Well I imagine what OP meant was that the testing phase in this "science" is not verifiable/repeatable/etc. Perception is an awful way to perform tests, and is very frequently wrong.