A couple of years ago I wrote about our need for an “opposing view reader” that would intentionally show us commentary from people with opinions different from our own. As far as I can tell relatively little progress has been made on this front. Twitter, for instance, will generally recommend to me that I follow people who are like the people I am already following. So I was happy to see that there is at least some academic research on this topic. The Technology Review mentions a paper titled “Data Portraits: Connecting People of Opposing Views” that is available on arxiv. Unfortunately when I actually read the paper I found that n for their study was tiny (37!) and I don’t think supports the conclusions with any significance (their stats notwithstanding).

In my original post I failed to explain fully why this problem has been growing in importance. One of the fundamental forces of the Internet has been the unbundling of content. If you are reading this right now you are getting my opinion from my blog or in your Tumblr dashboard (or via a feed reader). The post stands by itself. Whatever I link to was chosen by me. Put differently, there is no context that could provide for an opposing view, no additional editorial or purposeful pairing by an editor. 

When you combine unbundling with a huge increase in available content, you wind up with a situation where someone could spend all their available time just reading content that will confirm their existing views. Existing recommendation and search algorithms are all built to re-enforce this by trying to present you with content (or people) that match your interests. That leads to what Eli Pariser has called the “filter bubble” – his TED talk on the subject is worth watching (or you can read his book).

I also should have given credit to my then fellow PhD student and now BU Professor Marshall van Alstyne who wrote an excellent paper on the subject as far back as 1996, titled “Global Village or Cyber Balkans?” (co-authored with Erik Brynjolffson). Here is a great quote from the introduction:

Information technology can link geographically separated people and help them locate interesting or compatible resources. Although these attributes have the potential to bridge gaps and unite communities, they also have the potential to fragment interaction and divide groups by leading people to spend more time on special interests and by screening out less preferred contact.

The paper itself is quite theoretical but I am pretty sure that today we could easily use Twitter data to examine the hypothesis.

If you know of a recommendation service that does in fact surface opposite views please let me know. I am also interested in additional research on the topic. In the meantime I am thankful for folks like David Pinsen whom I follow on Twitter and who routinely helps challenge my thinking.