> So "useful math must be motivated by practice" is empirically false

That was not the claim. The claim is that useful physics originates in measurements.