David Heinemeier HanssonModify

It's pretty incredible that we're able to run all these awesome AI models on our own hardware now. From downscaled versions of DeepSeek to gpt-oss-20b, there are many options for many types of computers. But let's get real here: they're all vastly behind the frontier models available for rent, and thus for most developers a curiosity at best.

This doesn't take anything away from the technical accomplishment. It doesn't take anything away from the fact that small models are improving, and that maybe one day they'll indeed be good enough for developers to rely on them in their daily work.

But that day is not today.

Thus, I find it spurious to hear developers evaluate their next computer on the prospect of how well it's capable of running local models. Because they all suck! Whether one sucks a little less than the other doesn't really matter. And as soon as you discover this, you'll be back to using the rented models for the vast majority of the work you're doing.

This is actually great news! It means you really don't need a 128GB VRAM computer on your desk. Which should come as a relief now that RAM prices are skyrocketing, exactly because of AI's insatiable demand for more resources. Most developers these days can get by with very little, especially if they're running Linux.

So as an experiment, I've parked my lovely $2,000 Framework Desktop for a while. It's an incredible machine, but in the day-to-day, I've actually found I barely notice the difference compared to a $500 mini PC from Beelink (or Minisforum).

I bet you likely need way less than you think too.

I haven't done a full-system backup since back in the olden days before Dropbox and Git. Every machine I now own is treated as a stateless, disposable unit that can be stolen, lost, or corrupted without consequences. The combination of full-disk encryption and distributed copies of all important data means there's just no stress if anything bad happens to the computer.

But don't mistake this for just a "everything is in the cloud" argument. Yes, I use Dropbox and GitHub to hold all the data that I care about, but the beauty of these systems is that they work with local copies of that data, so with a couple of computers here and there, I always have a recent version of everything, in case either syncing service should go offline (or away!).

The trick to making this regime work is to stick with it. This is especially true for Dropbox. It's where everything of importance needs to go: documents, images, whatever. And it's instantly distributed on all the machines I run. Everything outside of Dropbox is essentially treated as a temporary directory that's fully disposable.

It's from this principle that I built Omarchy too. Given that I already had a way to restore all data and code onto a new machine in no time at all, it seemed so unreasonable that the configuration needed for a fully functional system still took hours on end. Now it's all encoded in an ISO setup that installs in two minutes on a fast computer.

Now it's true that this method relies on both multiple computers and a fast internet connection. If you're stuck on a rock in the middle of nowhere, and you somehow haven't discovered the glory of Starlink, maybe just stick to your old full-disk backup ways. But if you live in the modern world, there ought to be no reason why a busted computer is a calamity of data loss or a long restore process.

The only way many tech workers in the US can get a long break is by quitting their job. So lots of them do that every few years, which is partly why the average tenure in our industry is at an atrocious 18 months. But this terrible rate of churn is often avoidable by one simple benefit trick: Sabbaticals.

We've been giving everyone at 37signals a six-week sabbatical every three years for the last fifteen years or so. It's been magical for retention because a break like that allows the mind to reset in a way a two-week vacation never could. And when employees yearn for such a reset, the typical option is usually just to quit.

I know the idea of a six-week sabbatical might sound strange to many Europeans who'd be forgiven for thinking "isn't that just August"? And they're not exactly wrong. Europeans usually do enjoy more vacation time, but in the tech industry, that also comes with much lower pay. Easily half to two-thirds less.

I think it's entirely possible to have it both ways: Work for an American tech company with American pay levels, but also enjoy a regular full reset, without having to quit to get it. 

And the argument for the boss doesn't even have to be some humanistic plea about long-term happiness. It can simply be about retention: it's very expensive to see smart, trained people walk out the door.

I'd even argue that bosses — be they founders or professional executives — benefit just as much from a regular sabbatical like anyone else. Whenever Jason or I have taken one, we've always come back with fresh ideas and perspectives that invariably lead to positive changes or new ambitions that wouldn't have come otherwise.

Six weeks is also just long enough to remind tired founders that selling their company isn't likely to be the bliss they imagine. That mojito island gets boring quickly. That by week five, they're probably already antsy to get back to the action. There are endless stories of founders who regret selling their business when all they needed was a six-week break from the startup sprint.

Bottom line is that we all need a long break every now and then. Not just two weeks on Mallorca, but time enough to get bored. To get hungry for the intellectual stimulation of work and the social connection of colleagues. The sabbatical is a great way to deliver that and keep founders from wanting to sell and employees from wanting to quit.

As Omarchy was taking off this summer, and thousands of happy users started expressing their delight with the system, I kept waiting for the universe to balance the scales of passion. Nothing of note in this world is allowed to succeed without spawning a counteracting force of haters. And now they're finally here.

The same happened twenty years ago with Ruby on Rails, but back then I still thought you could argue your way to understanding. That if you just made a logical case to counter whatever objections were raised, you'd be able to persuade most haters to change their perspective. How naive.

It was Kathy Sierra who changed my perspective on this. From being annoyed by straw men and non sequiturs to accepting them and the haters as a natural consequence of success. That if you want people to love your creation, you have to accept the opposing force. Yin and yang.

Here's how Kathy presented the choice:

It's safe there in the gray middle. Nobody is mad at you, nobody is making any bad-faith arguments, but also, nobody cares. Lots of work exists in this zone. And that's fine. We don't need every project to reach the moon! But when escape velocity is achieved, you can't avoid drawing energy from both sides.

All this isn't to say that all objections, skepticism, or criticisms come from haters. Far from it. But once sufficient success is secured, a large portion will. It's just that kind of planet, as Jim Rohn would say.

The trick is to see this in aggregate as a necessary milestone. One that's even worth celebrating! Have you even made something worth cheering for, if there isn't a contingent there to boo at it too? Probably not.

So embrace the boos as you embrace the cheers. They come as a pair.

Omarchy didn't even exist before this summer. I did much of the pre-release work during the downtime between sessions at the 24 Hours of Le Mans in June. And now, just a few months later, we've delivered a petabyte of ISOs in the past thirty days alone. That's about 150,000 installs of the Omarchy Linux distribution!

I've been involved with a lot of successful open-source projects in the past quarter of a century or so. Ruby on Rails, first and foremost. But nothing, not even Rails, grew as quickly as Omarchy has been growing in the first few months of its life. It's rather remarkable.

This is what product-market fit looks like. Doesn't matter if the product is free or not. The fit is obvious. The stream of people who don't just enjoy Omarchy but love it is seemingly endless. The passion is palpable.

But why? And why now?

As per usual, there are a lot of contributing factors, but key is how Apple and Microsoft have been fumbling their relationship with people who love computers in general and developers in particular.

Microsoft is killing off Windows 10, which in turn cuts off a whole slew of perfectly fine computers made prior to around 2017–2018. They also seem intent on shoving AI into everything, and wavering on whether that might be optional or not. Oh, and Windows is still Windows: decades of patching cracks in a foundation that just never was all that solid to begin with.

Apple too has turned a ton of people off with macOS 26 Tahoe, liquid glass, and faltering software quality. They're also cutting off all Intel-based Macs from future updates. A Mac Mini sold as recently as 2023 is now end-of-life! This is before we even talk about how poorly the company has been treating developers depending on the App Store bureaucracy.

Meanwhile, Linux has never looked better. Hyprland, the tiling window manager at the heart of Omarchy, is a sensation. It's brought an incredible level of finesse, detail, and style to the tiling window management space: superb animations, lightning-fast execution, and super-light resource consumption.

The historic gap in native GUI apps has never mattered less either. The web has conquered all as the dominant computing platform. In the past, missing, say, Photoshop was a big deal. Now it's Figma — a web app! — that's driving designers. Same too with tools like Microsoft Office or Outlook, which are all available on the web.

I'm not saying there aren't specialized apps that some people simply can't do without, that keep them trapped on Windows or Mac. But I am saying that they've never been fewer. Almost everything has a great web alternative.

And for developers, the fact is that Linux was always a superior platform in terms of performance and tooling for most programming environments. With 95% of the web running on Linux servers, all optimization and tuning needed to get the most out of the hardware was done with Linux in mind.

This is why even a $500 Beelink Mini PC is competitive with an M4 Max machine costing thousands of dollars for things like our HEY test suite, which runs on Ruby and MySQL. Linux is just really efficient and really fast.

Finally, I think the argument that owning your computer, fully and deeply, is starting to resonate. The Free Software crowd has been making the argument since the 90s, if not before, but it's taken Apple's and Microsoft's recent tightening of the reins on our everyday operating systems to make it relevant for most.

Omarchy is a beautiful, modern, and opinionated Linux distribution, but it's also yours. Everything is preconfigured, sure, but every configuration is also changeable. Don't like how something works? Change it. Don't like the apps I use? Change them. Don't like how something looks? Redesign it. The level of agency is off the charts.

Turns out that plenty of people were starved for just this. All it took was someone to actually put all the pieces together, ignore the Linux neckbeards who insist you aren't worthy to run Arch or Hyprland without spending a hundred hours setting it up from scratch, and invite everyone to the party!

We're fed an endless stream of consternation over AI slop these days. The content apocalypse is nigh! It'll rot your brain! Okay, sure, maybe, but have you seen the kind of content sludge that perfectly ordinary humans are capable of producing? It's thrice as tragic.

The web is full of it. Garbage writing and brain-dead shorts. Content mills pumping out nonsense pages and gagging videos to appease whatever the high priests of SEO now think they've divined will please Lord Google or Master TikTok. 

It's been infecting websites everywhere with "calls to action", "white paper available upon sign up", and "10 ways to supercharge your productivity". Links stuffed into every crevice to juice rankings, capture "most searched for" keywords, and convert, convert, convert.

It's an affront to humanity to make sentient beings do this work. Turning human potential, creativity, and ingenuity into content sludge is a process no more dignified than turning pink slime into chicken nuggets.

I'll take AI slop over human sludge any day. Let the little robots barf up tokens to unlock the next basis point of incremental conversion. Better them than us, I say. This is exactly the soul-crushing, creative drudgery that machines were made to munch through without complaint.

But couldn't we do without sludge or slop, you say? Sure, right after we reach a shared state of nirvana. As soon as the average 4.5 hours of screen-on time is turned into real reading, real making, real pursuits. So that'll happen exactly never.

Case in point: the most important attribute of a phone for most people is still the battery life. These little content slop and sludge faucets can already spew out nearly an entire day's worth of nonstop eyeball junk, and yet you crave more. More! MORE!

So stop whining about the AI slop. You're already steeped in human sludge. And the door to exit both was always there. But you're not going to open it, are you?