Tony was giving a talk yesterday as part of a workshop with me and Grainne, to the OU Library and he said something I hadn't really appreciated before - namely that because Google refines its search results based on your history (if you are signed in to Google), the results, say, that Tony gets will be different from the ones you or I get. I know it came out in 2005, I told you I was slow on the uptake.
This made me think that your search history is actually valuable, because the results you get back are a product of the hours you have invested in previous searches and the subject expertise in utilising search terms. So, if you are an expert in let's say, Alaskan oil fields, and have been researching this area for years, then the Google results you get back for a search on possible new oil fields will be much more valuable than the results anyone else would get.
There are a couple of interesting implications to this. Firstly, just as people who don't have the time will pay others to build up their characters for them in virtual worlds such as World of Warcraft, so the time you have invested in your search history may become a valuable commodity. Time is the commodity we don't have enough of, and anything that can only be realised through the investment of time has some value to someone.
Secondly, if you can assemble and utilise the expert search of a network of people, then you can create a socially powered search which is very relevant for learners. Want to know about really niche debates in evolution? We've utilised Richard Dawkins, Steve Jones and Matt Ridley's search history to give you the best results. Or if you prefer, the search is performed as the aggregation of a specialist community.
You can see how this would play in a SocialLearn type context - the benefit of belonging to the network is that you gain access to a collective search refinement. And in a Google World search almost equals knowledge. Of course, this would require some means of you opening up your search history to be used by others, in a way that a) didn't reveal all of your searches (if you're an eminent historian you don't really want everyone to know the time you searched for 'Who is Paris Hilton?') and b) didn't corrupt your search data, in the same way as when you buy gifts for someone in Amazon and your recommendations get screwed. (Because I buy books for my parents I get recommended a range of London in the Blitz books - who knew there were so many?).
Anyone know if such a thing is in operation anywhere?
So to protect our interests, should we obfuscate by throwing lots of additional searches for things we're not really interested in? Tottenham Hotspur, caravans, Spongebob Squarepants?
Posted by: AJ Cann | 09/10/2008 at 09:39 AM
I did wonder about that - you'd probably need multiple search identities, which might get confusing. Even worse if other family members use your computer - mine would also be mixed up with 'sponsor a pony', Ben 10, and Australian holidays.
Posted by: Martin | 09/10/2008 at 09:46 AM
Aren't the search results that the expert actually follows up and bookmarks more powerful? Every academic should be publishing the RSS feeds for their social bookmarks, classified by key terms. The user can choose to filter these according to the social rating of the URL and aggregate results from a group of experts according to their reputation in their field and their online expertise in finding valuable sources.
Posted by: Laura | 09/10/2008 at 10:24 AM
It's reasonable to assume that many patents have been applied for on this. Microsoft Research have published some of their early work on leveraging the search trails of others (See eg http://portal.acm.org/citation.cfm?id=1377490)
Posted by: Paul Lefrere | 09/10/2008 at 12:26 PM
Not sure this is exactly what you are looking for, but a while back I played with a front end to Google that built a tag cloud of all the search terms that went through it. It was inspired by the http://www.gnosh.org/ site; the idea being that for a department or workteam, it was useful to get the kind of "ambient" knowledge of what your co-workers were looking for, without any real additional knowledge.
Posted by: Scott Leslie | 09/10/2008 at 05:45 PM
...that last line should read "without any real additional *effort*." Doh!
Posted by: Scott Leslie | 10/10/2008 at 12:03 AM
"but a while back I played with a front end to Google that built a tag cloud of all the search terms that went through it."
I keep thinking that intranet sites should expose this sort of thing for searches carried out on them?
Posted by: Tony Hirst | 10/10/2008 at 08:33 AM