I'm talking about a level playing field between sighted and blind QA, they are both subject to missing problems if the test is poorly designed. The fact that "most" tests are broken is related but tangential to my observation.