Just as an example, the keyword "Yahoo" only shows 49 results even though there's
over 294MM results total. What I think is happening, is that Google always returns 1000 results (assuming at least 1000 are available). It then removes all instances from the 1000 which have more than 2 results on the same domain and as a result the actual results drop significantly at times.
Like for yahoo, unfiltered results are all on the same set of domain/sub-domains:
http://www.google.ca/search?num=100&...meta=&filter=0
WG