Yeah, I see the parallel. I treat every feature like a mini speedrun. First I sketch the exact behavior in a unit test – that’s the “finish line” I can hit quickly. Then I write just enough code to make the test pass, run it, and refactor. I keep the loop tight: write, test, fix, test again. I also profile early, so I know if I’m over‑engineering a hot path. And when I hit a roadblock, I break it into micro‑tasks that can be finished in 10‑minute bursts; that keeps the momentum. The trick is to avoid the “perfect first time” trap – I aim for a clean, working version fast, then polish it later.