No non-trivial design is ever perfect, and tests are meant to test where the empirical reality (implementation) diverges from the design (expectation). It isn't

No non-trivial design is ever perfect, and tests are meant to test where the empirical reality (implementation) diverges from the design (expectation). It isn't clear at all how these two things are related in the way suggested, much less to the extent that they can be considered unified under an (as yet undiscovered) abstraction. Rather, I think the essay accidentally supports something else entirely: developers easily get distracted testing the wrong things and thinking about testing as being easier, more valuable, or less fallible than it is.