Skip to main content

University Library, University of Illinois at Urbana-Champaign

Ways of the Web: Filter Bubbles and the Deep Web: Home

This guide is a companion to the Ways of the Web Savvy Researcher Workshop.

Introduction

Most of us use common search engines like Google several times a day, yet if we knew that the results were somehow skewed, we might take more time to evaluate our results and rethink how we search.

This guide will introduce you to the concept of the "Filter Bubble" - one phenomenon suggesting results may be skewed. It will present differing viewpoints about Filter Bubbles, and suggest ways to "burst" your filter bubble, as well as ways to get around it. So what is a filter bubble? Read below to find out.

What is a filter bubble?

"Filter bubble" refers to a phenomenon that occurs with many of the websites that we use: algorithms (mathematical equations) use our search history and personal information to tailor results to us. So the exact same search, using exactly the same search words, can return different results for different individuals. This is called personalization.

Eli Pariser believes this is a problem, as it can affect us and the way that we think. Our beliefs and values are confirmed rather than challenged. Certain kinds of information do not even reach us. Above is a link to the catalog record for Eli Pariser's book.

Eli Pariser's TED Talk

Eli Pariser wrote the book The Filter Bubble (2011 - pictured to the left), and he gave a TED talk based on that book. Here you can view the TED talk in its entirety, with subtitles.

Who does this?

These are just a few of the websites that tailor results to you and your clicking history:

Google Amazon Washington Post
Netflix Yahoo News New York Times
Facebook Huffington Post

A Filter Bubble Demonstration - Try this at home!

One way to see how filter bubbles work with search engines that do personalization (like Google) is to take a word that can have multiple meanings in different contexts and build up different search histories using those contexts. Then, when you search for the same word after having built up different search histories, the search engine should return results that look a bit different.

For this demonstration to work, you need to be sure to clear your search history before you start each round. This works even better if you have 2 or 3 people working side by side at different computers. That way you can compare the results more easily.

Try this with the word Tea.

1. Have someone build a search history using names of countries where tea is popular or names of countries where teas orgininated. Remember, do not use the word "tea" as a search term quite yet. Examples would be England, Japan, China, Latin America, etc.

2. Have another person build a search history using different spices, herbs, and flowers that make up common teas. Examples would be roses, cinammon, chrysanthemum, lavender, etc.

3. Have a third person search for anything related to politics, such as names of political parties (not the Tea Party just yet, though!), names of political movements, words like "activism," or "conservative" and "liberal."

4. When you are performing these searches, click on some of the results (preferably general ones that might somehow later be connected to tea!). This will contribute to your search history.

5. Finally, have everyone search for the word "Tea." Have fun comparing results!

Note: Your results may still look very similar; the differences may be subtle. Whether or not the filter bubble is really something to be concerned about will be discussed in the next tab.

Ask a Librarian

University of Illinois Library's picture
University of Illinois Library
Contact:
Main Library
1408 West Gregory Drive
Urbana, IL 61801
217-333-2290
Website / Blog Page