Google Defuses Googlebombs

John C. Worsley writes "Google announced today a modification to their search algorithm that minimizes well-known googlebombing exploits. Searches on 'miserable failure' and their ilk no longer bring up political targets. The Google blogger writes: 'By improving our analysis of the link structure of the web, Google has begun minimizing the impact of many Googlebombs. Now we will typically return commentary, discussions, and articles about the Googlebombs instead.'"

Comments

Big changes?

[LiquidCoooled]

Surely this changes lots of things.

If google is now discounting the wording other people use to link to a page, then isn't google themselves becoming like old fashioned engine, ie only specifically accounting for information on the actual page and not based on what other people who link to this page thinK?

By improving our analysis of the link structure of the web, Google has begun minimizing the impact of many Googlebombs. Now we will typically return commentary, discussions, and articles about the Googlebombs instead.

reworded becomes:

By ignoring the link structure of the web, Google has begun minimizing the impact of many Googlebombs. Now we will typically return only results which are from the actual page itself rather than looking at how other people link to each other.

A googlebomb is not a bad thing, its making use of the algorithm to expand the keywords which a page is associated with.

Sidenote:

I did a search for google, and the snippet that comes up under each google entry does not exist on the page itself, where does it actually come from?

for example:

Google
The local version of this pre-eminent search engine, offering UK-specific pages as well as world results.
www.google.co.uk/ - 4k - 24 Jan 2007 - Cached - Similar pages


I thought google weren't meant to display a different page to bots as to users? (didn't they get in trouble for something similar not so long ago?)

Re:Big changes?

[drinkypoo]

If google is now discounting the wording other people use to link to a page, then isn't google themselves becoming like old fashioned engine, ie only specifically accounting for information on the actual page and not based on what other people who link to this page thinK?

Not necessarily. It depends, like most other things, on how it is handled. The practice of googlebombing is the practice of mob rule and google quite sensibly worked to put a stop to it.

A googlebomb is not a bad thing, its making use of the algorithm to expand the keywords which a page is associated with.

A googlebomb IS a bad thing, it's a group of people with an agenda railroading the functionality of a resource upon which the health of the internet depends in a very real way. Again, it's mob rule; a certain segment of the population runs away with the whole idea.

I thought google weren't meant to display a different page to bots as to users? (didn't they get in trouble for something similar not so long ago?)

Are you saying that bots are getting different search results than users? Because absolute shitloads of websites serve different versions of their pages to google for a wide variety of reasons. For example some premium sites allow google to index part of their content in order to rope people into buying a subscription.

Re:miserable failure

[inviolet]

Google bombs don't have much to do with PageRank. They're about link text being abused.

I'm with you on this one, but it also makes me wonder...

The purpose of link text is to impose additional, personal meaning on a link, like this: "Today in the news we learned about Windows monoculture". The "Windows monoculture" link text is my own meaning imposed on the link. Google is, or at least was, putting some trust in that imposition: Google would elevate that slashdot page's ranking under the category of "Windows monoculture", on the assumption that I'm probably not misrepresenting its content.

A google-bombing can therefore occur without any conspiracy: if lots of people imagine themselves witty for jokingly linking the phrase "miserable failure" in their blog to www.whitehouse.gov, the result is an unintiontional google-bombing. And as other posters in this thread have pointed out, there is some truth value to that.

Now we hear that Google is changing this, which means paying less attention to my link text, and instead devoting more computation towards analyzing what the target page is actually saying. I suppose Google is going to read the slashdot page I linked, and decide for itself what it's about rather than taking my word that it's about Windows monoculture. That's got to be computationally expensive.

It's the same general problem as we see in academia with scholarly references. Let's say some guy writes a thesis and uses some other paper as a reference, claiming it lends support to the new theory. We can trust his citation (i.e. Google can trust the link text), or else we can mistrust him and go and dig up the reference text and read it ourself.

Obviously that kind of mistrust is expensive (but isn't all mistrust?)... but after a certain amount of abuse, it's a price we have to pay in order to maintain the same degree of certainty. As for rethinking, they're doing this all the time at Google. They're constantly updating their ranking algorithms.