Okay, since this is hot topic in SFWA right now -- actually, it's always a hot topic in SFWA -- I might as well pull on my asbestos gloves and handle the flames. Does the best work get nominated for Nebulas? If not, why not? Is it a conspiracy, an old-boy club, a cabal, a crooked process? Or not?
As usual for me, I'm going to come down in the middle, with a resounding yes-and-no. The situation is more complex than the conspiricists seem to think. Does the best work of a given year end up on the ballot? No, not always. Is that the result of people manipulating the process? No.
Here's my take on what happens: When a writer first enters the field, he or she is dazzled (I know I was). It's thrilling to go to cons and meet other writers. Usually, the writers you meet most are the ones also new to the field, of about your own age and publishing level. This is because the older, more established writers have already made groups of friends, with whom they go off to dinner and sit with in the SFWA suite (at least, in the days when it was possible to sit in the SFWA suite, instead of being jammed in sideways). So our new writer meets other new writers.
Then he or she goes home and reads these people. That's natural -- they're new friends, you're curious, and there simply isn't time to read everything published in SF. So you read your friends first, and you can't recommend what you haven't read, so you recommend your friends for awards. Not because of log-rolling, but because those are the stories you're reading.
Meanwhile, more established writers recommend fewer stories. There are exceptions, faithful SF writers who still read a whole lot of SF (Mike Resnick, Jack McDevitt). But most writers I know say that as they age, they read far, far fewer stories than when they were younger (and, often, read more non-fiction, but that's another post). So they recommend fewer things.
This is not, of course, a total explanation of what ends up on a ballot. There are many other factors, and sometimes a story is so terrific that word of mouth spreads and everybody reads it. But I think my explanation does account for a lot of final-ballot choices -- and without charges of log-rolling or corruption. Just an imperfect process wielded by imperfect humans.
You are, of course, free to disagree with me.