you, first of all, for having me on. And second of all: everything. Everything all at once is going on, because we have spun up so many sub-projects within the Omarchy fold that even I am having a difficult time keeping up, and I literally spend 16 hours a day most days on this project right now. It's simply incredible to see the outburst of creativity and enthusiasm, and just people having fun with computers in a way where they experience something they haven't in a while. Especially for folks who've used computers for a long time and originally had a sense of wonder when they got into computers, and then maybe that sense of wonder got a little lost along the way, and now they found it again with Omarchy.

So the creativity that we're able to combine from the artistry side (we have a lot of people who are extremely creative, making beautiful themes, making all these designs) and marrying that with a hardcore engineering effort to secure the system, to make it more performant, to build our own bespoke kernels, what have you, is quite special, I think. Because if you look historically at Linux, there's been a lot of people who were very strong on the engineering side and could build very fast and secure systems, but, I hope I don't offend them here, they weren't always the best at color coordination. They weren't always the best at making things look pleasant, because they made things look functional, because that was what they were interested in. I think what's going on with Omarchy right now is that we're bridging the gap between those two worlds, between art and engineering, and we're getting something very unique.

And that's before we even go into the specialness of the Omarchy Quattro release, which is of course the focus on the agentic OS: something you can change even if you're not a programmer, even if you're not a Linux specialist. You can now mutate your malleable operating system into becoming whatever you want it to be. In just the first three weeks, I think, we had 3,000 plugins created. That's 3,000 mini apps, or major apps in some cases, created by people who in many cases never programmed. They were simply vibe coding, as we were calling it five minutes ago. I think that term actually needs to retire, because at this point even the individuals who are very versed in this technology are vibe coding, and there are some negative connotations associated with it that do not befit the current intelligence we have available in the models.

So put that aside. The excitement is really around just enabling so many more people to create things with their computer, to make it their own, to share things that they come across, and forming a new type of community too. I think if you look at both the Linux community and perhaps also the larger tech community, there have been times where it's been a little rough. The historic norm here was RTFM, read the fucking manual, and a lot of people get turned off by that, because the atmosphere is just not that pleasant.

So I'm trying to create both a beautiful, malleable machine that's actually the perfect computer, at least for me, on first installation, but also a community that's full of people who are simply happy about what they're building and want to share, and don't mind that there's new people coming into the community who don't know everything, who know nothing in fact, and are still willing to help them and be excited about things. Really focus on the progress we've been able to create together, the optimism, and not be pedantic all the time. Just: here's a space on the internet where we're going to be happy about computers, we're going to have fun with them, we're going to make them beautiful. And then you can go somewhere else if you want to, I don't know, be depressed about it.

Okay. Well, that's already a lot to unpack. I think anyone who has been following you for a while now, or recently, has seen the clowns and the circus and the things that they said. But you do have a point there: some of these projects are surrounded by, let's call it, drama. There's people interacting with each other in ways that have nothing to do with the tech side of things. So yes, I think that's a very inspiring and interesting thing that's happening around Omarchy. Am I saying "Omarchy" right?

You got it right. Omarchy. I even have a special pronunciation website to keep the "r" silent. Exactly: Omarchy. You can call it whatever you want. There's "jif" and "gif," and a lot of other words that go back and forth with different pronunciations. Some people call it Omarchy. Fine, whatever you want to call it. I call it Omarchy.

Why a Linux distro

I think that is one of the coolest things about what's happening, because you have such a large brand, and your energetic personality, bringing that to Linux and the community around it, is something very special. So I want to ask you: you've been open that money is not your motivation anymore to do things. That is kind of settled for you in your life. So of all things in the world that you could be doing right now, where you could race cars all day if you wanted to, or lie on the beach, how did you end up putting so much energy and time into a Linux distro? Why this?

It's a good question, and it has a simple answer. I just love computers, and I've loved computers since I was five years old. I got my first Amstrad 464 back in '84, and before that I spent a lot of time in front of a Commodore 64 with the other kids on my street. And from that first moment (Kung-Fu, that was the first game I played on a computer) I was just enthralled. This was the most interesting thing, in my opinion, in the world for a very long time. Now, I've found other interesting things, whether it's family or racing or what have you, but I have just this deep love of the computer. I just really like using them. A lot of times it has a purpose; I've been able to make a career using computers to build things. But I also just like using computers for their own sake. I just like sitting in front of computers and doing stuff with them.

So when I discovered Linux, and the access you have to the bowels of the system, how much you can change, that really clicked something in me, very similar to what happened when I discovered Ruby back in 2003 and started building Basecamp, which then became Ruby on Rails, the framework that underpinned it. It clicked something in me where, in the Ruby case, I suddenly went from "Well, I'm just using programs because I want programs; it's a tool, it's merely functional, I'm seeking an outcome" to falling in love with Ruby as a programming language, and as a way to express your ideas and your visions about software in this particular language.

And I didn't really have that with the computer. I was a Mac user for 20 years, and I liked that a lot. I think Mac and Apple have some really good instincts about what makes good computers, and they have good taste. But I've also found that the kind of company Apple has become in later years was diverging from some of the things I originally liked about the Mac. Now it's less about empowering the users, more about controlling the whole flow, having your toll booth sit in front of all software distribution, collecting 30 percent of the whole damn economy. And I just found myself drifting away from that.

I also had some skirmishes with Apple over the years. We launched an email system called HEY.com back in 2020, and Apple tried to kill it if we didn't hand over 30 percent of our revenues. And you know what's funny? Even with that, I couldn't even imagine using something else. I'd been using the Mac for so long, literally at that point nearly 20 years, and however much they insulted me or tried to damage our business, I still couldn't see that there could be something else, because at the time I really only thought that the alternative was Windows. I'd used Windows before, and I even gave it another try, and I realized it's just worse in all the ways. I'm just not a Windows person. I used Windows for a few years back in the '90s, when that was the only way to play video games, but I didn't want to run it as my personal computer.

And I never really considered that Linux was a third option, because I'd been using Linux for, I don't know, 25 years, on the server always. But whenever I would encounter the Linux desktop community, I thought I saw a bunch of nerds that I could not see myself in. They were interested in other things, and they didn't care about font rendering and color coordination or ease of install or curated defaults, or any of the things that I always care about, and what I liked so much about the Mac. They really cared about those things; they really cared about the design, they really cared about the fonts, they really cared about the defaults.

Duplo and Lego

But eventually push came to shove, and Apple put the final straw on the camel's back and it broke, and I went to Linux. First Ubuntu. I gave that a try and found a lot that I liked. I didn't actually love Ubuntu out of the box, so very quickly I tried to personalize it, and I tried to build it up in such a way that it was something I could replicate across machines. That was what became Omakub. That was the first kind of dabble I did in this. I ran that for about a year.

And then a little over a year ago I got snookered by YouTube. This was from Typecraft and ThePrimeagen. They just had a lot of videos on other ways of using Linux, using tiling window managers and whatnot, and I thought, I'll give it a go. I tried Hyprland on top of Arch, and in one way it was the opposite of everything I want out of a computer. Everything was an atomic bit I had to put together myself. It was just a Lego set dumped on the floor, and then an invitation to build your own Millennium Falcon. That's not typically what I want out of a computer.

But what I realized was: if I want the perfect computer, I have to start from the smallest Lego blocks, and the blocks that Ubuntu gave me were not small enough. They were Duplos, not Legos. You could build something out of Duplos and you're going to have fun with it, but most kids grow out of Duplos by age five, maybe, at the latest, and then they like Legos, because they can build more things with Legos. And I realized that transition from the Ubuntu Duplos to the Arch and Hyprland Legos was the unlock that I needed to fall in love with the Linux machine like I had with the Ruby programming language.

And once that happened, I realized I now had a straight shot to building the perfect computer for me. Then I spent a year on Omarchy in that shape, getting it up to version three. First we went from just the dotfiles, then we had an ISO, then we had our own mirror, then we had our own package repository, and we just kept accumulating all the terrain that a large distribution would use.

The agents

And then something happened, and what happened was that AI got unbelievably good in about the spring of this year. It was already really good in December of last year, November last year. That was when Opus 4.5 came out, on November 24th. That to me was a real inflection point. And what's been happening the last three, four months is pure magic. I have not written a line of serious production code for about four months. Everything is being created by the agents, because they now write code that is arguably better, the majority of the time, than what I could write myself, as a professional who's dedicated my entire career to this for 25 years.

Wow.

The agents are just better, especially when you use them in adversarial mode, and you have Codex and Fable, or whatever other models you use, and you use them to critique each other. The kind of implementations that they can produce are just truly magical. And especially in Linux, because all these models have pre-training that includes the 40 million lines of Linux code, so they know everything. They know every problem, where to look, where to diagnose it. And I have not had a single issue with Linux, probably since November, that an agent couldn't fix.

That was also one of the other things with Linux historically that I didn't actually appreciate so much: whenever it would break, which would be somewhat frequently, it would spit out an error message about a config file in a sub-package I had no clue about, and then I had to go trawl the internet and find some forum where a passive-aggressive nerd would have the answer, but not really want to share it before they demeaned you and shamed you for being an idiot that didn't read the source or compile it themselves, or whatever else they thought the bar was for participation. And agents cured all that. Agents are very polite, and when you ask them about a problem in Linux, they will fix it for you and tell you why they did it, without any prejudice.

So I just thought: hey, you know what, that's a good time. You combine all of that together, and now I'm obsessed. Now I see that not only can I build the perfect computer with Linux in theory; I can build it in practice. And that's what Omarchy has inspired, I believe, in so many people: they can see what I ship as a bunch of really good defaults. They don't have to figure everything out; everything's set up, everything's beautiful. But as soon as they learn it, they can take that base and turn it into their perfect computer, their mutation, their aesthetics.

And in that space there's just a lot of excitement, and a lot of what the business world would call blue ocean. There's not a lot of competition, because if you look at the historic alternatives, Windows and Mac, they're incredibly hostile to the agentic OS idea, because they're very locked down. Windows is a mess in all regards. You have to use computer use through your agent, which is literally moving a cursor around to click things. It's totally archaic. And then Apple, on the other hand, has some stances around security, actually, which we could talk about, and other things that just make it very difficult to mutate the system.

Linux is the perfect fit for AI. Everything is a CLI, everything is a config file, everything can be scripted, everything can be coded, everything can be captured, and therefore everything can also be debugged and evolved. So we have this perfect storm, and then you'd think the Linux world would embrace that. But no. Much of the Linux world, at the very best, is neutral. They're like, "Well, whatever you want." And then there's a huge portion of the Linux world that's actively hostile. You look at the GNOME project, you look at the Asahi project, you look at a bunch of projects that have in their articles of incorporation: no AI, write it all by hand.

And I just go: okay, well, if the old systems aren't capable of getting to the agentic OS, and the other Linux systems by and large don't want to, we actually have a moral obligation to deliver Omarchy for the people who do want to run Linux and who do want an agentic operating system.

The beginner's mind

The phrase that comes to mind is: that escalated quickly. Just to rephrase it: you went into Linux, you discovered it, you were on YouTube ("Oh, this is a tiling window manager"), and then you went into it, and then, what is it, one year later you're now here building the perfect computer. I think that's just so amazing.

This is the gift. This is the gift of the beginner's mind. This was exactly the same thing that happened with Ruby. When I came to Ruby, there were some other efforts at making web frameworks and whatnot, but I came to it as a novice, as a beginner. I didn't know all the things you were not supposed to do, that others didn't think you could do, or that, once you've been in a community for a while, you just take for granted.

When I arrived at Linux, I saw a million things that were broken and could be better, and then I saw a million things with an amazing potential, and I thought: I have to capture all of that right now. If I stay in Linux for three years and I don't make the effort to change things now, I will have forgotten everything that's difficult and that prevents other people from adopting Linux. So there's this magic moment when you arrive in a new technical universe where you can see all the flaws that none of the people who've been there for a long time are able to see, because they've acclimated to them.

Now, they know more things. I didn't know anything at all, and I had to learn a lot of things. But you know what? If you spend eight to 16 hours a day on something, eventually you'll pick it up, if you're smart and driven, and I'd like to believe I'm both of those things. So I applied the intensity and the focus: let me just learn everything there is to learn. That's not possible, but enough to be dangerous, let's call it. And then let's combine that with the fresh memory of what a Mac is like. Why can't Linux beat the Mac on ease of use, great design, font rendering, all the things? Of course it can. We just have to decide to do it.

Linux on Rails

So what sucked you in was building the perfect computer, right? That's what you are basically doing for yourself. And you said in other interviews that Omarchy will always have "this curated set of visuals and programs and software that I like, because I'm initially building for myself." So you could just do that for yourself and not even talk about it, right? But there is so much drive and energy and 16-hour workdays behind this. So what is the layer underneath that goes from building the perfect computer to what you are doing now? What is in between?

I feel I have an obligation: to world productivity, to world appreciation of beauty, to a re-energized set of virtues around computers, of what they're capable of and what they can make us feel. If I have discovered something that enables that better than the options we have today, I must share. This was exactly the same thing I had when I discovered Ruby. I discovered Ruby at a time where it was a very, very small and niche community in the West. It had a large following in Japan and had been around there for a long time, but in the West it was tiny. And I discovered that Ruby was simply the perfect programming language for the kind of applications I wanted to build on the web. And once I knew that, I thought: you know what, it would be incredibly selfish to keep that secret to myself. I should allow others to experience the same joy I'm experiencing with Ruby.

And now it's a replay of that 20 years later. Omarchy is Linux on Rails. Omarchy is my evangelization of discovering better ways to use a computer, and then thinking: I wish someone had told me earlier. Why did it take me so long to find Linux as a way to use it on the desktop?

Why now

Now, there are a bunch of reasons, and some of those reasons are not just about lack of discovery; it's also about timing. I think this is the perfect moment for Linux, for a variety of reasons. A: the tiling window manager concept has been around for a long time. We've had i3 and whatever since the early 2000s, I think; some of these early projects got started then. But it wasn't until Hyprland that I found a version of the tiling window manager concept that was captured with such poise and style and beautiful bezier animations that I could see myself using it. So that was one thing. We had to have Hyprland.

The second thing is that the computers you could run Linux on, up until literally this year, maybe last year, were not very good after Apple had come out with the MacBook with the M chips, because the M chips were a revolution for the favorite computer format of almost anyone, which is the laptop. So we went through this walk in the desert where Linux really did not have great hardware choices. Now, on the desktop it doesn't really matter, and in fact those choices have been great for a long time. But if you want to do broad penetration, even for myself: I want to not just run Linux on my desktop and then another operating system on my laptop. I want to run Linux everywhere. I want to run the same system everywhere. A little difficult to do that before we had really good computers. And now, with Intel Panther Lake in particular, which just came out in the last 12 months, we've gotten something that's competitive.

And then the final piece of the puzzle, which didn't even come fully together until after I'd been working on Omarchy for a year, was this agent revolution. So you take all of these things together. You needed some of the fundamentals. You needed Hyprland, and in fact now the essence of Omarchy is also based on Quickshell. Quickshell is just as important at this point as Hyprland is, because that's what's empowering this one unified shell where you can do all the plugins and all the agentic extensions and so forth. So we needed the software, then we needed the hardware, and then we needed the agents, and all of that stuff has literally only just come together now to deliver the superior combined package.

You could have had individual parts of it. Obviously there's been people running Linux on the desktop for a very long time. But it kind of stalled for a long time, and we just hovered at low-single-digit percent market share, because it was a niche thing. It required an enormous investment up front to produce something that was towards the peak of what Linux could achieve. You could of course always install Ubuntu, but as we discussed, those are Duplo blocks, and I actually didn't think they were superior to the Mac. If it was just stock Ubuntu versus stock macOS, I don't know, it's not that easy. I would probably go stock macOS. And therefore all of this stuff had to accumulate and come together before we could fulfill the prophecy.

That's one of the things I really love about the Linux community: we have these grand tellings. The Linux prophecy. The year of Linux on the desktop was foretold in, I believe, 1999, at some conference. And that makes for a great prophecy. It's a shitty prophecy if you make it and then next year it comes true. But if it's a prophecy that's literally been around for more than 20 years, then it's worth fulfilling. And I think that's the magic moment we're in: all the pieces are coming together. We have the lore, we have the prophecy, and now we're able to deliver on it.

Escape velocity

Amazing. So would you say, then, that it's building the perfect computer, but it's also about getting as many people to use it as possible?

Not as many as possible, I would say. There's a magic line that I like to call escape velocity. You need enough people on your spaceship to sustain it through the galaxy. You can't do it with two people. You can't do it with ten. You probably can't even do it with a thousand. But you can do it with 10,000, and you certainly can do it with 100,000. That's what I want to get to. I want to get Omarchy to be the perfect computer for me; that's mission number one. Mission number two is to get enough people on board with that vision that they like my perfect computer, and then they will turn it into their own personal perfect computer, and together we can sustain it.

But that's never going to be everyone. There's just an enormous amount of people in the world who really don't like computers. Now, I think the reason they don't like computers is that they've been using shitty ones, and if we show them Omarchy, maybe some of them will come over. But we're not going to get all of them, and I don't want to get all of them, because Omarchy is also a bit of an acquired taste. It's like coffee, it's like champagne, it's like 80 percent dark chocolate. You give that to someone who's not tasted any of those things, and the first taste, it's sour: "No, no, thank you." Right?

An acquired taste

I would like Omarchy to stay an acquired taste. This is actually a funny parallel to even the name. I don't mind that it's difficult to pronounce. I don't mind that there's some up-front investment you have to make in everything, from the pronunciation, to learning a handful of key bindings just to be able to find your way around, to coming to terms with a tiling window manager, to adopting Linux as a whole package. That's never going to reach 100 percent. But could we reach 20 percent, 30 percent? I believe we could.

I actually think that if you look at the long history of computers, the current desktop-based paradigm was also an acquired taste. By the time Windows 95 arrived, people were used to WordPerfect in DOS, and it was a little strange: now I have to drag these icons around? People forget that today, because we've been living that paradigm for 30 years. But I saw what was before that paradigm, and I saw how difficult it was for people to adopt it.

And I see what happens to children when they're first exposed to a computer. They don't have these preconceived notions. It's an entirely adult thing to believe that a tiling window manager is difficult. No, no: it's difficult because you have to unlearn a lot of things that you've picked up to use a traditional operating system. If you give it to a child who's not been preconditioned in that way, they don't have any of those biases. And therefore I want to reach all the kids. I want to give them a better, more beautiful, more extendable computer, one that's not locked down, one that they can mutate.

The kind of experience I had with the Commodore 64, with the Amstrad, with the Amiga: all of them shipped with developer tooling built in, and an invitation to use it. With the Commodore 64, for example, when you wanted to play a game you literally had to type, I think it was, LOAD, space, then comma eight, something. I forget; I should look that up. You were programming the computer to be able to load the game you wanted to play. Now, that had friction, and then eventually we got rid of all that friction and we ended up with an iPad. And I don't actually hate on the iPad. I think there are wonderful things about the iPad. But the iPad certainly did not teach kids computer skills. It taught them to be consumers of Netflix and games. There's plenty of things that can be good about that. I'm not at all against video games. I'm not at all against shows. But I am pro-computer. I'm very much pro-technology and understanding it, and I think kids who just get an iPad get a very poor understanding of what a computer can be.

Beating the Mac

And then I also want the adults along who want to come along. Now, again, we don't have to win everyone over. We have to get enough people on the spaceship that we get escape velocity, and we can travel through the galaxy for as far as we want to go. Maybe that's 20 percent, maybe that's 30 percent. But if that's all it is (which, by the way, is totally delusional: when was the last time a new desktop operating system captured 20 to 30 percent of the market? Actually, I think it was Windows 95, because if you look at global stats for macOS, I think they hover around 15 percent; if you look at the Mac in the U.S., maybe it's closer to 20, 30 percent, I think 25 percent maybe), so that was the last one, right? The last one was a quarter of a century ago.

I believe we have the ingredients to do it again: to massively move the market-share pie chart, to make Linux go from wherever it is now. There's conflicting reports here. Some say five percent, some say eight percent, now some say 12 percent. It's hard to know when you have all these agents, and we have Steam Decks and a bunch of other things. But wherever it is, it's third, solidly third, behind the Mac and behind Windows. If we could just beat the Mac, with a just-as-well-designed, beautiful, but also open, also free operating system: now is the chance. Now is the window. The window has been closed for 30 years. It's open right now.

Absolutely love it. Because I'm very similar to you. I love working with computers; I made my career out of it. And as you say, kids, or even the generation behind me growing up now, they're using iPads. They don't even have the concept of a file system anymore. They just think in apps. So I think anything that helps people rekindle, or even be introduced to, what using an actual computer is like is only a positive thing. So I really admire that mission, that you have this as a driving force. I did a tweet once where I said that people get assigned their operating systems by society, and I got a little bit of pushback on that. But I truly believe that in school or somewhere you are introduced to Windows or Mac, and then you think that is what a computer experience is supposed to be like.

Transferring enthusiasm

So when you think: okay, we have this mission, we're going to make it as easy as possible to expose this to as many people as possible, with your brand. What are the next steps? How are we going to get more people to adopt Omarchy, or Linux in general, in the general population?

There's multiple steps. One of the most important is that we continue the transfer of enthusiasm. We get more people excited for it because they see the excitement. This is one of the other big bugbears I find in both the Linux community and the technical community in general: they have an instinctual aversion to marketing. They think anything that has been marketed, or is slick, or is compelling, or is color-coordinated, is somehow vapid or empty or undeserving of attention. I think that's an unfortunate conclusion to draw. It's not completely unreasonable. There are many things in this world that are popular not because they're good but because they were pushed. But don't fall into the rabbit hole of believing that anything that has good marketing and is presented in a nice manner is therefore de facto void of any technical merit or attention. I think this is the trap Linux has been trapped in for a long time.

And I'm bringing a new kind of energy to it. I'm bringing an energy that was honed in the commercial market. I've been making commercial software for decades. I know what it takes to get people excited about things. And in fact it's not even a ploy. I'm transferring my enthusiasm because I genuinely hold it. I'm excited about computers. I can't shut up about them. I can't stop improving them. And once I have something that feels this unique in this moment, it feels like a moral obligation to share. Which is also why I believe it's resonating. Because there's plenty of companies who have lots of money, and they tweet stupid things that no one cares about and get zero engagement, because who wants to just be marketed at? No one wants that. But they do want to be infected with enthusiasm and good vibes and excitement for something new.

And I've committed to that mission fully. This is why, in the pie chart of those 16 hours a day, a large amount goes towards the engagement: towards talking to people on X, replying to folks with 17 followers or 17 million followers, and just doing the retail politics of evangelism. Go out, meet people where they are, hear their concerns, write them down, then fix everything.

We can fix everything

This has been one of the taglines that I've been promoting: we can fix everything. I literally even registered the domain. WeCanFixEverything.com is just a monument to the idea that in this new world, between Linux, open source, and agents, there is no paper cut, there is no dependency, there's no incompatibility that we cannot cure with enough tokens, intensity, and momentum. So I'm going to do that work.

I also just love it. I really love seeing someone discover a better way to use a computer and get that aha experience, because I saw it with Ruby programmers who had been using Java or .NET and thinking that that was the extent of it, that was what programming was. And then they would see Ruby, and many of them literally could not believe their eyes the first time they saw the contrast. They thought that Ruby was pseudocode: not executable code, but the kind of pretty code you would write in a magazine when you're explaining concepts. And when they realized, no, no, this is the actual code, this is what you will write, a lot of them really fell in love with Ruby the same way I did.

And I think the same thing is true on the Linux side. There were a lot of people who perhaps were curious about what Linux had to offer but really just didn't have the time and attention to spend the tens of hours, hundreds of hours, to come up with the kind of beautiful bespoke systems that you would see on r/unixporn. Those things were artisan creations that required artisan hours, and as beautiful as they were, they were very niche. So I want to broaden the appeal of the best technology and the best art that we have in the computing world to just more people. So we're going to do that.

Raising money

Then we're also going to continue to raise money. I like money. I don't need it personally anymore, but I know what an accelerant it can be for projects: that you can invest in people, in projects, in marketing, in all the elements it takes to propel the rocket forward faster. This one is also very important when you approach partners. We already have a lot of commercial partners on board with Omarchy, and the more of them we get on board, the more money we raise, the easier it becomes to have the next conversation with Intel or AMD or Nvidia, or whatever other company the Linux community actually needs to create the best systems for new computers. So I think that's a virtue in and of itself.

And at this point I'm also just doing it for sport. I haven't the faintest clue how we're going to spend the 20 million dollars that I've raised so far. But I think: hey, if we could raise 20, why not 50? Why not 100? And I think there's just value in the sheer momentum large numbers produce. It says: we're serious, we're going to stick around, you can trust in Omarchy to not just be someone's dotfiles. It was just DHH's dotfiles a year ago, and I wouldn't think badly of someone who goes, "Yeah, I don't want to build my home computer around that. He might lose interest in it in five minutes," because that's usually what happens with a collection of dotfiles. They go out of date, they don't stay up to date, and so forth. Now we have 20 million dollars, we have a foundation, we have employees, we're going for the moon here, and more people are interested in jumping on top of that rocket, and therefore we can spread Omarchy and great Linux to more people. So it's a very virtuous cycle.

Everything in the box

Then there's the technical part too. The current version we have, Omarchy Quattro, is amazing. And of course I'm going to say this; this is what I just told you about. I'm hype man number one. And I'm hype man number one because I love the system, but also because I think it serves a utility. So of course I think Omarchy is amazing, and that the latest version is the best version of Omarchy yet. That's the tagline from every keynote of every major tech company. But it's also true. It's really good.

But it could be so much better. Of the 3,000-plus plugins, a substantial number should be in the core. We should be integrating all this great creativity right into the box, such that more people can open more computers and have this amazing experience where everything works. I do not want to hear, "Well, my Bluetooth didn't work, my Wi-Fi didn't work, my RGB lights didn't work," and then the answer being, "Well, go look at a wiki." No, no, no, no. Whatever information is in the wiki, we should be applying to software. I actually think that, as much as people adore the Arch Wiki, and as beautiful a community project as that is, in the modern age it's a failure state. If you have to write down a workaround, in my opinion you should put in the work to just fix the bug, or not even the bug: fix the problem. Why doesn't the webcam work? Well, there's some weird driver you have to do if you're on this version of that machine made by this company. No. Let's fix it all. Let's put it all in the box. Perfect compatibility with everything.

Now, that sounds crazy if you were thinking of this in terms of a small hobby project run by eight volunteers next to their day job, without agents. Impossible. Could not do that at all. But that's not what I'm talking about. I'm talking about the Omacom Foundation, with 20 million dollars in the war chest, full-time employees, hundreds of thousands of community members. Once you have all of those things, and then you combine it with the age of agents: yeah, we can fix everything.

I'm setting very specific targets on that, both because it's fun and because it creates a milestone. One of them is perfect compatibility with the entire range of vintage Intel Macs, going back to the earliest ones that are compatible with GPU-accelerated tiling window managers; I think about 2009. Now, Apple actually produced an enormous amount of variety over those years. From 2009 to 2020 is only 11 years, but I think if you count up all the permutations there's going to be, I don't know, 50 different machines, maybe 100. And many of them used different Wi-Fi chips and different Bluetooth chips and different Nvidia chips and whatever else have you, and it's a bit of a mess to try to figure out how you get it all working. And the standard answer, as I said, was: go to the wiki, try to look up your specific model numbers, see how well it's supported, and if it's not, then live with it. No, no, no, no. Let's just put in the work to make it perfect. And then, once we've done the entire vintage Mac range, let's do everything else. Let's do all the Lenovo machines, all the Dell machines, all the everything. Because once you have agents plus literally hundreds of thousands of people, that's possible.

I'm not saying it's easy. I'm not saying we can do it tomorrow. But I can say that if we apply all of this momentum to it, could we do it in a year? Yeah. I actually think the vintage range is going to be 100 percent covered earlier than that. And one inspiration I'm drawing on there is actually the retro gaming community. One of the things I love about retro gamers is both that retro games are fun and also that many of the people in the community are obsessive collectors. They need every version of every ROM in perfect mint condition, and emulation. And I'm thinking: let's apply some of that energy to perfect-mint-condition compatibility with the entire vintage Apple range. It's kind of like: we've got the N64 library pinned down; there is not a beta version of a game that cannot be run on this emulator; we've done it all. Let's do that with hardware compatibility.

A generation of makers

And then let's continue to embrace the age of agents. One of the things we've shown that we're going to ship in one of the upcoming versions is voice mode. You should be able to talk to your computer, and it should just change, magically. Like: I want this plugin to track my favorite sports team, that plays in, oh, whatever league. That data is available on some community website somewhere, not covered by major media, but I can get an agent on it, and now I can have a little widget up in my top bar that keeps track of the score. Just giving you one example. We can do everything, and we can continue to lower the bar on what it takes for someone to create.

And that's what I'm really interested in. As we talked about with the kids, my interest there is to grow a generation of doers, of makers, that I recognize from my childhood (they came out of that), and that I'm not recognizing coming out quite as much in the iPad age. So there's a million missions, and I could talk for hours about all the projects that we have. If you were to look at the internal Basecamp we have for the Omarchy organization, the Omacom, there's just so many projects firing on so many cylinders. I haven't even talked about the language stuff. We want to have perfect Chinese, Japanese, Korean inputs, everything out of the box. I want to do all of it, all at once, as quickly as possible.

And this is exactly why I think this is such a unique situation that's happening. I've been using Linux for ten years. I started daily-driving Arch Linux maybe four or five years ago. And as you say, it was always just this niche thing, and if you had a problem you had to look there and be embarrassed, that kind of thing. But I think it's such a unique thing now that someone with your skill set (because you've built multiple companies, you have built teams, you know what it means to get stuff done, and you understand marketing), and, let's face it, your means as well, and your network, applies that to a Linux distro. I'm not sure that has ever happened before. So I think that is something that's truly unique, and I really admire that you dedicate all of your time towards this, because I can only see this having a positive effect on Linux as a whole: getting people to use computers, to fall in love with computers again. I've made some moderately spicy videos about Omarchy in the past, but I really appreciate that we could have this conversation, so I understand a bit more what's actually driving you here.

Security

So if we then think of broader adoption, getting kids to use Linux: for me, as a DevOps engineer, I start thinking immediately about security, because with more freedom there also comes more responsibility, right? I know that you've raised 20 million now, and part of that is going to a security team. We've had some discussions about security on X as well. So before I suggest a couple of topics that we could discuss, could you just update us on what's happening on the security front in the Omarchy project right now?

Yes. It's hugely important. It's actually been the number one priority since the launch of Omarchy Quattro. It's one of the reasons why we've not pushed out a lot of new features, because I wanted to make sure that the underpinnings we had were solid. And of course we needed help with that. I am not, on top of all the other things that I do, an expert security researcher. I rely on the fact that the Linux community, broadly speaking, has a ton of people who are very good at security, and that most of the components Omarchy is made up of are very secure, because they get lots of scrutiny and because Linux runs on everything. So we have a very secure foundation to build on top of. This is not Windows 95. I'm not building something on top of a very shaky foundation that is just full of holes. I'm building on top of damn Linux.

And this is also the curious objection I've heard from people who want to claim both that Omarchy is just a bunch of dotfiles, and also that somehow we have all these very unique security vulnerabilities. Well, isn't it one or the other? I understand that there can be security vulnerabilities in the dotfiles, in the configs. We had one example: I think if you had installed some other terminal (I forget which; Kitty), Kitty has an option where the default was turned on, and some people thought that was a bad option. Totally fair. We can discuss dotfiles and what holes you can open in those. But we're using the standard OpenSSL lib, the standard kernel, the standard this, that, and the other thing. So there's a little bit of cognitive dissonance there, and I don't always think the objections are in good faith. But I don't even care whether they're in good faith or not, because if they find something, we will fix it.

And now we have a team. The first team that got incorporated after the core team was actually the security team. We made sure that we have a bunch of seasoned professionals who work with this on a daily basis be part of a volunteer force to do it. Then the second thing was to start hiring people with deep expertise in this. Krzysztof, the Linux kernel maintainer, not just a developer but a maintainer, who is responsible for part of the PCI subsystem and other areas of the Linux kernel, was, how did it go, part of the first group of hires. A lot has been happening, and we keep doing the announcements, so which order did we do these in? But he came on in part not just because I wanted to have the fastest kernel with the best compatibility (that's a big motivation, of course) but also because he comes from a background of Linux kernel hardening, and we need professionals like that on board. The same thing when we hired outfoxxed from Quickshell. He has a lot of opinions about how to write secure systems.

Now, are we fully there with all the things? Of course we're not. We just got started. The whole Omarchy project is one year and change old. We've been blessed with attention from a lot of motivated security researchers online (I think that's the polite way of putting it) who are donating their time and energy to finding issues. Some of them are responsibly disclosing that to security@omarchy.org. I thank them; we thank them. We now have a special security credits page that calls them all out and gives them credit for that. And in fact the first person who did a responsible disclosure was an intern at Cloudflare, who shortly thereafter got a permanent job offer. So I think there are mechanisms here where Linux can be a vehicle for you to pad your CV in positive ways, by making contributions and being able to point to those contributions as recognized by other peers in the industry. So I think that's very good.

Then we also get the other kind: people on Twitter who find something and who have whatever bone to pick with Omarchy, so they just disclose it irresponsibly. And I thank them too. A lot of the ones that we found that way were upstream problems that were actually issues in all Linux distributions. So Omarchy is drawing in a new kind of energy and attention. Sometimes it's kind of hate-finding. Okay, fine, I'll take that too; I'm not picky. If you find a security issue in Omarchy, I hope you will report it responsibly, and if you don't, we will still fix it.

So we have all of that. We have all the investments, and we've got a bunch of programs in the pipeline too that we haven't been able to announce yet. Actually, just before hopping on this call, I was on a call with someone on the agentic side who can give us access to cybersecurity models that are trained for this kind of stuff. We will do all of it. And I am committed to Omarchy being as secure as a Linux distribution possibly can be, because it is the foundation of everything. If we keep going and keep getting adoption and this thing becomes Windows 95, it's all for nothing.

But it also requires diligence, it requires focus. It currently is the number one priority. But it's also something where we lean on the rest of the Linux community. We don't have to do this all by ourselves. We're not building everything from scratch, right? Most of these things are battle-tested. The Linux kernel has got 30 years on it and a lot of attention on it. So I think it's a serious issue and requires vigilance and attention and investment and hiring and focus. And I also think the specific focus that Omarchy is getting in that area is completely overblown.

But on the other hand, I think what you said is that, because of the popularity and the large exposure, and the negative attitude of a large part of the community, they then, out of spite, start looking for negative things, for security flaws. And in the end they end up contributing to Omarchy, and even to Linux. So in that sense it's only a very positive thing.

The AUR is not vetted

I also did a stream giving my takes on security in Omarchy, and one thing that I talked about was the Arch User Repository. So yes, to your point, you're saying the kernel is very secure in that sense. But there were several comments under that livestream, where I explained that the Arch User Repository is actually not vetted, it's not scanned. That's correct: anything could be in there. And there were people in the comments saying, "I've been a senior software engineer for years, I was running Omarchy, and I had no idea."

Well, it's important on that point: let me just correct the record there. Omarchy does not use the AUR for anything that's in the box. What the AUR is there for is as an optional install for people who want, of their own accord, to go find software there. Now, I think the AUR is full of great stuff, and what we generally point folks there for is things like Brave, or whatever other browsers, where the repos are controlled by the companies and so forth. But yes: I think it's 117,000 packages, many of them done by volunteers, and the AUR design is one from a naive era, kind of like the original email protocol design, where the people who did email didn't think that there'd be spammers in this world. So the protocol's not really made for that. The AUR is made on a very admirable idea, actually: hey, we're just a bunch of fun people wanting to package software up. They weren't thinking so much about the malicious aspects of it.

And they haven't hardened their distribution network either. They had a lot of outages when they got DDoSed, for a long time. This was actually what caused us to switch off the AUR. The original version of Omarchy did have AUR dependencies. That got fixed very quickly, partly or majorly because of the availability issues that they had, and that was why we stood up our own mirror, and then we stood up our own package repository, and then we simply went: you know what, we don't need the AUR for any of this stuff. We should be controlling our own packages and our own build pipeline.

But I also think that you can go too far, and I think that the AUR is an incredible resource, in the same way that npm is an incredible resource for JavaScript packages, that RubyGems is an incredible repository for Ruby software, and crates, and Go. There are actually a lot of software repositories where, if you run those things without reading every line of code, you are exposed to potential issues. npm has seen a lot of them, and npm still runs every fucking major website in the world, right? It's still part of those pipelines. So I think we need to take it seriously, and then also not throw the baby out with the bathwater. It's actually good that individuals can publish software on the internet. Now, there are issues with it. We should have controls around it.

Agents as defenders

And I think this age of agents actually makes that way easier than it would otherwise be. If there's one thing the agents are very good at, it is finding security issues. This is one of the reasons people are freaking out, because that can be used offensively. But the defensive capability of agents is incredibly good. Therefore we have a straight path to a modern version of the AUR that embraces agents, that perhaps has some slightly less naive policies around adopting packages and some of the other problems that they've had. But we should not be throwing out the baby with the bathwater. It should be possible for anyone to distribute software on the internet, and then we build in controls and checks and whatever to make that as secure as it can be, while accepting that if you run code written by someone other than you, there's a risk. And it should not scare us away from computers.

Yes, exactly. So just to get the current state of Omarchy: when someone installs Omarchy right now, the AUR is not enabled, and they have to take an action in order to enable it and start downloading from it?

When you install Omarchy right now, no AUR packages are used, no AUR packages are updated. There's nothing in the flow that relies on the AUR. What there is: if you go to install a package, you will not be shown the AUR package repository. The default is "Install packages," and there you will just be shown the Arch main repos plus the Omarchy repo. These are the packages we guarantee, right? Well, as much as you can guarantee. But Arch has a really good reputation for diligently doing package management. We've been building our own package repository. Some of it is based on original AUR packages that we then reviewed and put into the package repository once we were happy with them.

But then, if you can't find what you're looking for there, we give you the option of using the AUR, and if you do use the AUR you're presented with this interface where you can sort the 117,000 packages. And I think if you push Alt-D you get to see the package build, so you can review all this stuff as you see fit. And you can also decide that the AUR, as a concept, is just not something you want to use. You think there's too many risk factors to it? Great, don't use it. In fact, I don't use the AUR very much, because we've managed to package everything I need in the Omarchy package repository. In fact, that's how it goes: whenever there's a package I want, it goes into the Omarchy package repository, because then I'm not runtime-dependent on the AUR and so forth. So sometimes the AUR just serves as kind of an invitation for us to do our own package management. And in time there's not even going to be an AUR; we're working towards full sovereignty around packages, around hosting them and building them.

But as we get there, I also want to allow people to know: hey, here's the Wild West. You can go in there eyes wide open about what that means. Someone might shoot you in the saloon, but it's also a little exciting and there's cool horses. So you choose what you want to do. I don't want to create Apple. I don't want to recreate the locked-down computer. I want to have an open computer that is also safe, where the line between the Wild West and civilization is demarcated. I actually believe we have that right now. Now, you could maybe disagree and say, "Well, someone could stumble into the AUR, and they didn't know anything, and then they install something." Yep. That's the liability of free adults.

Attempts versus results

Well, just one idea here. Even if it's not enabled by default: if you're about to install something from the AUR for the first time, you could have something popping up that at least gives some information: okay, what you're about to install is something that has not been looked at. Just one time, just to educate people. Because if a senior software engineer doesn't know that this is actually the case with the AUR... Many people don't even read these things, but at least you've tried.

Yeah, but I don't actually care so much for attempts. I care for results. And what I think we've learned, if anything, is that these pop-ups don't have a lot of effect. It's been demonstrated, right? People have been trained that if they want to use their computer, they just click OK until the software appears. "Agree to terms." I don't actually care to build software on the basis of hopes and dreams in that regard. I care to build it on top of what works and what is effective. I don't think nag screens or warning screens are terribly effective, because I think we've trained users over the past several decades to ignore all of them.

And I think the Mac is actually particularly egregious about this now, because there's 17,000 pop-ups telling you this and the other thing. They literally used to have an ad for this, the "Mac vs. PC" ads back in the 2000s, making fun of Windows for having all these pop-ups, and now they've become that themselves. Again, there could be a line here; I'm not even totally against it. I think we just have to be realistic about what something like that does. If you put in a thing that says, "Hey, be careful, you're on your own accord now, you're in the Wild West," you could maybe say to yourself, "Well, at least I tried." But if it doesn't actually work, I'm skeptical that that's the path to go. I'd rather fix the problem at the root.

A foundational distro

And long term, the root of the problem is: we should be reviewing all the packages. We should have full sovereignty over the entire build pipeline. Omarchy is on track, and has stated its intent, to become a foundational distribution with its own package handling and whatever. We're bootstrapping on top of Arch right now. We use the Arch repos, we allow people to use the AUR. But the final state is that we're a foundational distribution that depends on no one for nothing. We will upstream, we will collaborate, we will do all the things, but we will take full responsibility for the entire package. Now, I don't know how long it took Debian or Arch or whatever to get there. A while. We're not getting there tomorrow, but we're going to get there quickly, because we have the resources, we have the momentum, we have the agents, we have the community. We can totally do this. I don't know when the last time was that a true foundational distro appeared, but we're making one. So that's the designated path here.

Omarchy for kids

Yeah. I'm not going to give in just like that on the security thing, though. I would really, at least if you have a kids version, if that is even going to be a thing...

There will be a kids version.

...and they won't be able to install anything?

We're making it so right now: they will not have sudo, which means they can actually not install packages of their own accord.

Well, okay, that's great. Because when I hear "we're going to make something for kids," and then they can just search the AUR and install something, I would think that is a recipe for disaster. And the reason I say that is not because I want to hate on Omarchy. I would hate for something with Omarchy, with this publicity, to get a huge hack and get huge negative publicity, right? That's the thing that I would hate to see right now, with this momentum.

Same here. Same here. That's why literally in the groups I've stated that the number one priority is security, that we need to make sure of this. Now, that doesn't mean we're going to catch everything. The number of CVEs that are published every day for the Linux kernel is long. Most of them are not zero-day remote execution issues; some of them are rather obscure issues, right? We need to fix all that, whether it comes from upstream or whether it's an AUR package. And a hundred percent agree.

When I talk about the Wild West, I'm talking about adults, informed or less informed. Once you're an adult and you own your own computer, it is my belief you should be able to do whatever the hell you want with it. Well, within the laws, right? Don't use your computer for whatever. But even if you do, that's actually not within my remit. I'm selling you a fork and a spoon. If you use that fork and spoon to, I don't know, dig out of a prison or shank someone, I can't be taking responsibility for that, and I don't think that's a good path to go.

For kids, very different. With kids, you need to enable the adults in their life, their parents first and foremost, to protect them reasonably. And that's what we're going to do with Omarchy Kids: here's a version of Omarchy, or a path of Omarchy, that you set up for a child, where they do not have access to sudo, they have access to a limited whitelisted DNS for using the web, if that's what you want them to use, and they have either time limits or time nudges. That's what all that work needs to be. I would not today think, "Oh, I should just give an Omarchy machine to a child," with no instruction about anything, no help.

Now, I'm saying that, and then I also remember my own childhood, running an elite BBS system out of my bedroom at 14, trading pirated software. So I don't know; sometimes I have it a little bit both ways. But either way, I want Omarchy to be a place that parents can trust, and we have work to do there. It's not there yet, but we're going to get started very soon. There's a whole team. There's a Discord channel with literally hundreds of people working on this, not just from the technical perspective but also from the art. We're doing very specific kids themes that are just wholesome and nice, and all the other stuff.

Adults get the Wild West

And then for adults I'm saying right now: hey, you know what, Omarchy is not Ubuntu. It's not trying to be. At this point it is still in the pioneer, early-adopter stage, and those people should have access to the frontier. They should have access to the Wild West. It should be separate. There should be a part of it that's like: here's town, there's a sheriff, and if someone comes in there he's going to take care of it. That's our own packages, and that's our Arch bootstrapping. And then, right over there: Wild West.

Yeah. Well, I'm really glad we could clarify this, because when you just look at your X timeline, when you say "Omarchy for kids," things can get taken out of context. Myself included: I also came to conclusions based on just that information. So I'm really glad that you say that security is the first priority, and that you're working on Omarchy for kids. There's nothing more I could wish for, so that's really awesome to hear.

Why podcasts beat X

This is why podcasts are so much better for reasoned exchange. On X we have 240 characters, and even if I tweet about security, most of the time, if I'm just tweeting positive things about security, it doesn't travel, because what people want to see is a dead body in the streets, right? So it'll travel if there's someone who finds something, but it won't travel if I just go, "Hey, here's all these positive things we're doing about security." So podcasting is a much better format for that. And we also have a fairly in-depth security page at omarchy.org/security that details the responsible disclosure protocol, the team, what we're doing, and all this other stuff. So we take it incredibly seriously. It is foundational. If you are not focused on that in today's world, you're not going to make it.

Exactly. Well, I think that's a really good statement to end the podcast with. David, thank you so much for what you are doing for Linux, for desktop Linux. I can only imagine what positive effect it's going to have for kids to grow up with a tiling window manager instead of an iPad. If I had a kid, that would be the first thing I would do: try to find some Linux distribution that they could start using from day one. So thank you for all of the effort, time, and money that you're putting into this, and keep up the good work, man.

Thank you so much. Thank you for having me on.

David, where can people find you?

They can find the Omarchy project at omarchy.org. That's O-M-A-R-C-H-Y; I'm spelling it out because it isn't always pronounced like it's spelled. Omarchy.org, that's the portal to everything. And then my personal website is just dhh.dk. That's where you can get everything I'm involved with, both on the commercial side, with Basecamp, Ruby on Rails, books I've written, whatever else, and then omarchy.org is everything for the project.

And we'll be putting all those links down in the description below.