I never used Obsidian.
I watched people get exalted over it. The graph view. The backlinks. The plugins. The community. Fair enough – I could understand the rave. But for me it always looked overburdened. Features I had no idea about. Workflows I did not need. A knowledge management system for people who wanted to manage knowledge. I just wanted to write things down.
I used Apple Notes. I used Bear. I used whatever was already open when a thought arrived. The thought was the point. The tool was irrelevant. I did not need a second brain. I needed a notepad.
Then I built note2cms. Four endpoints. Markdown in, permalink out. The protocol did not care what notes app you used. It accepted an HTTP POST with a JSON body containing Markdown. Where that Markdown was composed was not the system’s concern.
Then a stranger in Russia – @garrul4ik – looked at those four endpoints and saw something I did not see. He saw an Obsidian plugin.
The Plugin I Did Not Imagine
He built it overnight. While his wife hissed at him. Without coordinating with me. Without reading my source code beyond the API documentation. Without asking permission.
The first version was simple. Open a note. Press publish. Get a permalink. The plugin prepared the Markdown payload, attached the bearer token, and POST-ed to the endpoint. Four lines of logic wrapped in Obsidian’s plugin API.
Then he kept going.
Publish from a specific folder. Publish by tag – mark a note with #publish and it enters the queue. Bulk publish everything in a directory. Preview before publishing. Offline queue for when you have no connection. WiFi-only mode to avoid publishing over cellular. Queue management to clear failed attempts. Permalink modal with copy-to-clipboard. Post management with search and delete. Mobile confirmation dialogs. Automatic frontmatter injection when the title is missing. Error correction when the user forgets a required field.
Version 1.0.0. Then 1.0.1. Then 1.0.2. Then 1.0.3. Four releases. Nineteen commits. Tested on Windows, macOS, Linux, and Android. Submitted to the official Obsidian community plugin registry.
I did not ask for any of this. I did not spec it. I did not review it. I did not even know it was happening until he tweeted about it. The architecture invited the contribution by being simple enough to build against without guidance.
The Recursion
And then something happened that I did not expect.
I started using Obsidian.
Not because of the graph view. Not because of the backlinks. Not because of the knowledge management features that I still do not need. Because of the plugin. Because someone built a publishing tool inside Obsidian that was so good – so polished, so thoughtful, so mature – that it made Obsidian worth using for the first time.
I installed Obsidian to test the plugin. I configured the plugin with my API endpoint and token. I opened a note. I wrote frontmatter. I pressed publish. The permalink appeared in a modal. I copied it. My blog post was live.
And I thought: this is actually good. This is better than curl. This is better than the iOS Shortcut. This is what publishing should feel like – write in a real editor with real Markdown support, press one button, get a permalink.
The tool I built made someone else build a plugin. The plugin made me adopt the app. I am now an Obsidian user because of a feature that exists only because I built a CMS that I built because I did not want to use Obsidian.
I made Obsidian good for me by building something that had nothing to do with Obsidian. Someone else connected the dots. And now I am on the other side of a bridge I did not know I was building.
The Shape of Recursion
This is not a circle. A circle returns to where it started. This is a spiral. Each pass through the loop is at a higher altitude than the last.
Pass one: I do not use Obsidian. I build note2cms because I want to publish from any notes app.
Pass two: Someone sees note2cms and builds an Obsidian plugin because he wants to publish from Obsidian specifically.
Pass three: The plugin becomes so good that it makes Obsidian the best way to use note2cms. The generic tool acquires its best client through a contribution the author did not request.
Pass four: I adopt Obsidian. The person who built the “publish from anywhere” tool now publishes from one specific place – because the community made that one place the best place.
Pass five: My adoption validates the plugin. The plugin author sees the protocol’s creator using his plugin. The bazaar’s founder shopping at the bazaar’s stall. The signal is unmistakable: this is good enough for the person who built the thing it connects to.
Pass six: Other people see the creator using the plugin and adopt both. The recursion becomes a flywheel. note2cms makes the plugin possible. The plugin makes Obsidian worth using. Obsidian usage drives more note2cms publishes. More publishes prove the protocol. The proven protocol attracts more plugin builders.
Each pass adds value that did not exist in the previous pass. The system is feeding itself. Not by design. By architecture.
The Architecture of Accidents
I did not design this recursion. I did not plan for an Obsidian plugin. I did not imagine that my CMS would make me adopt a notes app. The entire sequence is accidental.
But the architecture made the accident possible. The four-endpoint API with a bearer token is so simple that building a client for it is an afternoon’s work. The Markdown-in contract is so universal that any notes app can satisfy it. The permalink-out contract is so clean that any plugin can display it.
The architecture did not specify what clients would be built. It specified the interface they would speak. The interface was simple enough that the right client appeared without being requested – because building it was easier than not building it.
This is Raymond’s bazaar in its purest form. The plausible promise – a working system with a clear interface – attracted a merchant who built something the market owner could not have imagined. And the merchant’s product improved the market so much that the market owner became a customer.
You cannot design this. You cannot plan for it. You cannot put “create conditions for recursive community value” on a product roadmap. You can only build something honest, make the interface clean, and wait.
The recursion is the reward for architectural honesty. The community built something better than what the creator could have built alone. And the creator was humble enough to use it.
The Confession
I am writing this post in Obsidian.
I am going to publish it with the plugin that @garrul4ik built. The plugin will prepare the Markdown, attach my bearer token, POST to my endpoint, and return a permalink. The pipeline will parse the frontmatter, render the HTML through the Terminal theme, push to my GitHub repo, and deploy to GitHub Pages.
The notes app I never wanted to use, running a plugin I never asked for, publishing to a CMS I built because I did not want to use any particular notes app.
One big beautiful recursion.
And I still do not use the graph view.
You cannot design a bazaar. You can only deserve it. And sometimes, the bazaar builds something so good that the person who built the bazaar becomes its most grateful customer.