BONUS EPISODE: 37signals is moving to Omarchy

0:00Welcome to Rework, a podcast by 37signals about the better way to work and run your business. I'm Kimberly Rhodes from the 37signals team. Today, joined by David Heinemeier Hansson, 37signals co-founder and CTO. If you've been following David on any of his social media channels, you've seen over the last year or so that he has switched from Apple products to a framework laptop. He's now on Linux and has been working on a new project called Omarchi to bring all of these things together. We're here to talk about it today. David, I'm going to open it up to you. Why don't you give us kind of a high-level overview about what all these things are that we've been doing?

0:37And then I want to dive into an announcement you made internally at 37Signals about how we're going to be moving forward. Certainly. So I switched to Linux about 18 months ago. And the first port of call when I switched to Linux was Ubuntu. That is the most popular Linux distribution out there. It's been maintained for many years by a company called Canonical. And it's really nice. I found that that Ubuntu default wasn't really set up for developers. It didn't have all the tooling. It didn't have all the keyboard affordances that I wanted. So I quite quickly built something called Omacube, which was kind of a remix of Ubuntu.

1:17And I ran that for about a year. And then just a few months ago, I decided to try another adventure into Linux land and experiment with this combination of Arch Linux, which is the distribution kind of like Ubuntu. And then this window manager, it's called Hyperland that sits on top of Arch and drives all the windows and how everything works together. The big difference is that Arch is incredibly bare bones. When you get it out of the box, it has basically nothing. You have to run terminal commands to even set up your Wi-Fi. You have to install everything from scratch to select each individual package.

1:55And I initially thought, you know what, that's not what I want. I'm trying to get the opposite. I'm trying to get my ideal environment up and running in 10 minutes or less. But then I started scratching a little bit at the surface and I started realizing, you know what, there's an advantage here to the fact that Arch gives you nothing in that I can do whatever I want. I am no longer bound by the choices canonical made for Ubuntu. I can put everything together just perfectly right. And that is the combination of factors that then led to Omachi. And Omachi is Arch Linux plus Hyperland plus my Omikaze opinions, choices, configurations, all put into one single installed script and very soon a full installation device, ISO, that you can just run and get up and going.

2:46And I've been running this for a couple of months and I've basically been obsessed, completely obsessed with designing the perfect development and really computing environment. For me, first and foremost, I want my computer to be perfect, not just like sort of kind of okay, but perfect. And then I realized, just like I did with Rails, that by the time I'm done turning something like Arch and Hyperland into the ideal setup for me, there's a bunch of other people who are just like me, who would very much love to be up and running with a super productive, beautiful system in 10 minutes or less. So why don't we do, just like I did with Rails, just like I did with Omicube, put all this stuff into a box, get more people excited about it, get more people to help out and build something great together.

3:35And that's what Umachi has been doing over the past couple of months. It's really sparked an enormous amount of interest from a bunch of people who didn't necessarily think that they were going to run Linux, which is exactly where I was 18 months ago, not thinking Linux was a thing that I was going to do. That was something for a different kind of nerd. And I realized, you know what? No, Linux is basically for all nerds. They just don't necessarily know it yet. They just need a compelling package that lowers the barrier of entry while it gives them something that is uniquely different. And what I've come to realize is that the work I put into Omacube, making Ubuntu easier to use for developers at the first time, it's good.

4:17I liked it. I ran it for over a year. I still think it's a wonderful package. But it was quite similar in a lot of ways to what you would be able to get out of a Windows box or a Mac box. Omachi is very different. And the key reason it's very different is that it runs Hyperland, which is this thing called the Tiling Window Manager. Most desktop operating systems, you drag your little windows around with a mouse. And if you need to see something underneath another window, you drag it over here and then you... It's very mouse-centric. With Omacube, I used a bunch of tricks to kind of get around that.

4:50But there's nothing like a Tiling Window Manager where all of that is taken care of for you automatically. You open a new window in Omachi, boom, the existing window you have slides over to the left and pop. There's the new window or application on the right. You do that a few more times and now you have a perfect grid. There's just so much less organization in it. And it's funny because I remember seeing people run these systems. Tiling window managers have been around for a very long time, even though Hyperland in particular is quite new. There's been versions of this for a long time. and I thought, you know what, this seems, I don't know, it just doesn't seem for me.

5:27And as I'd found with so many things, when I think something is not for me, some of the times it's just because I haven't tried the right thing. I mean, I have my mechanical keyboard here. I thought mechanical keyboards, again, what a nerdy thing. I'd heard from a bunch of people, oh yeah. You're like A and A for the keyboard. Ooh, ooh, ooh, the talk? What are these people talking about? I was just on a Mac, Apple, Magic keyboard for the longest time, not realizing that this could be something because I tried one mechanical keyboard and I didn't like it. Very much the same thing with Tiling Window Manager.

6:01I'd seen one Tiling Manager and I didn't like it and I didn't give it a second chance. And then I found Hyperland, which is this amazing Tiling Window Manager that's built by something very rare in the nerd Linux world. Someone with a unique sense of taste and aesthetics. There's a lot of window managers, tiling window managers even, that are very functional, very fast, but they don't necessarily look great. Hyperland out of the box looks fantastic. It even has animations, which I usually hate. I usually don't like animation. I've pulled out all the animations from Omacube to make it as fast as possible.

6:39But here's a set of animations for just the right transitions that don't feel like a chore, but actually feel like they add to the experience. And then there are other things like switching between workspaces that don't have animations because you want that to be as fast as possible. But the overall sensation of using this system is just on a completely different level. And the funny thing with the Hyperland is it has the same philosophy as Arch. These are just building blocks. They're not giving you a whole system out of the box. In fact, when you install Arch and Hyperland, like there's no login screen, there's no notifications, there's no application launcher, there's nothing.

7:16There's a system that knows how to swap windows around and bind key bindings and so on, but you have to do all the other work yourself. And again, that's actually not what I want in the end, right? Like I want a system where everything's just set up. I don't have to think about it all the time, but realizing that operating at this low level, I can put the ideal system together. Just like when I showed up and I found Ruby and there was no web application frameworks for Ruby at the time. I mean, there were some, but there was nothing like Rails. There was nothing that was just fully integrated, full stack, full everything.

7:49I had to build that bit by bit. And by the time I'd done that, we ended up with a highly opinionated Omakasa system where both others adopting it and myself using it didn't have to think about these decisions again. So it's just been an absolute joy and revelation to discover that that sensation, that like peak experience for me 20 plus years ago with Ruby can now have like a sequel in Linux. And it's as exciting in some ways, as exhilarating to discover that a bunch of insanely good technology was just out there. And I kind of just had it in periphery. I didn't really investigate further. And now that I have investigated, I found something that just really fits my brain like a glove.

8:38and just as satisfying is realizing that there are already thousands of other people, primary programmers, but not all, who've now had the same experience because they've been able to install Amachi and realize that they too, like tiling window managers, if everything is just set up and there's a base for them to iterate on and create their ideal environment, some people just leave the Amachi defaults exactly as they are and other people will take that as a starting point and really tailor their environment to it. One funny thing about that was just two days ago, I think, I got this pull request, a piece of code submitted to the Umachi project from a guy who'd never made a pull request before.

9:18Like ever in life? This was literally the first time he was contributing to open source in some way and he was changing something about how Umachi worked. And I thought like, this is really interesting because first of all, something like Umachi is pulling someone into open source, starting their journey with open source. And then also, could you imagine submitting a pull request to turn down Liquid Glass in macOS or pull out some of the junk in the Windows Start menu? Completely unfathomable. These are not affordances we have in the commercial operating systems. Linux really is unique in this way that you can tailor just to you, that it doesn't have any of the gunk or the junk that you'll find in the commercial offerings.

10:03And then you can collaborate with other people who share your taste. This is one of the criticisms that's often leveled against Linux. It's, oh, there's all these choices. There's all these distributions. Yeah, because we don't all want to run the same operating system. We at least don't want to run the same flavor of it. I mean, underneath it all is all Linux. I mean, Omocubus Linux, Ubuntu's Linux, Arches Linux, Omarchi's Linux. It's all just Linux, which is to say it's the Linux kernel, the magic that makes all this stuff work. But then on top of that, we can really express ourselves in highly personalized ways, just like we do with programming languages.

10:40Ruby is a very unique programming language. It does not appeal to everyone. I've come to realize that after some time, but not everyone loves Ruby as much as I love Ruby. But they might love Python as much as I love Ruby. They might love Go or they might love another language. And again, that's what Linux gives us. There's no way that everyone in the world just like either loves Windows or love macOS. No, that's not what it is, right? So it's just been absolute, complete joy to discover this, to build Amachi up, to now I already have 20 releases onto it, thousands of developers into it, manufacturers of Linux friendly machines interested in engaging with this, whether Framework or B-Link or some of the other makers that I've been talking to, there's just a new excitement in Linux land that just kind of just ricochets, right?

11:35Like I get excited about something in Linux, then other people get excited, then I get more excited. And then we all just get super excited about a new piece of technology that's actually ancient, right? Like Linux is 30 plus years old. Hyperland is very new. Some of the other tools are quite new. And Omachi is obviously literally only two months old. but all of these things coming together just at the right time to give us a true alternative to something like the Mac that I'd been using 20 years prior to switching to Linux and all the other stuff. Other people on Windows feeling the same way.

12:05Now, finally, there is something that speaks to me, something that's different enough, unique enough, has a position and opinions that kind of align with my own. Now is the time to make a change and try something new. Okay, I have to say for all of your company check-ins where you share with the company what you've been working on, it is clear that you are so excited about it. Like just the energy behind it, even though it's just written words, you can definitely tell. If someone were to ask you five years ago, like imagine you're going to be on Linux, you probably would never have believed it.

12:35Two years ago. Look at you now. Not even five years ago. Two years ago, I would have gone like, there's no way. I'm not doing Linux. Linux is a thing I run on servers and that's for a different kind of nerd. I'm not interested, right? That's also why I'm so excited. I just, I simply love, first of all, learning. And I also love learning that I'm wrong. Like I literally had the wrong mental image of what Linux on the desktop was. I thought it was something else than it turned out to be. That's the best kind of wrong. The best kind of wrong. Where you're like, oh, I've been walking around with this misconception of Linux.

13:10And now I corrected that misconception and now I get to enjoy it. Okay, so let's talk a little bit about the announcement that you made recently to the company. So 37signals used to be an Apple shop. We all had Mac computers. More recently, it's been like, yeah, if you want to use Windows, let's diversify a bit. We should be diversified. Now we're moving more towards the Linux and Omachi angle. Talk us through a little bit about that and what you're thinking for the new policies of the company. Yeah, so about a year ago, just a few months after I had adopted Linux in general, I thought, you know what?

13:44This is good. Linux is good. I've been working with the Zama Cube, the first iteration of this idea of making our own distribution. And this is good enough that going forward as a default, as someone new coming into the company, this is what we should be using. So we made the policy, I made the policy about a year ago saying anyone new joining the technical teams that we have, if they don't have really strong preferences to the country, they should be using Linux. They should start on Linux. and we actually just hired four new programmers not too long ago and all of them have started on Linux.

14:18We have a couple of folks on the existing teams that switched to Linux and we have some people on the operations teams that've been running Linux and we even have a couple of designers that've been checking out Linux. So that kind of stuff was going well but we're in a different situation now. Omachi is a far more comprehensive take on what would our own distribution look like. what would it look like if we had full control over everything to set up a development environment just like we wanted it for many years we've had a project internally called shipyard and shipyard would set up a new machine such that it was usable for the technical staff so that they're ready to make commits and they're ready to deploy things and it was a bunch of stuff that was setting things up and i started realizing why are we doing this just for us like 90 of what's in these setup scripts are just like good practices that we should all be using.

15:14I mean, broadly here as developers, most people should be running something like this. What if we just took all these things and we put it into one package and we wrapped it around with all this other stuff? And that's basically what Apache is. It's taking all these developments we've had internally for what a great development environment would look like and then putting it in a box. And now I've come to the realization, hey, if we're doing that already, we're taking all the stuff that we needed to set things up. Now we put it into a package we're sharing with everyone. We should be running that package.

15:42In exactly the same way that now that we've spent years and years building Rails, and that has all the things that we want and we need to build new applications, we should obviously be using Rails. It would be very weird if we were not using Rails. If we said like, oh, you just used whatever you want. No. Omachi is now a 37signals Linux distribution. in the same way that Rails is a 37signals Ruby web framework. It's also for other people. Omachi already has, I think, about 50 contributors outside of the walls of 37signals. And Rails obviously has many thousands of contributors. So it's not just a 37signals project, but it's very much also a 37signals project.

16:26And once it gets to that stage, we should go all in. We should go all in and say, this is what we're going to put our efforts behind. When we do our setups and so forth, it should go towards one solution that we all improve together such that it is the best that it could be. Rails got so much better because it wasn't just me working on it, because there were a bunch of programmers outside the company working on it, and because all the programmers inside of 37Signals were working on it. Now that we have this reality around our computing, our core computing environment, it just makes total sense to me that that should be our default.

17:04So we've now announced that over the next three years, as hardware goes through its cycles, we're all on the technical teams, on the operations teams, and on the Ruby programming teams, going to move to Macchi, going to move to Linux, going to move to hardware from framework or primarily framework. They make these amazing laptops I'm so fond of. I have my beloved framework 13 right here. I actually have quite, I have two of these. And did I show you this? I haven't seen that. He's got this super neat purple translucent thing going on it now. Anyway, these are just great computers. And prior to that, it's funny, I actually have one here still that we use as a media machine in the bedroom now.

17:54Like this was standard issue equipment, right? Everyone would just get a Mac. Right. And that was what we were going to do. And now we're going to go like, boop, a WayMac in framework. But we're not going to throw away perfectly good Macs for people who are fine with that for now. So we're going to take it at a reasonable pace. Like three years is about the hardware cycle we've always had. Someone would update their laptop or desktop every three years. And that's what we're going to do. And that also gives everyone a time to sort of get used to it. That's not like one day to the next in the middle of a project.

18:26You got to just change everything. Sure. And then even with all of that, I'm even allowing for sort of a loophole in the commitment here that if for whatever reason, like your brain is just thoroughly incompatible with Linux, like, fine, run your Mac, right? We went from being essentially 100 % Mac shop to over the years, then we had a little bit of Linux and we have a tiny bit of Windows. We should actually have more Windows because half the customers on Basecamp run Windows and it's actually not perfect that we don't have more people on it. But we're going to be running all of it as we should.

19:00We should be running both Windows, Linux, and the Mac inside the company on a regular basis so that we catch all the problems that otherwise would be caught by customers. But on the technical teams, we can put more effort around one solution, make that super duper great and push that forward and help show the rest of the industry, other programmers, that Linux isn't just sort of a option in theory. It's a fantastic option in practice. And here's a company that goes all in on it. That's willing to go like, all right, that's it. That's what we're doing. And I think this is really what Linux needs.

19:40It needs new energy. It needs new aesthetics. It needs new commitments. And it should combine all that with the fantastic, wonderful work that's already been done over literally 30 years by super dedicated, awesome individuals and companies. This is all adjective. It's not like, oh, we're going to barge in and rewrite Linux. No, no, no. I mean, Omachi is a celebration of Linux. It's a love letter to Linux. Arch Linux, the distribution this is based on, and it's literally 20 plus years old. There's just so much fundamental core progress in Linux that we just need to like show people, which again, I keep coming back to this example of Ruby on Rails because it feels so similar to me.

20:22Rails was a love letter to Ruby. And I was basically going, Ruby is this incredible programming language. I need to present it to more people so that they can have the joy of writing Ruby. But to get that joy, they're going to need a little help. They're going to need a little nudge. They're going to need some tools. They're going to need a lowered barrier of entry. Rails is that. Omachi is attempting to do the same thing for Linux for a new group of quote-unquote customers. People, mostly Mac users, but also some Windows users. But I think in our sort of corner of the industry, web development, SaaS, all that stuff, the Mac has, what, 97 % market share?

21:02And I don't have any delusions that we're going to turn that fully around. But could we just get Linux to 10%, maybe even 20 %? Who knows, one day 30 %? I think that's entirely possible if you look at the developer community and the web developer community even more specifically. So this move that the company's making, it's reminding me of, we always talk about dogfooding our own products, but also like we're using Basecamp every day. Like that's how we're able to know exactly what it needs because every single person in the company is in it every single day. As you explain it, it just makes sense.

21:37That's exactly what this is. That now that Omachi has kind of shown itself as being so well suited for the kind of work that we do, we should be using it. The technical team should be using it. We should be improving it. We should be helping pushing it forward. And we can just do that so much more effectively if the majority of the technical staff is running it. And it's not any different than what we used to do with the Mac. We were all in on Mac. Like every single person virtually I can remember. I think like Noah, our own data analyst, like ran Linux for a little bit. Maybe one person on Ops also did for a little bit of some sort of offshoot experiment.

22:15But otherwise, we were all in on the Mac, right? And it's funny because when I announced this externally after we had announced it internally, I got a lot of the pushback, right? How dare you mandate the tools that people are going to use? Like, first of all, what are you talking about? Like companies in general, as default, mandate what people use, right? Like, hey, this is a Windows shop. We use Windows Manager something, something. Here's your machine. It's already set up for you. In our corner of the industry, that's what the Mac is. Here's your Mac. It's set up with all this stuff. That's what you're going to use.

22:51You're going to get a standard issue company machine. So for us to say the same thing about Linux, I don't find that at all kind of out of character with what everyone else is doing. In fact, it's even less in character because we're still embracing a diversity of operating systems at 37 Sickles. We want to have some Windows. We want to continue to have some Mac, but we also want to have Linux, right? And then the other thing is, this is a case where I'm not just saying like, hey, Ubuntu, let's run that. No, no. I'm saying at 37 Sickles, we're building a Linux distribution. We should be running our own Linux distribution.

23:22Of course we should. And if we do, we're going to make it better and we're going to make it better faster. And also every single programmer now inside of the company has a direct line to changing how the entire computing setup that they use every day, how that works. That's a way to get like just a tailored environment that everyone needs to be more productive. Not just a tailored environment, not just about productivity, but it's also about aesthetics. Umachi in a large sense is about aesthetics. That your computing environment shouldn't just be functional. It shouldn't just be fast. It should be beautiful.

23:55Absolutely gorgeous in just the same way that I want beautiful code. I want gorgeous Ruby all over everything that we work for. Umachi is allowing me and us to deliver that to our base computing environment. Okay, David, I want to ask you about one section of your write-up to the company, because I think it's important for people who might be thinking about making the switch to Linux, not just our own employees. So it says, I ask you approach the transition with an open mind, give it five minutes, then another two weeks. And then you also say, work through the frustration for at least two weeks, except a temporary drop in productivity to get there.

24:36Kind of talk us through that as someone who's like, you know, I have a good system that works. Why would I go through this frustration to make the switch? Habits are really hard to break. And computing habits for most people are very deeply ingrained. when I first attempted to, or not just attempted, when I first switched to Linux, I literally had 20 years of muscle memory, of expectations, of aesthetic tuning, of developing an eye for like the Mac way of doing things. And I can now tell you from very personal experience, very recent experience, holy shit was that frustrating. I mean, I kept doing the wrong hotkeys.

25:17I kept breaking things the other way. I didn't like how something looked. I didn't like how something worked. I was used to a certain way of doing things, right? And it only happened 18 months ago. So it's really fresh in my memory just how annoyed I was for the first three days to the point where I was questioning whether this whole thing was going to happen at all. Like the original motivation to move to Linux was powered by a frustration with Apple. But you can't sustain frustration alone, right? Like if my adventure into Linux was just about not liking Apple, it wasn't going to last. Like I was not going to make the commitment to rewriting everything just on the basis of animosity.