Two weeks in, I'm starting over.
Not because something broke in a dramatic way, and not because I made a catastrophic technical mistake. The process itself became unmanageable, and I recognised that early enough to do something about it.
What happened
After Day 0 and Day 1, I kept going in several directions at once. I set up a VPS on Hetzner and configured Nginx, WireGuard, Docker Compose, CI/CD with GitHub Actions, and GlitchTip for error monitoring. I built a landing page at orangustudio.net. I started a vault system to track decisions, daily logs, and article drafts.
On paper, it looked like meaningful progress. In practice, I was spending more time managing infrastructure and producing content than writing actual application code — which is the only thing that actually matters at this stage.
During the implementation I had annotated 16 topics for articles, with the intention of posting them as I went. But gradually the philosophy shifted toward "publish everything once the milestone is ready", and the backlog kept growing without anything going out. The vault became its own project to maintain. The content pipeline became a second job running in parallel with the actual product. None of that was sustainable for one person.
Why I took everything down
I shut down the VPS, deleted the DNS records, took orangustudio.net offline, and removed the repositories from GitHub.
The landing page was the clearest signal that something was off. I had built a polished front door for a product that didn't meaningfully exist yet. A beautiful landing page with nothing real behind it isn't build in public — it's just a facade, and I didn't want to maintain one.
The infrastructure work wasn't wasted, though. I now know how to set up WireGuard, how to structure Docker Compose for a Django project, how to wire CI from day one, and how to think about a VPS deployment end to end. That knowledge doesn't disappear when the server does — it's already changed how I make decisions.
But keeping all of it running was consuming attention that should have gone to the product.
What I'm keeping
The domain. The Cloudflare account. These dev.to articles and the X threads that are already published — they are real records of real work, and I'm not pretending they don't exist.
The decisions I documented along the way — why AGPL, why split settings, why uv — are still valid and will carry forward. The mistakes I made are just as useful as the things I got right.
What changes
The rule going forward is simple: build something, then talk about it — not the other way around. A single commit can be a post. A small decision can be a thread. There is no milestone required to publish something, and there is no backlog allowed to accumulate.
Before, I was setting up monitoring before I had errors worth monitoring, writing drafts about features that didn't exist yet, and planning infrastructure for traffic that wasn't coming. That kind of forward-planning is useful in a team, but for a solo project at this stage it just creates overhead without value.
Starting over is not failure
There's an art form in Tibetan Buddhism called a sand mandala. Monks spend days or weeks creating an intricate geometric painting with colored sand. When it's finished, they sweep it away.
Starting over is not erasing the work. It is dissolving the mandala so the next one can be made with clearer hands.
The point isn't the artifact. It's what the process leaves in you.
I'm not restarting from zero. I'm restarting from experience.
Top comments (0)