suggy - 10:38 am on May 2, 2011 (gmt 0)
enigma1 -- I have noticed the same thing. I think it's some kind of over optimisation filter.
If you remember that with Panda google was effectively targeting 'written for search' content (ie content that was really produced simply to harvest search traffic). And, writers/ publishers of this content became increasingly in SEO savvy. The result: they wrote and optimised for search terms.
With Panda, I think that google must be comparing the search term to the page and saying "nah, that match is uncannily good (over optimised) and the page meets these other indicative criteria (eg length, lack of rich media, formatting, etc) for content spam, so it's filtered out. Add some search phrase modifiers (or take the exact match specificity away) and this filter is avoided.
I definitely think that google is using content criteria to determine likely spam (ref points on graph quote from Amit?). Unfortunately, this seems to back fire on google when pages have little or no content, hence why we see abysmal pages top of SERPs in competitive niches. Basically, the competitive businesses have optimised themselves out of the race!