Uses yeast-display deep mutational scanning to measure the effect of essentially every possible single amino-acid substitution in the SARS-CoV-2 RBD on folding/expression and on ACE2-binding affinity. Finds the wild-type RBD sits well below the measured fitness ceiling — many single mutations further increase ACE2 affinity — i.e. SARS-CoV-2’s RBD is not a computationally optimized binder, an independent experimental test of a “rationally designed for human ACE2” story (though it does not address non-computational routes such as serial passage/directed evolution). relevance_note: the direct experimental test of Andersen et al.’s “RBD is not predicted optimal” claim, run independently of any origin-debate framing.