The Risks of Personalized Search

October 28, 2009

Most online users are lousy searchers. These individuals don’t agree. In my research, In a study at a major trade association in July 2009, my team learned that most of the individuals in our sample perceived their search skills as excellent or superior. The perception that search is a slam dunk is widespread. The reality, of course, is that most people rely on public Web search systems. When it comes to locating data for a business decision, the actual work processes observed reveal inefficient and largely manual work methods. I was thinking about this interesting situation when I read about Google’s new real time and social search system. If you are unfamiliar with this “to be” service, navigate to the Google blog and read “Introducing Google Social Search.”

I think that the service can be quite useful for some types of research. On the other hand, I see three risks associated with this type of service:

First, users who think they are great searchers will rely on “recipes” (systems and methods) to deliver information tailored to their needs as determined by software agents. In short, the clueless will have no clue what their query has been interpreted to mean. The clueless searcher will get results and assume that their search expertise has returned exactly what the user wanted. Wrong. The smart system returns what the algorithms determine the user wanted. Perception is one thing; reality is another.

Second, searchers with confidence in their search skills often overlook the reality that sources may or may not be correct. Not even Google’s smart software can deliver phone numbers that match an organization’s current phone number. Even more egregious errors exist within any result set. Users who lack basic search skills are likely to accept information in a results list at face value. Users lack a foundation in determining what is reliable and what is not.

Third, training wheels in smart systems—whether in’s slick new product displays or Google’s semiautonomous agents—foster users’ confidence in their search skills. Users don’t know what they don’t know, and the false sense of confidence can get some people in hot water. I recall reading about a person following a Google Map into a pond. The information from the search system took precedence over the fact that the maps were wrong.

i don’t know how to fix this mismatch between a user’s search expertise and the user’s perception of his or her search expertise. Schools are not doing much to educate users about online. In the meantime, the trade press happily ignores these and other risks of personalized search systems that shift the burden of knowing how to find accurate information from the user to software. Risky in my opinion.

Stephen Arnold, October 28, 2009

No pay, no way.


2 Responses to “The Risks of Personalized Search”

  1. » Pandia Search Engine News Halloween Wrap-up on October 31st, 2009 9:21 am

    […] The Risks of Personalized Social Search […]

  2. ????? ???????????????????? ?????? | ?????? on November 1st, 2009 6:20 am

    […] //Beyond Search […]

  • Archives

  • Recent Posts

  • Meta