User Tools

Site Tools


Filter Bubbles and Selective Exposure

date: December 18, 2012

Here's a question: if you and I share a connection on the web, and if each of our search results (or information discovery) are influenced by that connection, is our overall search experience better or worse because of that connection? And, what factors make the search or discovery experience better or worse? Before we quibble over how we define better or worse, let's assume that better implies a certain level of quality and of diverse content.

So, a friend of mine is interested in filter bubbles1), which is a negative term used to describe what may happen when search results become influenced by data reflecting one's personal social realities. The negative effect is that you continually retrieve content that simply reinforces your previous views.

Donald O. Case (2007) summarized literature about a similar issue called selective exposure. 2) 3) According to Case, this is where

humans tend to seek information that is congruent with their prior knowledge, beliefs, and opinions, and to avoid exposure to information that conflicts with those internal states (p. 97).

At first glance, it may seem that filter bubbles and selective exposure refer to the same thing. It seems to be true that the consequences (smaller, belief-reinforcing worlds) appear to be similar, and that the two may have an interactive and reinforcing effect on a person's belief system. However, while they may refer to similar outcomes, the difference is that filter bubbles are an information retrieval issue, and therefore an issue of relevance, and selective exposure is an information seeking issue, and therefore an issue of preference.

More specifically, filter bubbles, if they exist, are the result of search algorithms (or also a kind of online information rhetoric—how information is presented [this is why usability is so important]). Such systems relevance rank information based on the notion that what is relevant has something to do with our histories and our social realities, measured by our connections and actions on the web. Thus, if filter bubbles are a reality, we can assign blame to the search engines and the social media platforms that we use to interact with various types of information. On the other hand, with selective exposure the blame would go to the searcher because it is the searcher's preference for friendly information that leads to what the searcher finds relevant.

Selective exposure is a problem, but I am not entirely convinced there is a widespread problem with filter bubbles (as opposed to a less negative state based on personalized information retrieval in general). I will admit that there are potential issues or that there could be issues, but I believe personalized information retrieval is an important area of research and the practice of it has, at least for me, been very useful in my own information seeking experiences.

In any case, we might be able to make a strong case that better search engines or social media platforms are search engines or social media platforms that discourage filter bubbles if a certain level of information quality and a certain level of information diversity is a result of our interaction with these engines and platforms. However, whether they do so is largely out of our hands simply because we do not control the search algorithms. But we might be able to leverage (or hack) these algorithms and use them to our benefit in order to counteract any potential harm that filter bubbles may cause.

I think it may be possible to leverage social search engines and platforms by taking advantage of the tools that may be creating the problem in the first place. If our histories and social realities influence our search results or how we discover information generally, then it seems that if we form the intent to attend to our and others' histories and social realities (our connections and actions on the web), we can influence positive outcomes. With the filter bubble / selective exposure problem, the issue could be examined by creating an experiment where we look at users of social computing / media / search engine platforms by dividing them up into groups based on some measure of the qualities of their connections. Such that users with more diverse social connections, as measured by some instrument, may go into one group, users with less diverse social connections may go into a second group, and a third group could be used as a control group, where this group may simply be a random selection of users, where such randomness mimics the variation 4) that might be found among the general population of users and their social connections.

Why does the title of this post refer to librarians? I originally conceived of this post a while ago when Google began talking about Search Plus Your World and, at the time, I wondered how those who use a service like Google Plus might experience search results if they have a number of librarians in their circles (or Google Plus library pages; e.g., the New York Public Library) compared to those use use Google Plus who don't have librarians in their circles. It seems to me that having librarians (or information professionals of some kind) in one's circles could influence search results (when using Google) in a positive way (a testable proposition). (This could also apply to Bing and Facebook since the two of them together provide a similar functionality to that of Google and Google Plus.) If so, then it could be that librarians, academic or public, might have further justification to embed themselves in places other than the library. I particular like Michalek's (2012) description of the outward facing library and the role of the librarian in such an organization. 5)

As a final note, my dissertation is about what it means for academic libraries and librarians in a world where the storage of content (information) is decentralized and freely accessible. This post is simply another way to explore those implications. I believe this implication, and other variants, is at the heart of the intersection between library science and information science; in particular, it's what makes information science relevant to library science. As such, for those of us who primarily study and research topics in library science, it's part of the reason why we need to study and research topics in information science too (a recent experience has led me to reflect on this relationship a little more than I normally do).

Case, D. O. (2007). Looking for information: A survey of research on information seeking, needs, and behavior (2nd ed.). London: Emerald.

Michalak, S. C. (2012). This changes everything: Transforming the academic library. Journal of Library Administration, 52(5), 411-423. doi:10.1080/01930826.2012.700801

blog/filter-bubbles-and-selective-exposure.txt · Last modified: 2017/03/07 16:35 (external edit)