This is a thought provoking question and I don't think there is a clear and distinctive answer since as a result of many parameters they are not mutually exclusive. Sean is right, and I too can't separate the two. I think some answers are influenced depending on the type of soils/sites an architect has worked with. I doubt there are many golfers who really could tell the difference, let alone care, as long as the course was firm, fun to play and memorable.
JES II frames it well with a great example at Huntingdon. If the site has a lot of great natural movement and I don't mean just broad sweeps, rolls and troughs, but smaller interesting character, with plateaus, benches...almost containing internal contour within the bigger site picture, then I believe in the long run this would outweigh the less than ideal soils. This has been proven and tested on countless courses across the mid-west, northeast and upper mid-atlantic.
I remember Peter's thread on soils and 'on the surface' I was initially swayed by great soils as my first choice, but this has to be considered in the context of the site. If given the option, I would take a great site in an out of the way location, you know what I mean...the type where all 18 are just waiting to be found, before I took a flattish sandy site in a heavy demographic area. If we don't disturb the natural drainage patterns and soil conditions/dynamics, then I think we will be far ahead in the long run and the course will be built cheaper. Sure, sand can be molded to create special cool features and made to look natural, but this usually comes at a higher cost. Tom Doak makes a good point and if the possession arrow is pointing in the direction of great soils on a decent site you have a good chance of building something special...it is up to the architect to bring out the best of the site.
"I think golfers and golf course architects alike need to look with fresher eyes at the possibilities and charms and economic good sense of the modest, non-sandy site -- though I'm not sure which of those two groups needs to lead or which is likelier to follow"
Peter, I think some of this is currently being accomplished, but I do agree that this needs to happen more often. The risk falls to the owner/developer to find the right site at a decent price, NOT EASY! I think the modest, non-sandy sites site lead during the Golden Age and for many years after, but it seems for the past several years, that developers have learned that the combination of geat soils and a good site may outweigh location. Sean also makes a good point as well, someone still needs to provide good golf to the areas between the sand and make it accessible.