Every extra file I have to look through when programming is an added cost; abstractions add more than just CPU overhead. The highest-priced EC2 instance on Amaz

Every extra file I have to look through when programming is an added cost; abstractions add more than just CPU overhead. The highest-priced EC2 instance on Amazon is under $5/hr; you won't find a decent programmer whose time is worth anywhere near that little. This may be relatively minor on a small project, but adds up quickly. It's extra overhead when onboarding new developers, more time spent debugging (deeper stack traces, etc.) more time simply waiting for grep to run. I'll argue that you actually spend more time refactoring as well, since you're encouraged to spend more time on the upkeep of this over-abstraction for the sake of "clean code" rather than getting real work done. Probably more important is that you don't know your six-months-from-now problems now, you know them in about 5.9 months. I've wasted a lot of time building the wrong abstraction layers because I made bad assumptions about our future problems. At first, I thought that was a failure of my programming abilities; I've come to realize that was my failure of my psychic abilities. This isn't an attack on you or your style, merely a few observations about what kind of effect it would have on my own work. If it works well for you, awesome! However, it seems like most of this would only be really effective by v3 or so of a product: v1 answers "is this useful and worth investing more in", v2 covers "ok, let's turn this into a real product", and v3 is that giant scary internal rewrite (that goes great thanks to your now-high unit test coverage) that creates what you would have made from day one if you actually knew what you needed.