Table of Contents
I like to focus on things that last. Most things are ephemeral, and the people who make them seem to be perfectly fine with that.
Disappearing Software
When I worked in fiat mines, most of my code lived for weeks or months, especially in startups. I often said we weren’t building software but interactive presentations, which most developers agreed on but didn’t mind as long as money flowed.
Most professional software engineering is building things explicitly designed to be thrown away in the near future. That’s often a solid way to make a profit, but a weird way to spend a life, when you stop and think about it.
Buildings are Durable
Most places on OpenStreetMap stay unchanged for decades. Say a guy named Satoshi opens a cafe in 2009. The place can easily stay unmapped until 2012, no hurry. Then someone finally maps it, and someone else updates its opening hours later in 2019. Now it’s 2026, the place is still here, and the information is as relevant as ever. All the previous efforts were not in vain.
Most roads are capital-intensive and carefully planned, a key difference between real engineering and software engineering. Mapping roads is a great social service. They are quietly used by navigation apps, delivery services, and basically anything that needs to be spatially aware.
OSM data is openly licensed and stored with great redundancy. Some areas may accumulate rot, but it goes away as soon as there is a single local maintainer equipped with some local knowledge. The whole thing is also free, so you don’t need to be rich to benefit from public infrastructure.
That’s what a healthy project looks like. Well-thought-out contributions from diverse sources accumulate and benefit everyone. The protocol may seem outdated and boring, but boring is what makes it exceptionally stable.
Apps and Protocols
RSS and Atom are decades-old protocols that still work fine with no changes. As a publisher, all you need to do is choose software that supports webfeeds, and most publishing software does. The reader apps people wrote decades ago can still subscribe you to your content. Nothing about webfeeds needs to be “reimagined” in 2026.
Compare that to news and social media apps, which are replacing each other or fully rearranging themselves every couple of years, hoping to align themselves with the new trends. The top apps are in a messy permanent state of reinvention, and most of them don’t even have the decency to export your data properly when they shut down.
I use my own feed reader for a reason. The main layout hasn’t really changed since 2022. I know, I’m missing out on a stories panel, infinite feed and 300 MB of dependency bloat. I just wanted beautiful images, titles, summaries, and dates, all wrapped in a standard Material interface.
Some mistakes were made under the hood, such as unstable ORM dependencies which forced a massive rewrite, but lessons were learned and ORM is fully eradicated as a class.
Open Source is Anti-Short-Termist
Open source software is fundamentally anti-short-termist, which is one of the reasons I like working on it. Nobody is going to rewrite a well-reviewed library every six months just because the new retard in charge wants a “modern stack”. If the code is useful, it stays useful, and that’s the end of the discussion.
There is no point in building skyscrapers if your foundation is garbage, and the same is true in software. A focus on fundamentals is what scares the hell out of hipsters, which is an underrated second-order effect of removing the noise.
The people who stick around in FOSS tend to be the ones who care about whether the thing actually makes sense and will work in five years. The ones planning their next pivot into an AI-powered blockchain orchestrator are naturally filtered out.
You also end up in the company of actually pleasant, serious people instead of lazy navel-gazers. Open source communities are full of adults who care deeply about quality and stability. I’ve met many amazing people there who I now call my friends.
Anti-Ephemeral is Antifragile
Anti-ephemeral things are also antifragile, and BTC Map is a good example. Since buildings are durable, we build upon that foundation and bundle all the merchants right into the app.
The upstream data lives in OpenStreetMap, but it gets included in our APK at build time and is rendered locally on the device. If our or OSM servers go offline tonight, or even forever, most locations will remain useful for years, because we shipped them to every user the moment they installed the app.
This works because real-world businesses are not short-termist. The same place that accepted bitcoins in 2022 is most likely still there in 2026. You can’t remove a merchant from a binary that’s already on thousands of devices. We will also cache communities at some point, enabling people to locate other bitcoiners fully offline if/when shit hits the fan locally or globally.
Files Beat Algorithms
The same logic applies to writing. Why complain about being censored or suppressed by some algorithm if static websites are easy to run, with no DDoS concerns and no censorship concerns, especially behind something like Tor or any of the cheap VPS providers that don’t care about your content?
The site you’re reading right now is a handful of Markdown files (mostly plain text with extra tweaks) compiled by Hugo into a folder full of HTML pages and rsynced to a $5 throwaway server. Files are easy to grasp, serve, and back up. There is no need to make it more complicated than that, and my notes will last at least as long as my backup with no subscription cost or maintenance.
Conclusion
I prefer to focus on fundamental things that last, and I like to build on boring protocols. I wanted to cover software distribution, but it didn’t fit the format, so probably next time. In short, distribution matters too, and apps should be simple, self-contained binaries, like bitcoind or .exe files from the Windows days.
Postgres? Too complicated unless you operate at Google scale. An SQLite file is the best database for most use cases. Overcomplication and navel-gazing create maintenance burden, add fragility, and make projects more likely to be abandoned, which is fundamentally lazy and short-termist.