Friday I reported that Google would be treating sub domains like sub directories. That turned out not to be 100% true. Matt Cutts clarified at his blog that Google tweaked their “host crowding” algorithm to reduce the amount of search results shown from the same domain.
The examples I gave in Friday’s post reflect the new algorithm, so the point is the change went pretty much unnoticed by most SEOs. Matt explained that the algorithm “change doesn’t apply across the board.” You will still see google.com results dominate a search on google, as you will for ibm.
Ionut Alex Chitu feels this change resulted in Google finding less total results per query. I am not sure if we can relate the two, since Google is always fiddling with those total results figures and it would be almost impossible to isolate the variable that controls that.
Postscript: See How Changes To The Way Google Handles Subdomains Impact SEO which goes into depth about the changes.