(I carefully wrote this before Tom D. responded, so I'll post as written. I see he has refuted some of what's here)
As advertised, I flipped through the pages of the Confidential Guide, counting 7s. There are about 100 in the book. It was an interesting exercise. Tom Fazio is the modern architect with the most courses ranked 7 or better, though the courses Tom Doak selected as 7s (Wild Dunes, Wade Hampton, Pelican Hill, for example) are not the current darlings of the "best of" lists. You would either conclude that Fazio builds a lot of 7s, or that much of his his modern work would be regarded as 6s. Nevertheless, there's a great deal of respect there.
Let's say that Tom missed a few in his travels. Add in the second Golden Age courses from 1995-2006, plus some sympathetic restorations of classic courses, and maybe there are 150 courses worthy of a 7, with about 80-100 in the U.S.. That's 0.5% of all courses. Eric, I think we're still looking at a bell curve centered around 3 or 4, truncated at 0.
There was a recent thread that the "best of" lists experience a significant drop in quality about halfway down the list, that any of 300-500 courses could represent the 75th best modern course. This would support Doak's delineation between 6 and 7, that once you get two thirds of the way down the list, you run out of excellent courses. I think it shows that raters need to figure out a little better what the next best courses really are. Where does "very good" actually become "excellent"?
I'd like to think there are quite a few 7s being built these days. Let's take an example. I played Colorado GC once last year, and though it did not thrill me, it is an excellent course, with a great deal of variety and challenge, with a reasonably interesting walk. It is also artistically built and shaped into the terrain. I thought it was about a 6 when I played it, but a strong case can be made to call it a 7. Anyway, this is me making my usual case for the modern architects. They don't get the best pieces of property, but a lot has been learned in the last 20 years, and excellent courses are being built.
All of this analysis of Tom's method may drive him crazy. It's his own personal rating system which became popular with his friends, so he published it. On the other hand, the Golfweek rating system is similarly designed, so the method has been adopted.