OCLC and squeaky wheels
Had quite a shock today while teaching class. I was reenacting the “little big” search experiment I described in a post last week, and much to my surprise, had wildly different results this time around. If you search for this book on worldcat.org now, up it pops in first position. While it is gratifying to see OCLC react so quickly to feedback, I was curious whether this represented a change in their search algorithm, or rather just a quick fix for this one title to appease me.
Sadly, I suspect the latter. Doing some testing, I discovered that when you search for the term little or big, Crowley’s book now comes up in first position. Not sure how his work merits top billing for either of those terms, not when there are plenty of other books with two word titles where little or big is one of them, or one word titles that are even better matches. It is as if a database administrator got my feedback passed to them and said, fine, we’ll give you what you want and attached a high weight to this work using whatever mechanism they have for doing so. The underlying search algorithm is still a bit odd.
For example, searching for the word little brings up Crowley’s book in the first slot, the book Little Eagle in second position, but then items 3-10 are books written by someone named Little. So, two there by title keyword, eight by author. Furthermore, results 3-10 are all for German titles. Sure, I am sitting at a German IP address, but I selected English as the interface language, so why use an identifier over which I have little control rather than a deliberate user choice to rank my results? Incidentally, when I visit amazon.com from this IP address, it recognizes my IP address and adds a little note to the entry page (in German) encouraging me to visit amazon.de, but if I insist on searching at amazon.com, it returns results as if I were in the US. Amazon lets me choose how I want my results fed to me. I see no such option with worldcat.org, other than the language facet on the left, but I want my initial search to be spot on, not to have to tinker endlessly to find what I want. Librarians do that. Any usability test will reveal that an average user does not. We all know that.
Even when I log in to worldcat.org, I see no option to set my preferred locale for searching. And now that I think about it, I think it is kind of weird to change the ranking based on IP address. I did, after all, enter an English language query, so why weight German books high because I am at a German IP address. What a mess.
At any rate, Little, Big is now at the top of the list because somehow it has now been given a high weight, not because any underlying problem has been fixed. Searching for similar short, pithy titles (Kerouac’s Big Sur, Wilder’s Our Town, etc.) shows that they still land buried beneath works that are less exact matches.
Then again, I have no idea what people are seeing who might be sitting at a French, American, Russian, or Japanese IP address. Changing the ranking algorithm based on IP without giving the user a clear warning that this is occurring nor the ability to shut it off is just not right. Frankly, it renders worldcat.org useless to me when I am sitting at a German computer. Doing neat things behind the scenes with queries (aka post-query processing) is something that Roy Tennant and others have been saying for years needs to be a priority in libraries. Worldcat.org has evidently taken up the challenge, but this is not a good nor consistent implementation. Where I am does not define who I am or what I want.