Yesterday Ben wrote an entry named Google Cloaking Its Own Pages! where he discribes the forums discussing the slip up on Google's part. The thread at WebmasterWorld on the second page, message # 27 by GoogleGuy, we hear the official Google response.
Those pages were primarily intended for the Google Search Appliances that do site search on individual help center pages. For example, http://adwords.google.com/support has a search box, and that search is powered by a Google Search Appliance. In order to help the Google Search Appliance find answers to questions, the user support system checked for the user agent of "Googlebot" (the Google Search Appliance uses "Googlebot" as a user agent), and if it found it, it added additional information from the user support database into the title.
The issue is that in addition to being accessed via the internal site-search at each help center, these pages can be accessed by static links via the web. When the web-crawl Googlebot visits, the user support system thinks that it's the Google Search Appliance (the code only checks for "Googlebot") and adds these additional keywords.
That's the background, so let me talk about what we're doing. To be consistent with our guidelines, we're removing these pages from our index. I think the pages are already gone from most of our data centers--a search like [site:google.com/support] didn't return any of these pages when I checked. Once the pages are fully changed, people will have to follow the same procedure that anyone else would (email webmaster at google.com with the subject "Reinclusion request" to explain the situation).
Many other forums and blogs are discussing this. Mikkel at Search Engine Watch Forums asks some tough questions. ThreadWatch.org has their normal humerous look at the topic. And even the folks over at SEO Chat are discussing it.
