It seems like there’s been a lot of publicity about Netflix recently. About how effective their recommendation engine is … about how they’re leveraging their understanding of consumer tastes not only to distribute other people’s content more effectively, but to create their own original content aimed at specific markets, like House of Cards.
In fact, the title of this post is paraphrased from an article that ran in the Sunday Washington Post, Feb, 23, entitled, “Watch It, HBO.” The article is actually a version of an article that ran a few days earlier in Slate — you can read it .
What strikes me above all in this article is the power of data analytics. Netflix, along with Amazon and other companies, is using what it knows about us, and about our aggregated behavior, to get smarter and smarter about serving us.
Meanwhile, there are libraries that throw away all their user behavior data every night — in the form of circulation records — in the interests of protecting user privacy. When they do that, they throw away the opportunity to use that data to improve their services. I think that’s a mistake.
I’m not advocating that librarianship abandon its commitment to protecting individual privacy. But letting privacy concerns cut us off from implementing modern, more effective services is a primitive, ostrich-like response to the challenge. Wouldn’t a more mature approach be to recognize that we need to offer the benefits that come with retaining and analyzing user data — but at the same time we must manage the risks?
If we don’t, what will happen as people become more used to recommender systems, and the systems get better and better — and libraries are stuck in the mode of purely passive 1980s-era approaches to service?
There are several ways I can imagine that scenario playing out — none of them favorable to libraries. So, I hope that the leaders of libraries that still throw out their user data will rethink this, and develop sophisticated programs that address both the risks and rewards of retaining and analyzing user behavior data.
The Embedded Librarians Café
February 27, 2014 at 10:31 am |
Recommendation engines are great as they help people find new things that they’ll likely enjoy. Libraries should consider applying some data analytics tactics to their services.
However, I also believe that librarians should play an important role in expanding people’s horizons by introducing them to data, information, and knowledge that they wouldn’t otherwise come across. I’m not sure that recommendation engines — as they generally are applied in widely known arenas like Netflix and Amazon — would do that important task.
March 3, 2014 at 9:10 pm |
Steve, what a great idea! I agree, we’re not in the mass entertainment business, like Amazon and Netflix. We’re in the “creation of knowledge in our communities” business, to paraphrase David Lankes. It would be fun to work on a project to build a tool that could support this.
March 4, 2014 at 4:53 pm |
David,
You’re making me miss grad school! This would be an excellent project.
Having said that, one doesn’t have to be a grad student…
Steve
March 4, 2014 at 5:14 pm |
David,
Put more simply: What would an “anti recommendation engine” look like?
Perhaps “anti” is the wrong term. There would need to be some criteria that would exclude items — for example — that one is already interested in, reaffirm one’s worldview, are beyond one’s cognitive ability to comprehend (ie, introducing Calculus topics to a toddler or with some aging could grasp such topics), and so on…
Steve