Sunday, November 11, 2012

Crowd Sourcing or Crowd Surfing

Crowd sourcing is typically supposed to tap into the wisdom of the crowd.  In the case of many library initiated crowd sourcing ideas it seems that crowd sourcing is intended to take the pulse of your local community.  In his 2009 post, Libraries and crowdsourcing -- 6 examples, Aaron Tay lists some common uses of crowdsourcing for libraries:

  1. Digitisation of newspapers and correction of OCR errors (National Library of Australia)
  2. Tagging of photos on Flickr (Library of Congress)
  3. Rating,tagging and contribution of user reviews of books (via Endeca, Encore, Aquabrowser, Koha and other next-generation OPACS etc)
  4. Frequently asked questions (Social reference tools like LibAnswers)
  5. Comments and suggestions (GetSatisfaction , UserVoice etc)
  6. Collaborative Cataloguing (Library Thing, Biblios.net and Open Library)
Half of these uses (uses 2, 3, and 6) are basically folksonomies (a kind of group tagging popularized by the site delicious.com -- formerly known as del.icio.us).  There are definitely advantages to this kind of cataloging.  Massive numbers of resources can be tagged in far less time than it would take a diligent S.W.A.T. team of librarians to do it (in the case of the Internet, assigning your library cataloging task force is almost akin to giving them life sentences).  And so things are allowed to take on a kind of organization that wouldn't be possible otherwise.  But the trouble with folksonomies is that they aren't thorough or consistent.  Casual taggers aren't following Ranganathan's faceted colon classification system; they're developing something much more complex and infinitely less searchable.  They create subjective cataloging.  While Ranganathan might praise the way in which this system divests hierarchy, he'd probably be pretty dismayed when he went to search Flickr for pictures of Russian Blues and found that no one had used that tag.  And in fact that there was no consistency to the language or spelling of the tags that had been used: "cute cats," "kittens," "qt kitty," "meows," or whatever other variations you can imagine.  It's nice that things are getting classified, but it's not useful.

                   Here's your Russian blue S.R.


Another type of crowd sourcing Tay talks about is the use of comments. Allowing comments and suggestions in your OPAC can mimic Amazon reviews -- a popular element of that website.  They give voice to your community and make it feel like a conversation has started.  But there are major disadvantages to this as well.  One only need look at a handful of poorly designed products on Amazon to discover that sometimes manufacturers, authors, and publishers post their own misleading reviews.  Furthermore, these kinds of reviews are typically only written when the customer is very satisfied or very disappointed (just look at a site like Ratemyprofessors.com to see an example of the way in which including only the most effusive and most vitriolic reviews can skew one's perception of an instructor); they don't actually benefit from the aggregation present in typical crowd sourcing -- like the Google rating algorithm -- because they aren't drawing groups that are actually representative of your constituency.  Finally, Amazon reviews are useful for one reason: they help buyer's beware.  Libraries have less need of this.  Our patrons aren't spending hard earned cash.  They can take more chances.  If the book they pick up at the library doesn't catch their interest after the first 50 pages, they can bring it back and pick up another at no cost.  Granted, sensitive patrons might care to know if a book is particularly gruesome or sexually explicit, and these reviews could provide that kind of information, but, for me at least, libraries are about discovery and about being exposed to new ideas.

There are two major problems with the types of crowd sourcing Tay discusses.  First, as Derek Powazek suggests in his article The Wisdom of Community, the best crowd sourcing is done by reducing the task to its simplest parts.  He warns

Conversational inputs are too complex for Wisdom of Crowds systems. Online discussion systems do not lead to wisdom on their own.
And he goes on to explain exactly why:

One of the reasons discussions do not lead to wise results is that there’s no aggregation—the conversation just happens. But WOC systems are there to produce a result. This requires an aggregator (like you) and an algorithm.
The benefits of crowd sourcing, he argues are only visible after aggregation.  Folksonomies and comments resist aggregation.  Even a site like rottentomatoes.com, which attempts to aggregate subjective movie reviews is only helpful to a point.  Skyfall, Lincoln, and A Royal Affair (all movies which came out this week) received a 91%, 91%, and 90% respectively.  But that doesn't suggest that people who enjoy the new James Bond movie will also like the romantic costume drama of A Royal Affair.  These kinds of qualities resist being made objective.

A further problem with crowd sourcing is the possibility that it can become crowd surfing.  Kristina Grifantini, in her article Can You Trust the Crowd Wisdom?, cites evidence from Vassilis Kostakos from the University of Madeira in Portugal that suggests that small numbers of users can distort the overall pictures.  While you may think a whole crowd is working on your projects, it may in fact be a handful of web savvy patrons.  If you put a lot of stock in the crowd sourced material, you may unwittingly be raising the profile of just one or two web savvy patrons.  They now surf above the surging masses (your other patrons) like a kid at a Pearl Jam concert in 1996 (or two young girls at Relient K.  You know, whatever you can find on youtube.)

    No comments:

    Post a Comment