Most of the software I use every day updates itself weekly. Buttons move. Menus get reorganised. A feature I relied on disappears into a settings panel. Each change is small, but together they add up to a constant low-level tax on attention.
This article makes a simple argument: software that changes slowly is easier to learn deeply, easier to trust, and easier to build on. That is worth more than most new features.
What "slow" means here
Slow software is not unmaintained software. It gets security fixes and bug fixes. What it avoids is churn: redesigns for their own sake, features that exist to hit a quarterly metric, and interfaces that assume you have nothing better to do than relearn them.
- It keeps its interface stable across years, not months.
- It stores your data in formats you can open without it.
- It treats new features as opt-in rather than default.
- It has a clear, narrow purpose.
Depth needs stability
You cannot get good at a tool that keeps changing. The keyboard shortcuts, the edge cases, the small tricks that make an expert fast all depend on the tool holding still long enough to learn. I wrote more about this in Tools That Think With Us.
The best tools disappear into the work. You only notice them when they change.
Building on stable ground
Stable tools are also better foundations. When I build a small script or workflow on top of a tool, I want it to keep working. Every breaking change turns a finished thing back into a maintenance chore. My Seedbank project is an example of a tool built on plain files for exactly this reason.
What I'm doing about it
- Choosing tools with long track records over new ones.
- Keeping my notes in plain text.
- Turning off automatic updates where I can, and reading changelogs when I can't.
None of this is new advice. But it is easy to forget when every product announcement promises something shinier. For the longer history of these ideas, see The Quiet Craft.