you're mixing up 2 different results a) error-correction needs small level of errors to begin with to amplify signal - we finally got to that point, and larger

you're mixing up 2 different results a) error-correction needs small level of errors to begin with to amplify signal - we finally got to that point, and larger correction setup deals with more errors b) "standard" benchmark problem now 100% computes something uncomputable with classic chips (in practice) - the problem is that it's so quantum, neither it is verifiable with classic chips anymore