I found it unsatisfactory. He is correct to discount the use of an unknown to explain another unknown. However, he then precedes to do exactly that not once but twice. Nether divided psychological intentions or the valsava maneuver are more salient than "genetics" -- all three are obscure features about the human body that we can observe and describe mechanistically but are unable to discern an etiological relationship to stuttering. Correlation does not imply causation (and I would probably argue that divided intentions and higher rate of valsalva maneuvers are the *result* of stuttering rather than the cause).