The Great Google Algorithm Shift
Google rolled out its latest index at the very end of last month, and changes to its search algorithm have caused a huge outcry on some of the search engine forums. Interestingly, however, is that during the same time this happened, I received only one complaint from a reader about the change, this being a web site owner concerned that the ranking swap had hurt him. No searchers wrote to complain that Google had suddenly become useless.
OK, now I've set myself up to get those complaints. Feel free to fire away! Absolutely, when webmasters post about changes, you have to pay serious attention to what they say. I've always said that there are two "canaries in the coal mine" when it comes to how well search engines work.
The first canary is/are researchers who depend on search tools as part their everyday work. If something changes -- and for the worse -- they definitely begin to complain. The other canary is/are site owners, who often depend on search engines for their traffic. No doubt, they have a vested interest in saying that a change that hurts them personally must be bad. However, they also honestly know sometimes what's really relevant for the queries they target better than anyone. You cannot simply dismiss their complaints.
It's certainly not hard to find oddities at Google. For instance, "house" brings back things like the White House, Random House, Columbia House, the House of Blues and the US House of Representatives. These are big sites with "house" in their names, but other than the US House of Representatives, I doubt anyone would consider these relevant for just the single word "house."
Of course, what IS relevant for just the word "house?" Before condemning Google, take note that AltaVista and AllTheWeb.com picked up some of the same sites as did Google, and the remaining ones didn't scream out great relevancy. Teoma was dominated by pages from inside the US House of Representatives web site. I also don't know what the query "house" brought up at Google before the change. It may have been the same as now.
Here's a different example: "fish." The second site listed at Google is AltaVista's Babelfish translation service, which has nothing to do with fish other than having the word as part of its name. It certainly stands out from the other sites, which all seem pretty relevant to the word "fish" in some way. Is this an example of Google's new algorithm letting something bad through?
I don't know. Once again, I have no idea what the results were for "fish" at Google before the change. The algorithm swap may not have caused this. Or, if it did, it may have improved things on other searches I haven't seen.
For its part, I've had an initial briefing with Google which confirms that they have made some algorithm changes. Of course, algorithms changes are always on going, so anyone who feels the results got worse after the last monthly update may find improvements will be coming. It's also important to remember that the index is always undergoing changes, as Google is updating a selected number of pages much more frequently than once per month -- and that number of pages is growing.
I'll be coming back to tell you more from the Google-side of things next newsletter. In the meantime, here's a rundown on articles and comments about the issue.
WebmasterWorld.com: Google News
There are many different threads from webmasters on the changes. Scroll down to the "New update, pagerank death?" and "What do you think of the quality of Google *non-commercial* searches?" threads for two good ones.
And what do librarians think? Apparently, any changes haven't been noticed by the members of this highly regarded librarian newsgroup. Scanning the subject lists for September and October 2002, I found no mention of Google search quality degradation.
Google explains new page rankings
The Register, Oct. 8, 2002
"The effect was to bump obscure but densely-linked sites high in the rankings,"states this article. Maybe. But I had one reader who recently sent some good examples of "bad" searches that contradict this statement.
One of these was a search for "new york giant tickets." The problem with that search isn't that "obscure" sites have moved high but instead major sites with little or nothing to do with buying tickets have moved up, such as the New York Times sports section, SportingNews.com and the New York Mets web site.
To its credit, Google does list the official Giants web sites among the top results -- though barely, at number nine. In contrast, AllTheWeb.com puts it at number two. All the other pages at AllTheWeb.com also seemed to be exactly focused on the subject of purchasing tickets, as well. However, three of those links were dead -- and these were also all from the same web site, giving it more of a presence in the top results than I believe AllTheWeb's setting are supposed to allow.
Google Degraded? Geeks Aghast
Wired, Oct. 5, 2002
Covers the outcry from site owners who feel Google has stumbled. However, anyone who thinks this is the first time Google may have gotten worse since 1998 is absolutely wrong. I can recall distinctly that Google had some relevancy drops that I perceived when it increased its index size substantially in 2000. These didn't last long and commonly happen when a crawler adds more pages.
Reaction To "Google Degraded? Geeks Aghast"
Search Engine Forums, Oct. 5, 2002
This is a good thread in that people have posted a variety of specific examples where they feel Google's relevancy has dropped.
When an engineer flaps his wings
Dive Into Mark, Oct. 3, 2002
Mark Pilgrim tries to summarize the Google changes and in particular states, "Google is now apparently cross-checking link text with the linked site, and discounting or ignoring links whose text does not appear in the linked site."
Sounds good, but I don't think the changes are as simple as that. For example, do a search for "pier one." The top site that comes up is for Pier 1 Imports, though the page itself doesn't say "pier one." Instead, almost certainly what helps is that some people are linking to Pier 1 but using the words "Pier One" in or near the anchor text.
Sadly, you can't prove this via Google, because it doesn't offer the advanced syntax that AltaVista does. At AltaVista, a search for:
link:pier1.com anchor:"pier one"
brings back 37 seven pages that link to the pier1.com site and use the exact phrase "pier one" in the hypertext anchor links. Wish I could do the same search on Google!
Microsoft no longer Google's No. 1 vision of 'hell'
Computerworld, Oct. 2, 2002
As part of the changes, Microsoft and other prominent sites lost their top rankings for the search "go to hell."
Why the search engine is not always right
Reuters, Sept. 26, 2002
Last month, a search for "go to hell" listed sites such as Microsoft, Disney and AOL in Google's top results. These Google Blips, as I call them, get discovered from time to time. Interestingly, Google clearly took action to fix this shortly after news started going around. Missed seeing it? No worries, Fantomaster has a screenshot: http://fantomaster.com/google-hell-search.html. For past blips, see http://searchenginewatch.com/sereport/01/02-bush.html, which tells you what happened to President George W. Bush, as well as Liv Tyler and how Microsoft in the past was rated "more evil than Satan" by Google.
Meet Your Favorite Search Engine Watch Contributors
Many of SEW's leading expert contributors will be at ClickZ Live, the new online and digital marketing event kicking off in New York (March 31-April 3). Hear from the likes of: Thom Craver, Josh Braaten, Lisa Barone, Simon Heseltine, Josh McCoy, Lisa Raehsler, Greg Jarboe, Dan Cristo, Joseph Kerschbaum, John Gagnon, Eric Enge and more!