The method of pilot projects that are then scaled up is very mature, well understood, and universally applied. It works, way, way, way better than thinking hard.
While there is some truth in your statement, a lot of unintended consequences only show up after considerable time and at scale, and then only if you know where to look.
The original steam engines standing in as a pilot project didn‘t do much to the atmosphere. Billions of combustion engines did.
The great insect apocalypse went almost unnoticed until two decades in and an order of magnitude of change.
Some things are nonlinear (that's kind of what "second order" means). With a pilot project, it's easy to observe things that are O(n), but not so easy to observe things that are O(n^2).
So sure, do a pilot project. Let's make sure that even the first-order stuff works the way we expect. But also think hard about second-order effects. And even after that, continually watch for un-thought-of second-order effects emerging as (and after) you roll it out.
