TED Talk: Beware Online Filter Bubbles

Eli Pariser | TED2011 Beware online "filter bubbles"

Internet and tech companies are responsible for the filtering and personalisation of web results, determining how online users discover and act on information. There are many advantages that can be taken from web content personalisation. Relevance, speediness, and customisation can be included among them. However, it can be argued that the personalisation of web content and search results significantly restricts users’ online experience. Hopefully, algorithms will be developed with a sense of civic responsibility, as stated by Eli Pariser, allowing web users to have control of global access to unbiased information.

Download the entire lesson below in A4 PDF format:

TED Talk_ Beware Online Filter Bubbles

Warmer questions

  1. Internet companies are curating and personalising our web content experience. But what about the content we are not seeing?
  2. Editors have traditionally been tasked with curating the world of information. How does the world around us change when we get curated −or personalised− content, automatically for us?
  3. Can some level of personalisation be useful to our online experience?
  4. What are online users missing that they need to see?

Reading section

Mirror mirror on the wall

Tech companies are now fashioning an Internet that is no longer worldwide but is entirely tailored. They have developed algorithms that track online habits generating online experiences that are totally personal. Machine learning algorithms are used in almost every platform to predict online users’ intentions based on what the platform has learned from behavioural and historical data. In this particular, the term “filter bubble” makes reference to filtering, or personalising search results and news feeds to reflect the online users’ preference. Personalised search thus refers to web search experiences that are custom-made specifically according to an individual’s interests.

There are two general methods of personalising search results; one includes modifying the user’s query and the other is about re-ranking search results. Relevance is the focus of Facebook’s personalisation algorithm, for instance. The slightest characteristics of your online life are gathered to create a complete indication of an ideal Facebook experience. Facebook’s News Feed delivers customised content that “most interests” individual users. Google is doing this, too. When searching for information, Google displays results according to its relevance algorithm. Personalisation app Google Now pursues to “give you the information you need throughout your day before you even ask”. Amazon’s recommendation engine uses personal data tracking along with other users’ browsing habits to recommend relevant products.

Although this personalisation clearly saves a lot of time, detractors argue that the problem is not with the information the users are getting, but what is being left out. Customised content is exposing online users to the whole Internet through their own interests, habits, and experiences using information that individuals are often not entirely conscious of. Their scope of experience is thus being narrowed to what is believed the users want to be experiencing. The “filter bubble” theory explains this concern thoroughly. Proposed by Eli Pariser, an internet activist, this theory proposes that personalisation can affect online users’ experiences. Instead of being in contact with universal and diverse content, users are algorithmically exposed to material that matches their pre-existing viewpoints. Examples of popular filters include:

  • Google’s Personalised Search Results
  • Netflix “Popular Queue and other Personalised queues”
  • Twitter “No Replies” setting for brands
  • Twitter “Top Tweets” in Search
  • Facebook “Top Stories” News Feed Ranking
  • iTunes “Top” Apps, Books, Songs, Movies, etc.

For users, relevance means less time spent finding the content they will enjoy. But personalisation isn’t just for websites, it has made an impact on mobile apps, too. The Essential Wine App from Delectable Wines, for example, allows users to take a picture of their favourite wine, the app will remember the users’ choices and recommend related wines based on their tastes.

Vocabulary matching

Match the vocab on the left with the correct definitions on the right.

Vocabulary Definitions
1. Feed a. A slight change in position, direction, or tendency.
2. Relevance b. Make or adapt for a particular purpose or person.
3. Shift c. Convert (information or an instruction) into a particular form.
4. Tailor d. Supply (someone) with (information, ideas, etc.).
5. Algorithm e. A person or thing that controls access to something, usually information or access.
6. Bubble f. (Of a microprocessor) designed and built as an integral part of a system or device.
7. Gatekeepers g. Used to refer to a good or fortunate situation that is isolated from reality or unlikely to last.
8. Embedded h. A method or set of rules to be followed in calculations or other problem-solving operations, especially by a computer.
9. Encode i. The quality or state of being closely connected or appropriate.
10. Personalisation j. Having custom or tailored information from online networks sent to you.

  1. d
  2. i
  3. a
  4. b
  5. h
  6. g
  7. e
  8. f
  9. c
  10. j

Vocabulary Gap Fill

Using the words in the previous exercise fill in the gaps below.

  1. You should ___________ your spending to your income.
  2. The most rudimentary ___________ repeats a single instruction.
  3. Economists are concerned that the housing ____________ is going to burst and signal the beginning of a recession.
  4. What he said has no direct ­­­­­­­­___________ to the matter in hand.
  5. The two parties _____________ confidential data in a form that is not directly readable by the other party.
  6. His name lies ___________ in the minds of millions of people.
  7. Social media companies are able to swfitly _____________ public opinion on poltical figures.
  8. Law schools are the ___________ of the profession.
  9. You are not supposed to ___________ bears in the wild
  10. The internet has allowed people to have access to _______________ such that they may never encounter an opinion different from their own.

  1. tailor
  2. algorithm
  3. bubble
  4. relevance
  5. encode
  6. embedded
  7. shift
  8. gatekeepers
  9. feed
  10. personalisation

TED Talk: Beware online filter bubbles

As web companies strive to tailor their services (including news and search results) to our personal tastes, there’s a dangerous unintended consequence: We get trapped in a “filter bubble” and don’t get exposed to information that could challenge or broaden our worldview. Eli Pariser argues powerfully that this will ultimately prove to be bad for us and bad for democracy.

Watch the video above and then answer the questions below.

  1. What did the Internet mean to the speaker when he was growing up in Maine?
  2. What is the speaker’s political tendency?
  3. What happened to her Facebook’s conservative feed?
  4. How did this happen?
  5. Did Facebook consult him before editing that feed out?
  6. Is Facebook the only site doing this algorithmic editing of the Web?
  7. How is Google tailoring its queries?
  8. What other sites are personalising the feed?
  9. What is the Internet showing us?
  10. What is a filter bubble?
1. It meant a connection to the world. It meant something that would connect people together.
2. He is progressive, politically.
3. The conservatives had disappeared from his Facebook feed.
4. Facebook was looking at which links he clicked on, and it was noticing that, actually, he was clicking more on his liberal friends’ links than on his conservative friends’ links.
5. No, it did not. Without consulting him about it, it had edited them out. They disappeared.
6. No, it is not. Google’s doing it too.
7. There are 57 signals that Google looks at −everything from what kind of computer the person is on to what kind of browser the individual is using to where the person is located− that it uses to personally tailor the query results. So there is no standard Google anymore.
8. Yahoo News, the biggest news site on the Internet, is now personalised − different people get different things. Huffington Post, the Washington Post, the New York Times are all flirting with personalisation in various ways.
9. The Internet is showing us what it thinks we want to see, but not necessarily what we need to see.
10. If you take all of these filters together, you take all these algorithms, you get what the speaker calls a filter bubble.

The advantages of a personalised web

  1. Personalised web content engages users much more effectively compared to generic “one size fits all” content. From a company’s point of view, personalisation increases the time users spend on a website and higher engagement levels are established by delivering personalised web searches.
  2. Internet companies identify what users like and give them more of it, saving online users a lot of time and exposing them to information that they would miss otherwise.
  3. Companies see personalisation as a means of getting online users what they want faster. From an Internet company’s perspective, online users get results they are the most interested in by using a lot of data, such as past searches, device usage, location, etc.
  4. A recommendation engine will give users fast and accurate access to similar content. It is intended to helping users find interesting content that will include audio, video, text or infographics they might otherwise have missed.

The disadvantages of a personalised web

  1. Personalised searches restrict the users’ possibility of search. This personalisation forces them to operate within certain limitations by omitting other links outside the users’ circle.
  2. Personalised search leads to a compromise in the quality of the search results. Non-inclusion of data in your personalised search would automatically keep a substantial proportion of the user’s information out of the scope of the search.
  3. Personalised search limits your options, by restricting the scope of the users’ searches, significantly.
  4. The concerns on personalised content come mostly from users, where they raise privacy issues as a problem of web Personalisation. They feel they are being tracked, followed and watched to be sold something.
  5. The filter bubble has increasingly made online users suspicious of how companies and web marketers are handling their data

Extended discussion questions

  1. Do personalisation algorithms risk narrowing our minds?
  2. Is personalisation the process while filter bubbles are the result?
  3. Internet privacy: Are your searches private or just personalised?
  4. Are relevance algorithms narrowing horizons and keeping people less informed?
  5. How can online businesses use filter bubbles to create meaningful connections between their companies and online consumers?
  6. Internet companies are curating and personalising our web content experience. But what about the content we are not seeing?
  7. Editors have traditionally been tasked with curating the world of information. How does the world around us change when we get curated −or personalised− content, automatically for us?
  8. Can some level of personalisation be useful to our online experience?
  9. What are online users missing that they need to see?

Potential debating topics

  1. Personalised web content saves individuals valuable time and is of great benefit to online users.
  2. Users get personalised search results based on what sites they visit. Cookie tracking blatantly violates online users’ privacy.
  3. Online individuals get “interest-based ads” based on the websites they visit. Web users are consequently being flooded with unwanted information and bombarded by annoying advertisements.
  4. Internet was meant to be global. Local search results based on IP address narrows online users’ scope of access.
  5. Personalisation algorithms should be integrated into software that would offer web and app users the possibility of overriding the recommendations made by the algorithms.
  6. Algorithms cannot really forecast humans’ unpredictable behaviour just by tracking the user’s clicking. Personalisation algorithms have only a partial understanding of online users’ interests and preferences.
  7. The content we encounter online seems to repeat the same things over and over again.
  8. Personalisation algorithms influence what users have chosen yesterday, what they choose today and what they will be choosing tomorrow. The system can understand web users only on its own terms.