Do Search Engines Have A Bias In Showing Their Own Sites?

Jan 20, 2011 • 8:58 am | comments (11) by twitter Google+ | Filed Under Google Search Engine
 

Ben Edelman released a paper named Measuring Bias in "Organic" Web Search. In short, he felt "each search engine favors its own services in that each search engine links to its own services more often than other search engines do so." Specifically, "Google's algorithmic search results link to Google's own services more than three times as often as other search engines link to Google's services."

Read the report, then come back here.

Now Danny Sullivan, arguably the most respected writer in the search space disagreed. His story, named Google “Favors” Itself Only 19% Of The Time felt Google does not have any ranking bias towards their own results. Now that you read Edelman's story, go read Danny's and then come back here.

What do you think? Do you think Google unfairly pins their own services higher with a bias?

Before I state my opinion, take my poll below.

So Matt Cutts obviously thinks otherwise and he tweeted "Why would something called "Google Translate" rank higher for [translate] than a product called "Babel Fish"? Surely, SE bias?!? But: no."

I asked Matt, "@mattcutts what bothers you more, stuff like that or spam?"

This is where he replied, "@rustybrick I don't mind dubious claims when people are just uninformed. But these folks should know better. :("

Personally, I don't think Google is specifically moving their results up in the organic free listings. The algorithm looks at lots of factors where Google tends to do well in by the nature of Google's popularity. But I personally do not think Google has a ranking bias toward their results.

Forum discussion at Sphinn & Sphinn.

Previous story: People Search Google For Dialing The Police. Why Not 911?
 

Comments:

Sebastian

01/20/2011 02:09 pm

IMHO the whole bogus "search neutrality" fuss is about satisfied users vs. MS, their lobbyists, and large scale scrapers. What a waste of energy, valuable time, and billable hours: http://arstechnica.com/tech-policy/news/2011/01/regulating-googles-results-law-prof-calls-search-neutrality-incoherent.ars

Barry Adams

01/20/2011 02:51 pm

So either a Harvard study led by a veteran researcher is inaccurate and/or misleading, or a Google fanboy (with an incentive to protect his buddy-status with Google) and a Google employee are defensive of Google's practices. Hmm.... Personally I found Danny Sullivan's retort against the Edelman study to be rather weak. He fails to poke any significant holes in it and instead tries to attack the study on its scientific merits - which is an area I somehow don't think Sullivan has a lot of experience with - and on Edelman's credibility - again, something Sullivan should really keep his mouth shut about. Of course search engines give preference to their own properties. That's just common business sense. Edelman has now provided us with solid proof for what the search industry has always known for years. Well, except for Danny, obviously.

ian_art

01/20/2011 03:06 pm

What about this story when Google was caught hard-coding some of the results with their properties tied on top? Did the whole thing die out or was it a hoax?

Michael Martinez

01/20/2011 05:01 pm

Matt and Danny have to explain what it is when Google shows its own data aggregation above the search results or its own videos above links to other services' videos in the search results if that is not "bias". Of course all the search engines are biasing their results to their own properties. They have an inside edge on crawling and indexing those properties. The issue should not be defined in terms of whether there is bias, but rather in terms of whether there is intentional market manipulation away from relevant competitors.

Matt Cutts

01/20/2011 09:47 pm

"either a Harvard study led by a veteran researcher is inaccurate" You neglected to mention that the veteran researcher has sued Google, been a paid consultant for Microsoft, and is a member of an organization called the "Alliance Against BAIT & CLICK" whose members have also sued Google.

Barry Adams

01/20/2011 10:54 pm

Aye, full disclosure and everything, it's good to know those things and understand the context and source. Which still makes Edelman more trustworthy than you and Danny, as he still has an academic reputation to keep intact whilst you protect naught but corporate interests. Also, speaking of Microsoft, you might want to read this latest piece by Aaron Wall about the discrepancy between how Microsoft handles instances of bad behaviour versus how Google handles them: http://www.seobook.com/breaking It's about trust, Matt. Microsoft is building it, and Google is losing it.

Thom Craver

01/21/2011 12:17 am

I agree with Matt and commented on this. The Harvard researcher's sample was statistically flawed. It was too small and biased as it contained one-word terms, over a third of which were actual Google product names. Since Google's algorithm attempts to rank brands and products, it makes sense. Several long-tail search terms do not show favor to Google products in the slightest. I went into some detail, giving examples and blowing up the professor's facts in my blog rant. (The Unfairness of Google's Ranking Fairness).

dannysullivan

01/21/2011 12:19 am

It was Edelman misunderstanding that a search for a stock symbol like CSCO is not the same thing as search for CSCO, -- he thought adding a common was somehow "breaking" the "hard coding" that caused a Google OneBox display to show you a little chart of the stock prices. Tossing in the extra character changes it from being a stock quote to something else, so the OneBox doesn't trigger. Google ought to fix that, because for whatever reason someone might add that extra character, they probably do mean the stock quote and would find the OneBox results useful. The whole thing is similar to how if you do a search for CSCO at Yahoo, you get a stock chart that shows at the top of the results for that -- with the chart powered by Yahoo Finance (similar to how Google's chart is powered Google Finance). Unlike Yahoo, however, Google also provides links to alternative services. Yahoo doesn't do that. All you get is Yahoo. Same thing at Bing, where you get a Bing Finance chart but no links to Yahoo Finance much less Google Finance. I've known Ben for years and corresponded with him regularly on his studies. He almost always asks for my opinion before he publishes something, I assume because he respects my opinion in regards to search engines. He's cited me in several of his reports as an authority. We don't always agree, of course. I've particularly liked and appreciated his work documenting spyware abuse or how Google indirectly funds some pretty unsavory characters. I just wrote about that on Search Engine Land this week, and Ben thanked me for mentioning his work when I was calling out Google for some of this (most of the attention instead had been directed at that whole Bing affiliate thing, as if Bing was the only one with unsavory traffic partners. Ben and I disagree on this bias aspect. That sort of thing happens. I don't think he has the fundamentals of his study correct, which means trying to even go into the statistical part makes no sense. But I explained my reasoning in my story, for those who care to read it. Some will agree; some won't.

ian_art

01/21/2011 08:14 am

Thanks for clarification Danny. I think Ben's study lacks keywords related to the money properties like AdWords, if there's bias it should be there (and from what I just checked there doesn't seem to be any). While studies like this are certainly needed to make sure there's a healthy ground for competition, I don't think Google needs any bias at this point. With the power their domain have accumulated over the years, they can easily rank for pretty much anything with a few internal links.

Jon_cramer

01/21/2011 03:54 pm

Hell yeas they are bias especially Goolge...very evil

Barry Adams

01/25/2011 03:52 pm

I have been informed since by several of my peers with relevant backgrounds and whom I trust in all things SEO that the study is apparently indeed flawed and its results not to be trusted. So I owe you and especially Danny Sullivan an apology. Sorry, I was too quick to respond and my judgement was incomplete and thus thoroughly flawed.

blog comments powered by Disqus