The Tagging Hall of Shame: Amazon

by Jennifer Golbeck

Readers are likely to say that I am against a lot of the new web-movements, what with my post about XFN and this one.


Let me reiterate that I think tagging is a good thing, and that when done properly it can be very powerful. On the other hand, the fact that it's the Hot New Thing on the web seems to be leading some sites in to implementing tagging without thinking through why or what purpose it serves.



I was doing some Christmas shopping on Amazon.com recently when I noticed their new tagging feature. Toward the top of the page, you can add your own tags to items. This is useful - the ability to annotate an item, and review those annotations later can help users in a lot of ways. For that, Amazon gets credit. I also hold this principle to be true generally (be it on Amazon, in XFN, on Flickr or del.icio.us, etc.)- creating your own personal tags of anything can be helpful in remembering and personally organizing information.


Of course, the buzz around tagging is not because we can create tags to help ourselves, but that we can create tags that will help others. This is where I think a lot of applications go astray. Going back to Amazon, they have decided to add a section to the page to show everyone's tags. Let's look at a few examples:

Freakonomics, by my former professor Steven Levitt - this is the #4 best selling book on Amazon at the time I write this. Thus, if something is likely to be tagged, this is toward the top of the list - lots of people look at it. Yet, here are the tags displayed from users (with the number of times each tag was used in parentheses after each term):



Good Book (1), konomics (1), linda or dad (1), ok (1), freakonomics (1), revealing (1), contrarianism (1), best (1), Madison (1), Nick (1), Dad (1), Creativity Booster (1), business wish list (1), Math Stimulator (1), debunking (1)




I don't think a reasonable person could argue that these are helpful to users. Interesting, perhaps, but questionably useful.

However, perhaps it is that one item. In thinking about a book that was very popular and had been out for a longer time, I went to The Da Vinci Code, by Dan Brown. With all the controversy and hype, certainly this was going to attract taggers. The results, however, are similar (note: spelling errors are from users, not me):



For Adam (1), dan (1), must read (1), Fiction (1), Dan Brown next novel (1), Mysteries and Thrillers (1), fiction (1), Mary and Jesus Books (1), Novels (1), critcal thinking (1), bestseller (1), novel (1), The Solomon Key and Beyond (1), Faux-Historical (1), The Solomon Key (1)


In addition, these pages are cluttered with further useless information by providing a list of people who tagged the book. These people are not connected to their tags, but rather listed in a separate section. Even if you thought someone was creating intelligent and insightful tags, there is no way to know which person that is.



I think this is a perfect example of tagging gone wrong. Amazon is not offering any real benefit to users by showing them all this data. They are just muddling up the pages. Yes, the feature is new and with more data, common tags may begin to be repeated, but if these tags are examples of what to expect, there is not a lot of hope. Seeing that 15 tagged The Da Vinci Code as a "bestseller" and "fiction" is still not going to help me.



Furthermore, Amazon doesn't need this. They provide a very insightful list of phrases similar to what we would want from tags in the "Inside This Book" section. For Freakonomics, they list the first sentence and the following terms (as "Capitalized Phrases (CAPs)"):


University of Chicago, Supreme Court, Stetson Kennedy, Paul Feldman, Los Angeles, Head Start, Black Disciples, James Alan Fox, National Bureau of Economic Research, Sudhir Venkatesh, Quarterly Journal of Economics, Adam Smith, American Economic Review, Peter Sandman, Black Gangster Disciple Nation, Chicago Public School, Jake Williams, Journal of Political Economy, Johnny Appleseed of Crack, Most Popular White Girl Names, Oscar Danilo Blandon, Arne Duncan, The Impact of Legalized Abortion, Harvard University Press, William Julius Wilson



Each of those phrases are linked to a page that shows the number of references in this book and references in other books. If you have read the book, you know that these are interesting and useful glimpses into the text.



As I've said, I do believe tagging can be useful (and my next post will discuss how). However, sometimes a community solution like shared tags doesn't offer anything but confusion. Unfortunately, Amazon's shared tagging falls into this latter group.