Why I take my time

I don't have to build this. I've got a job, an income, enough money; strackt doesn't need to succeed for me to be fine. There's no investor clock, no runway, no team waiting on a payroll. Nobody sets the pace but me.

So why put in the years? Partly because I like it — the challenge, the work, watching strackt come to life one piece at a time. But mostly because the freedom cuts the other way people expect. When nobody's waiting on the release, you stop building for the release. I'd rather wait until it's right than ship something sloppy and patch it later; done properly, it pays off in the end. That's what taking my time really means — holding out for the version I'd actually want to use myself.

And building the right tool is a long game — building, not just finding. strackt isn't one thing I wrote; it's a service stitched together from other products, mostly open source, each one picked to do its part. Choosing those well is a long game on its own, and that makes building strackt one by default. It was never going to be a "slap a script on top and call it done" job. The implementation is the easy part; anyone can wire something up and watch it work once. The real work is everything after — the testing, the finessing, getting each piece to fit and click. That finessing is what building a product actually is.

Because a platform can't live with "mostly." It has to be right every time, and that's exactly where the tools that look perfect fall down. Ansible sounds like the right tool for the job; running commands straight over SSH can seem that way too. And they mostly work. It's the long-term unpredictability that gets you — the one-in-a-hundred, sometimes one-in-ten, where the same input comes back a little different than you expected. When a tool does that, it has to go; you build something better and test it just as hard. That's how I landed on NixOS — not because it was fashionable, but because a rebuild does the same thing every time.

Some of that you can't rush anyway. The right pieces have to be found, tried, given time to mature; the details have to settle. But at some point you do have to launch. I could leave strackt in the cooker another four years — maybe it'd get better, maybe it'd just be what it already is. The honest truth is it's there now: usable, good to use, doing what it set out to do. I've finessed it to the point where I'd want to run my own things on it, day to day — not because I'd need to poke at it every day, but because I can trust it every day.

And that's the whole point. None of the work shows; nobody sees the tool I threw out, or the tenth rebuild that finally came back exactly like the ninth. They just see a server that quietly works. That quiet is the standard I was after — reproducible, right, something you can trust — and the only reason I could hold out for it is that nobody was rushing me. The effort nobody sees is the work. Taking my time is how it earns the trust.