Modern productivity software is slick. You sign up, and start organizing your life around the tool. Your data is safely abstracted away in the cloud, with marketing copy offering the comfort that you could download it – though just enough UX friction ensures you probably won’t.
That was the Silicon Valley way: raise gobs of cash, build for the cloud and get sticky.
With Obsidian, two engineers from the University of Waterloo have inverted that script. Obsidian is a note-taking app that stores everything in markdown on your drive. Markdown is a plaintext format, so users can read and edit the file without specialized software. Obsidian is austere - mundane even. And that's the point. If you want bells and whistles, you can build them yourself. Obsidian provides a glimpse into the future of software – an app for a world where software is so inexpensive to write, that you can build features to serve a market of one.
I decided to inhabit this world to see what it's like.
Market of One
While most journalism is now online, I'm not a fan of reading on the screen. The online reading experience is crap for plenty of reasons: it hurts the eyes, it's distracting, and the apps are addictive by design. I wanted to break free from these addiction machines, but the app to regain my attention would be complex to build, and would only really serve me.
Here's how it would need to work: When I come across an interesting article, I'd just add it to my reading list - perhaps adding a tag like jevons paradox
. At the end of the month, I'd be able to publish a curated anthology of articles and push that to my kindle
This may sound simple, but the devil is in the details. Here's a glimpse: First, you'd need to clip the article in a way that doesn't capture crap, like ads. Suppose you store this as a file. If you want to push the file to your kindle, you still need to fetch all the images from the web, convert them to the right format, and insert them at the right location. And then of course, you'll encounter these innocuous characters on the web that can break your code – and you'll need to "escape" those.
Ordinarily, in the pre-LLM era it would take a dev a couple of weeks to get something like this functional enough for daily use. The dev would need to be familiar with libraries for yaml, markdown, web access, file manipulation, image processing, and epub creation. Those skills don't come cheap.
Feeling brave one lazy afternoon in Karachi, I got to work. In 4 hours I had a kludgy prototype that worked enough for a dopamine hit. Feeling accomplished, I went downstairs for dinner and didn't return to the project – the code left to idle in the purgatory of aspirational hello-worlds. Then one frigid Saturday in Toronto, sitting at the library in North York, I whipped up Cursor on my laptop, and plugged in my iPad as a reluctant extension monitor. "The library is closing in 15 minutes" boomed the loudspeaker at 4:45. But that was okay, because the app was done.
Jevons Paradox
In 1865, English economist William Jevons, observed a strange phenomenon. After James Watt invented the steam engine, coal consumption should have gone down. After all, Watt's design was dramatically more efficient than the Newcomen engines in use at the time. In reality, as steam power became cheaper, people used a lot more coal. This phenomenon, where improvements in efficiency lead to even higher consumption came to be known as Jevons paradox.
The paradox appears across industries – and it's the environmentalist’s nightmare. When you build more roads, you get more traffic. When refrigeration becomes cheaper, you get more fridges – and more food spoilage. When you build water reservoirs, water consumption goes up. The list goes on.
When I interviewed economist Anton Korinek in 2023, he alluded to the Jevons paradox. He noted that cheaper software production may lead to more software demand. But then he cautioned: demand isn’t infinite. There may be a limit on how much software the world needs. In other words, it depends on whether the demand for software is elastic.
I think it’s fair to assume that software demand is elastic to a point. The software I co-wrote with Cursor was not economically feasible before. But now, I have a personalized anthology on my kindle for my TTC rides.
YouTube Moment
When I spoke with an engineering program manager at a big tech firm about GenAI assisted coding, he called it a "a YouTube moment." By reducing the cost of publishing video content, YouTube created an entirely new market for user generated video content. Similarly, by dramatically reducing the cost of software creation, we may see an efflorescence of user generated software.
We don't yet know what shape this new software will take, but unassuming though it may be, Obsidian may hold some clues. In a thread on X, OpenAI co-founder Andrej Karpathy wrote a "love letter" to Obsidian, calling its philosophy of software the state of the art. The philosophy is simple, your notes are all in plain text stored on your local drive, it's super easy to build extensions – and most importantly, there are no attempts to lock you in.
In a post titled File over app, Steph Ango, the CEO of Obsidian, writes that it's a "delusion to think it (Obsidian) will last forever. The app will eventually become obsolete."
That's not the type of thing you'd expect a CEO to say, but in a world where all software is ephemeral, perhaps software CEOs need a new playbook.
Maybe the next revolution in software isn’t scale—it’s humility.
NOTES:
If you’re feeling adventurous, then here’s the repo for the Obsidian-to-Epub code: https://github.com/rizvee1982/obsidian2epub
It’s not pretty, but it works. Let me know if you want to use it – happy to vibecode some features for you :)
Thanks for posting the repo, now you’re building for a market of two!
Obsidian is currently my knowledge/notes warehouse, but I haven’t found a frictionless way to get my reading material in there. Excited to explore this.
You may want to check out Zotero for some feature inspiration - its Obsidian integration also imports citations, notes, and PDF annotations. Might be interesting to build a few of these in
Interesting perspective. Nice!