Given that I’ve seen excellent mathematicians produce poor-quality code in real-world software projects, I’m not sure how relevant these benchmarks are.