Tom,
Not sure why you would single me out. All I recall writing on this subject is that Robert Dedman, founder of Club Corp (now Invited) once told me that returning nines allowed an extra 3500 rounds per year, so he insisted on that. For all but the top few courses, golf is a tough biz, so I did follow Mac's first rule, including the first part, "Where possible." Put another way, architects should be concerned with the fiscal as well as the physical in design, so on most sites, returning loops isn't particularly difficult and if easily done, why not? I know you like(d) your iconoclast positions (and they have worked for you quite well) but for most courses, this is a good idea.
I have done a few non-returning courses, including Wild Wing Avocet and Cowboys Golf Club. At Cowboys, it just didn't fit, but it cost over $500K to run water and sewer down to the halfway house, which no one on the team had anticipated and which generally, I would prefer to put in the golf course. That obviously varies with the site and situation of nearby utilities, but it is a cost to avoid if you can. At Wild Wing, at least the cost was split among adjacent courses.
Philosophically, is a "perfect" golf hole (if there is such a thing) is more valuable than a similar one, even of slightly less quality than a large magnitude unanticipated expense? Again, it varies by the site and the type of course, but for most courses budget prudence really is a bigger design concern than some theoretically better hole.
If you can, I would like someone to post a quote of the Doc's repudiation of rule 1. I suspect it came after he routed CP, which is a case where returning nines wouldn't have worked or would have been far less satisfactory.