ABX testing is useful if, and only if, only the variable(s) that are supposed to be under test are changed. This yields excellent results with many true scientifc processes.
Unfortunately,audio ABX testing with the typical setup introduces uncontrolled variables (almost always a common input signal, or common ground path between chassis or gain differences) that make the test setup invalid.
The sad fact is that the technoflakes seem not have the curiosity to learn and discover the real reasons why audio playback performance is not perfect, and don't seem to care, and the audiophlakes are pursuing absurd magic beads and bangles to make things better, and they too seem to have no curiosity about what is really going on and how to make real improvements.
Frank Van Alstine