Mathematics, a formal discipline, has proved unreasonably effective as an instrument for change in formally unverifiable domains. The change delivered by the instrument may not be fast, it certainly won’t be obvious, and it will probably be far out of proportion to what we expect, but it will occur nonetheless.