There's a "right" size for any project and going over it's hard to avoid over-thought, over-complicated, user-confusing, pedantic, semantic things.

This isn't the trite "too many cooks" argument. It's that the right kind of quick-and-dirty shortcutters who value leaving things alone produce better products; they use simpler paradigms and abstractions, have better combinable primitives, communicate purpose more effectively, and produce overall better tools.

The corollary is the stupid thing you can hack to do what you want is often better than the smart thing that is very carefully designed to do what you don't - because one of them you can technically use and the other you can't.

This is why earlier versions are frequently better than later ones and earlier products seem to do more, better, and faster. It's why smaller companies seem to produce better things, and counter-intuitively, products that take less time to finish (say a few days) seem to do better and gain more traction then the 6-month effort.

Some things really are profoundly complex, like databases and operating systems. Most things however, despite much we'd like to pretend, are not in that class of software.

And yes, these claims are cultural bomb-throwing. But the evidence for it is solid.