1script - 3:17 pm on Sep 2, 2010 (gmt 0)
I also think that this concept, if it's indeed a real thing, should work on all keywords, not just a subset of them. For one thing popularity of a keyword is a fleeting thing, it's popular one day and dead the next. To measure popularity by the number of AdWords ads on SERP (page) would reek of collusion and I don't think they would do that (but who knows). What else is there to reliably measure keyword popularity?
I donít think itís only on popular keywords.
So, yes, I think throttling down of Google traffic is real and applied to ALL your keywords. Exactly how it is applied (how the keyword pie is sliced or at least how many slices) probably greatly depends on your site's trust and other quality-related properties.
I also think that keywords get into their respective slices randomly, despite that it sounds crazy. Looking at keywords that continued to rank vs. those that were dropped I'm yet to find any reason for why they were allocated that way. I'm looking at # of internal and external BLs and whether the KW is in the title, h1 tag and how many times on the page. Those parameters appear to be exactly the same for KWs that still rank vs. dropped ones. Should I be looking at something else? If anyone knows, please enlighten me!