littlebabyjesus
one of Maxwell's demons
I find these things tough to fully understand, but each time I see grand claims for quantum computational supremacy, the test used seems to have been specifically set up to be suited to a quantum computer - ie you ask it a question about quantum states, in this case a quantum circuit. So it can model this in a way a classical computer clearly cannot as we're talking vast Hilbert spaces of possible classical states. But I'm not entirely sure what this proves. It feels like a cheat.