Thanks to everyone who replied for their comments, they've been very helpful. I'll quote dgaies here, but I realize many others made similar points.
Originally posted by dgaies The actual average numerical rating are pretty much useless in my opinion for a variety of reasons. The primary reason is that one person's "9" might be another person's "7" since there's not universal method by which individual rate the lenses. Also, some people rate things relative to the value. So something like the DA35/2.4 might get a "10" because it's a great lens for the money while the FA31 might get a "9" from the same person because for $1000 they expected a little more. That doesn't mean the person thinks the DA35/2.4 is a better lens than the FA31.
Bottom line, you're best off reading the text of the reviews and various threads on the forum and using that information to make a decision.
I agree with nearly everything you've said. And please note, I did say that comparing relative ratings was
one influence, not the only one.
The two reasons you give for the numerical rating being useless are valid, but by themselves are not the reason for the average rating being useless. One person's 7 being equivalent to another's 9 just means you can't assign an absolute value to a rating. Lowell says that for him a $30 lens rated at 8 means "capable" of taking exceptional photos if you know how to use it, but there's probably at least some people who have a slightly different scale. (If you take an average over lots of people of what a rating 8 implies for a $30 lens, he might be spot on. That's not the point.)
The fact that different people determine a rating based on different mixtures of quality and value is also, by itself, not a major issue. It just means the rating is neither a pure quality score nor a pure value score, but some unspecified mixture of both. I don't have any problem with that. It's probably a good thing.
I think that the real problem with the average rating arises from the fact that, in combination with the above differences between individuals, there are different sets of individuals rating the different lenses. (That's clearly inevitable.) This can lead to the very counter-intuitive result that
everyone who rates both lenses gives lens A a higher rating than lens B, using whatever personal rating scale they have and whatever balance between quality and value they prefer,
but the average rating for lens A ends up less than that for lens B, because of the influence of those reviewers who rated only one lens or the other. (Edit: Apologies if you saw my first two edits. I was being dumb. I put the text back the way it was originally. I need sleep.)
So getting back to the two specific examples I mentioned originally, if you had to rate both the 18-55 WR (average reviewer price $134.14) and the 18-55 AL II (average reviewer price $99.29), would you really rate the former 0.63 points worse than the latter? Possibly, but I'd be really surprised if that was common.
Now, the 18-135 is a lot more expensive than the 18-55, but when I read the reviews for the 18-135 I didn't get the impression from those that compared it to the 18-55 that it was of lesser value. In fact, I got the distinct impression they were quite happy with their purchase. I did just go back and read the reviews for the 18-55 AL II, and many of those used a phrase something like "great lens for the price", so perhaps exceptional value explains the slightly higher rating for the 18-55.
Overall, I got the impression most people don't care much for the average ratings. Although I don't either, it's hard to ignore completely. I can't mentally compute my own average for the "believable" ratings only, so it's tempting to think of it as a valid summary.