I just stumbled onto this thread and want to add my 2 cents. I understand the opinions given but I worry about the lack of implementable solutions. The complaint is that a small number of high scoring books skew set rankings, rewarding instant spenders with thick wallets instead of long determined work. As has been noted, the more skewed the book scoring is, the harder it gets to counteract.
Book scoring is based on fair-market value which incorporates desirability and supply/demand. Seems fair. Book scoring should only be determined by the book and not by set inclusion or any other factors. That keeps it simple. The sum of a collector's books should determine registry ranking. A set score could count a book at a higher or lower value.
Every set is different with various combinations of high value and low value books. A 50/50 split of points and completeness could work for some but not others. Another proposal would be to use fraction_complete * set_points. If a collector has one million point book in a set of 100 books, he would score (1000000) * (1/100) = 10000, and he could basically double his score by adding one book of any lowly grade. With a 50/50 split he would score just over 500000, and adding one more lowly book would be a small improvement.
A different approach could be to reward points and completeness in separate awards, but I suspect they need to be mingled by some formula here as well. Quality and quantity should both matter.
Set scores could also use something like 1+log10(points) for each book. This would vastly cut down the spread, since a million point book would only count as 7, while a single point book would count as 1. The spread here could easily be modified to whatever desired.
I'm mixed about this whole topic. I admire expensive sets but I also admire complete sets of good quality. I don't want to devalue the hobby.