Exactly right. The filter, or CIRCA, or stemming stuff, or whatever it is, has everyone chasing their tails. There is very little consistency. About the only thing you can say is that ecommerce is broadly effected, while noncommercial sites aren't even aware that there was a November update.
Google must be laughing their little asses off at that. Too bad they haven't quite got it to work so that the results are any good!
The stemming theories are really strange. No one has figured out whether it means you should add more stemmed variants of your keywords, or whether the new algo is so smart that it can recognize various stemmed derivatives as coming from the same thematic root, and thereby nail you on keyword packing extra fast. It does seem that certain two-word terms are really toxic, and simply breaking them up with a word in the middle, or making one or the other a plural, could be helpful.
But then, I suspect many of the "I'm back!" reports are premature. It could be the freshbot, which after picking up changed pages, is now injecting them into the SERPs independently of the filter, like it does with blogs. But be prepared -- a lot of those blog hits drop back down after they get their initial boost for being so fresh. I think a lot of the "I'm back!" people will see the filter reassert itself soon.
It's certainly a new algo rather than some bug, but the algo is very poorly implemented. An ecommerce randomizer for the SERPs could do just as well as all the fancy CIRCA self-learning software, from what I've seen.
I think Google will get away with this, as long as they don't start messing with the noncommercial SERPs. I'll go further -- I think Google intended to achieve something akin to a randomizer, whether for anti-SEO purposes or for Adwords/Froogle purposes. Perhaps Google knew how crummy the ecommerce SERPs would be, and that's why sites with a PageRank of 6 or higher were granted more immunity to the filter than sites with a more average PageRank.
It's starting to look more like a conspiracy to me, and the buzz about how the algo has changed may have inadvertently emerged as a convenient part of the coverup. The only people who believe that Google implemented a new algo in good faith are those who also believe that Google never tests these things off-line.
Yeah, Google knew. They even let the Scroogle exclusion trick broadcast its message for three weeks before taking any action. Or maybe they are so isolated in Mountain View, and so disorganized, and have so completely lost control over the far-flung ones and zeros of their data, that they really are confused.
Conspiracy or incompetence? This is six weeks after Florida, and there's not much wiggle room between these two choices.