I am mystified by the relevance of most of Kevin Hoover's points. 

His central point appears to be (his shout, not mine) THIS MAY IMPLY
THAT THAT THERE ARE SOME POINTS ON OUR CURVE THAT NO COMBINATION OF
EXOGENOUS-VARIABLE SETTINGS COULD EVER GET US TO.  BUT SO WHAT?  SUCH POINTS
WILL NOT BE MISLEADING, BECAUSE THEY WILL NEVER BE SELECTED AS PART OF THE
MODEL'S SOLUTION.  THIS IS NO DIFFERENT WITH THE AGGREGATE DEMAND CURVE ... IN GENERAL, NOTHING [would happen to AS if P were cut in half, M held constant] ... THE MODELS ARE CONSTRUCTED ON THE BASIS OF ASSUMING THAT SUCH ISOLATION [of AD and AS curves] IS EFFECTIVE IN A (sic) PARTICULAR CASES.

The textbooks mislead students because a "solution" which involves an instantaneous doubling of M/P would be non-sustainable (forces would be set up to push the economy away from that position).  Not only would such points "never be selected as part of the model's solution" they are not a sensible way to think about macro.  Therefore the textbooks should only derive an AD curve that corresponds to feasible M/P levels (or label the non-feasible sections, not AD, but MADD, Macroeconomic Attention Deficit Disorder, or "gravity-defying science fiction").

Are textbook writers entitled to assume that AD and AS curves have an independent existence?  It certainly complicated matters to assume otherwise (e.g. SRPC and LRPC curves in hysteresis models); but since M is held constant as P changes along only one of these 2 curves (AD but not necessarily AS), those who think it is legitimate to derive AD curve with fantasy levels of M/P invite the question: what would happen to AS if the "solution" occured in one of the MADD zones?  

Such AD "always and everywhere legitimate" advocates also invite questions about the second-round expectational feedback consequences (for an AS curve) of an AS shock (a new M/P associated with the new AD=AS "solution").     

Robert Leeson