His theory is that Google is too easily gamed by SEO (search engine optimizers) and sites that rely less on algorithms to rank content – Mahalo, for example, uses hand edited search results – will provide more relevant results to searchers.
He also sees an opportunity for tapping into the social gestures of trusted web users, such as how Techmeme sorts the tech news of the day based on the buzz among top tech bloggers.
He sees the social gestures – in the form of links – from top bloggers as a great way to sort search content.
Scoble raises some interesting points, but seems to fail to understand how Google and other search engines rank search results today.
Google, along with every other major search engine, are already looking at social gestures to sort search results. Link popularity has been a major factor in determining the trust of a given web page for as long as Google has existed. They quantify social gestures – in the form of links – to measure trust, or as they call it, PageRank.
Not all links are created equal. A link, for example, from a trusted source like Robert Scoble is worth more than a link from a brand new website since the new site has built up no trust of its own.
Another bone Scoble seems to pick with Google is their search results mix. By this, I mean what types of links one will see in the first page of search results for a given search. He seems to think that Google blindly serves ten results with no further analysis other than link popularity. That’s wrong.
A couple years ago, Google’s search results for retail searches were starting to degrade. The biggest problem was a lack of diversity in results, where the top-10 would be dominated by the same type of results. For example, a search for a Canon SD 800 IS Camera would bring back results from tons of comparison shopping sites, like Bizrate, Shopping.com, Pricegrabber, etc.
Checking that result today, I see results from Canon, Review sites, Retail sites, and blogs reviewing the camera. This was done without human editing by a Mahalo editor, and without relying on a handful of elite tech bloggers.
As far as I can tell, the thing that impresses Scoble most about Mahalo is the search refinement provided in search results for relatively generic terms. For example, a search for the term “HDTV” provides a diverse set of results, including links to popular manufacturers.
This seems to overlook the fact that other search engines attack search refinement as well. For example, Yahoo provides refinement suggestions below the search box:
I notice they don’t include “Manufacturers” as an option here, which leads me to believe that it’s not a particularly common refinement.
Ask.com goes further with their refinement suggestions:
As I think about it, it makes sense that “Reviews” and “Best” would be great refinements for the term “HDTV” since people are trying to figure out which ones to buy. How does a list of manufactures of HDTVs really help with this process? I just want to know which ones are best and why.
Google clearly has a good feel for what terms a person may use to refine a broad term like HDTV. Take a look at their suggestions offered in the Google search box on Firefox:
Okay, here’s the point on search refinement: If people often refine to the same term, why wouldn’t Google just promote results from the refined results page to the results page of the broader search term? Why not put more reviews on the main page if that’s what people prove to be REALLY looking or once they refine their results?
What’s to stop them from saying, “Billions of queries suggest that this is the type of results people are looking for when they conduct a product related search,” and roll out similar results formats for millions of product pages overnight?
Search refinements are another form of social gestures that can be used to improve relevancy, and nobody has access to more of those gestures than Google.