Reviewing Some Bad Google Search Results With Sergey Brin

After today’s Google search press briefing, where I raised the issue of some poor quality search results in Google at one point, Sergey Brin asked me to demonstrate a few. My pleasure! Below, what we reviewed and comments from Google’s cofounder.

First, I showed him the results for search engine on Google that I joked about yesterday at our SMX East search conference. Seriously, Dogpile ranks first? Followed by AltaVista, Ask and Lycos? Google’s the world’s most popular search engine, yet it doesn’t rate itself as the most relevant result, into number one?

He laughed a bit at the results and agreed they probably aren’t so good though countered reasonably that people who search for “search engine” on Google itself probably don’t expect to find Google. They’re already there, so putting it tops isn’t relevant.

I disagreed, a bit. We know plenty of people go to Google and search for “google,” so why not show them what’s commonly accepted as the best search engine for a search on search engine?

Sergey then said, unprompted, that Search Engine Land really should be showing up in the results for a search on “search engine.” Hey, I won’t argue with that (are you listening search quality team?). But I then noted that when searching for search engines (in the plural), you get a different set of results — more sites about search engines than search engines themselves. Well, at least that’s how it used to be. When we looked today, you still got back individual search engines.

The results illustrates one of the issues I raised during the press meeting, that much of Google’s algorithm seems to reward sites that have gained trust over time, regardless of how relevant they seem to a particular query.

Then I went on to an example I showed yesterday in a session, for california car rentals. I won’t name specific sites that surprised him, as they did me yesterday. The point is that in general, brands you’d expect to see like Hertz weren’t there.

Sergey pointed out that this contradicts my earlier point, that aged, trusted sites with authority seem to win in the search results. I agree, this is a case where “little” sites are winning — but how they’re winning is a mystery, but some certainly don’t feel among the most relevant results out there.

I also ran a search for car rental california where some big brands appeared but there was still a seemingly strange selection of results. On the second page of results, Google pointed to a listing from Expedia for car rentals in Orange County, California, that surprised me. I know it’s on the second page, but it’s content light and specific to one region, not all of California.

Don’t get me wrong. Google provides great results for many queries. It’s just that it’s very easy to find these type of exceptions, and I think many are wondering how do you move past links in a way to address better results (this came up in the press conference).

Finally, I talked to Sergey about often terrible experience with local results. He was well familiar with locksmith spam, but I pointed out other examples such the incorrect police phone number I found or reports of people getting wrong numbers for hospitals or any number of other weaknesses we’ve seen with Google Local listings that occupy the most prized results in Google’s search pages.

Some of this I covered in a keynote address (it’s short) at Where 2.0 that you can watch here:

[youtube]https://www.youtube.com/watch?v=ZiAGWS14VIM[/youtube]

Talking with Sergey today, I said that after reading how Google pays people $16 per hour to rate the relevancy of its results, how about paying some of those to review and improve the third party local information that Google depends on?

Sergey was thoughtful about it and said Google might move more that way, highlighting how today, the company has announced its own maps in the US.