Google says they’ve managed to neutralize the impact of a lot of Googlebombs.

A Googlebomb is a when a group of people get together to link to a certain site with a certain set of keywords, in order to convince Google that it ought to display that site at the top of search results when queried for these keywords. (This often worked because Google does a lot of backlinks analysis when they determine their search rankings – the words you search for need not be on the page itself which ranks well for it.) In the past, for example, people tried to connect George W. Bush’s official biography page with the word “failure”... and a counter-bomb tried to connect Michael Moore’s homepage with the same query. The impact of this trick can be increased by telling people who don’t know how Google works to “enter failure and hit the I’m Feeling Lucky button!”

Now Google always claimed that the impact of these Googlebombs wasn’t big in relation to all searches performed on Google; Google’s Marissa Mayer said Googlebombs are “distracting to some” without affecting “the overall quality of our search service,” and Matt Cutts agrees that tackling bomb phrases wasn’t high priority as they were “well off the beaten path.” But the truth is that the bombs still received major spotlighting within recurring blog and mainstream publications over the past years – and a lot of people misunderstood the Googlebombs, thinking it was Google who was delivering their opinions... a typical killing the messenger problem.

How did they attempt to do it?

Google doesn’t share exactly how they tried to defuse the impact of bombs. They reveal that they applied a scalable algorithm, as they prefer to do, which theoretically works against any kind of Googlebomb, so there’s no need for manual moderation of search results (only manual “moderation” of search result algorithms). Googlebombs made use of a core approach of Google’s algorithm: link text counts. To understand how they were able to detect Googlebombs, we need to ask: what makes a Googlebomb stand out? Which specific attributes differentiate a Googlebomb from a normal popular phrase for link texts?
Part of the answer may be found in some network analysis. I can imagine that Googlebombs have some criteria in common, like that they are always using the same link text (that much they share with some non-bombs), but that they also look like “bridges” from otherwise unconnected networks. Why would a large number of networks (neighborhoods/ communities) which are not tightly connected to a certain other network through general links show a distinct, repeated link into this opposing network? To use the “failure” example: why would the networks of US Democrats websites suddenly get together to link into an opposing Republican network?

How successful were they?

Google now says that instead of the Googlebomb target, they will “typically return commentary, discussions, and articles about the Googlebombs instead.” So how successful are they in that task? From the traces of the past that I was able to assemble, here’s an overview of the performance over time.* The short answer is that it’s hard to tell how successful the latest algorithm update was – some results remain bombed, some results were successfully changed, and some results were already “defused” in the past.

*Results may vary depending on your location and the Google datacenter you hit.

[Hat tip to Search-Engines-Web.com and Alek!]

[By Philipp Lenssen | Original post | Comments]

Bookmark/Search this post with:
| | | | | | | |