Given that it's not possible for a machine that represents numbers as any finite number of bits to accurately represent the uncountably infinite set of real num

Given that it's not possible for a machine that represents numbers as any finite number of bits to accurately represent the uncountably infinite set of real numbers, it's guaranteed that there are going to be conditions where floating point arithmetic produces inaccurate results. And given that it's not possible to avoid the floating point processor being mildly broken in some conditions, I suppose the least-bad solution is for the M1 to be mildly broken in exactly the same way that the existing Intel processors are, rather than being mildly broken in a completely different way. Principle of least surprise.