Skip to main content

Assignment - 4 (Echo Chambers, Filter Bubbles, Opinion Mining and Opinion Manipulation)


Please propose an alternative design for Google search that addresses problems posed in this paper. By design I mean here just a short description of user interactions, information filtering and presentation of results.

Google search, in its current form, does pose the issue of "Filter Bubbles" and as the paper states, could potentially be performing "global lobotomy". But at the same time, providing an alternative design for Google search is a monumental challenge in itself, as it needs to take into consideration various challenges, including but not limited to the following criteria:

TRUSTWORTHINESS OF SOURCE:- "Trusted Source/Vetted Media Source" Badge - Much like how Facebook & Twitter have currently adopted "Verified Account" tags for original and authentic business/public personality pages, even Google should adopt a Policy of awarding "Vetted Media" badges based on background verification to websites which have original and verifiable factual content, and not fake news. This badge could be based on recognition from international media organisations such as AP.

TAG BASED APPLICATION OF PERSONALIZATION:
- Personalization can not be entirely revoked - Google may not survive as a successful organisation if personalization is revoked, since as the paper describes, Google is almost an online extension of "System 1" type thought process among human beings. But extending personalization to news would not be better UX, but potentially disastrous to journalistic ethos as well as implications of removal of journalism from democratic processes. No journalism is almost the same as untrustworthy journalism, which is what personalized news and information search on Google is. Hence, in this aspect, I would opine that perhaps topical search results which are under "News" tag should not have personalization applied to them, but rather have results shown from trustworthy websites.

PROFITABILITY OF CONTENT & CLICKS:
- Economic Sustainability & Justification to Investors (by Google) - Profitability can not be entirely taken out from Google's search algorithm as it is, at the end of the day, a private profit oriented entity, and not a charitable organisation. Having said that, there are realms of Google such as academic content search results, news items, inappropriate/illegal content (country specific, and issue specific - such as violent material, child pornography, sale of weapons/drugs etc.), and personal data of individuals, which should not be privy of commercialization, assuming they have inviolable reasons to not be monetised - such as Human Rights Concerns, Illegality of Content, and/or Ethical Compromise upon commercialization. Although, profitability of search items could be selectively categorised on topical basis, and those deemed compromising the integrity of content due to conflict of interests ought to be omitted out of profitability which is derived from "user being the product". 

CHANGE OF UI & SEARCH RESULTS:
- Having a News Spread with multiple sources displayed alongside each other - While personalized information could be the primary source displayed in search result, if in case, Google can enable a feature within the display of search results to display items based on user relevance, for user to unavoidably look at, before moving forward, a small way of addressing "System 2" manner of rational decision making could be encouraged and  initiated within the user. 


MORE POWER TO USERS:
- A Simple informative widget which tells users what's going on outside the "filter bubble"  - A simple widget could be embedded along with the previous display grid which describes what percentage of viewers viewed which articles displayed in relevance grid. And perhaps a small description of user persona of each user "on hover" could add a little benefit to the user to understand better how (un)biased, (ir)relevant, (un)popular, (un)accepted, and (un)informed their news results and information that they consume.


This suggestion is explained with an illustration below, for greater clarity:
Fig.1 - Representational Image of Search Results.

Considering that the same problems apply to Facebook, is it possible to apply your design to Facebook too?

Although it could be possible to use the same design to Facebook, it may not be the best solution for Facebook, as before we design for Facebook, we need to take into consideration the reasons why people use a Social Network, over a Search Engine. Search Engines are more like Information Directories, and hence, the user expectations here is to get relevant or informative results, while Social Networks are usually places, designed for users to gather and connect with like-minded people, and discuss ideas/opinions, and not necessarily,  a journalistic platform.
Keeping this in consideration, Facebook does not have any onus to build on democratic theory or journalistic sanctity of information.

Facebook newsfeed algorithm redesign should be considered based on multiple factors which overlap at times with Google search algorithms, but not necessarily coincide.
Ideas which could be retained from the Google redesign suggestion include the ideas I proposed under the following titles:TRUSTWORTHINESS OF SOURCE:- "Trusted Source/Vetted Media Source" Badge 
A slight modification here could be that, whenever a news article or link is shared, Facebook should develop its own Vetted Media Badge, whose delegation and awarding could be a function of a recognised international Journalistic organisation like Associated Press, which should award Journalistic recognition to media outlets.

MORE POWER TO USERS:
- A Simple informative widget which tells users what's going on outside the "filter bubble" - same as shown with the Google illustration (Fig.1).

CREATION OF A SEPARATE 'NEWS' PROPERTY:- Facebook should delineate its Commercial or Social Pages, and Groups from the News pages, as function and implication of News content is different from Regular Pages and Group Posts -But more importantly, Facebook needs to create a new category of Properties on its platform which stand distinctly from Pages, Groups, and Profiles, called News. And a prerequisite to have a News presence, should be to have the Badge.



Blogs Are Echo Chambers: Blogs Are Echo Chambers
Eric Gilbert, Tony Bergstrom and Karrie Karahalios

How agreement and disagreement were measured in this paper?
The researchers used Machine Learning and NLP techniques to measure agreement and disagreement in this paper. They built a predictive model using machine learning algorithms, based on a custom text analysis code written in Java and Perl, and utilised the Weka toolkit.

They categorised comments in three sets: agree, neither, disagree. The context of agreement was with respect to the blog author. That is to say, the researchers measured commentator's agreement or disagreement of comments in relation to the author's views within the blog post.

When there were multiple comments or opinions between author and commentator, the agreement was assessed based "on balance".

The researchers kept in mind to sift between and omit the (dis)agreement with other comments, and focused only on the the (dis)agreement with the blog post and the author.

This approach was applied by two researchers to independently rate the selected sample of comments, then their rating were compared.


From your personal experience, give some examples of online communities, where there is very little disagreement between opinions (or disagreement is not tolerated).

Facebook Groups, Facebook Opinion Pages, Political Quora Threads, Whatsapp Groups - all tend to become echo chambers at varying points of time. 

Facebook Groups, Whatsapp Groups, and Special Opinion Pages on Facebook are usually echo chambers of like-minded individuals, in my experience.I've observed that Political Quora Threads can be used to downvote an unpopular opinion, facilitating removal of dissent by virtue of number of dissenters.

Another important set of Online Communities which have similar echo chamber issues are openly affiliated Media entities with political/ideological viewpoints like Breitbart, Salon, or Fox which attract singular set of audience, as well as opinions, and exist everywhere.


What features of Facebook wall (or Facebook in general), promote development of echo chambers?

Curated Newsfeed, Content from Friends who one interacts with the most, Curated Content from Pages we Like and Pages like the Pages we like, Facebook Curated "Trends", Groups, Indexed hastag Trends, Curated Advertisements targeted at users based on our Preferences, Online Personas, Likes and Dislikes. Almost everything on Facebook, except the content we see from friends who we interact with for different reasons, and notice to hold different opinions, facilitate creation of Echo Chambers.
-----------------------
The material under discussion could be found here.

Comments

  1. Avar Pentel14 Nov 2018, 15:00:00

    Good points about design, especially the widget which tells users what is going on outside the bubble.

    Avar Pentel

    ReplyDelete

Post a Comment

Popular posts from this blog

Assignment Blog Post on Lifestyle, Mobility and Location.

This Blog Post is in response to the Assignment #3 which can be found  here  or  here .  The Livehoods Project: Utilizing Social Media to Understand the Dynamics of a City 1.       What patterns are used by the authors to define a  livehood  cluster? Authors define Livehood clusters based on “… the spatial proximity between venues as given by their geographic coordinates, as well as the social proximity” derived “… from the distribution of people that check-in to them.” The authors use Foursquare check-ins by users in the targeted area to analyse and define Livehood clusters. The authors used only the Foursquare check-ins that were publicly disclosed by users on Twitter to collect the data, use Cluster mapping to define Clusters of user behavior which determined the borders of Livehoods within city municipal borders. For each check-in, authors documented the user ID, the time, the latitude and longitude, the name of the venue, and the category of the place. The
Post a Comment
Read more

Assignment-1

Find below the answers for the questions asked in the Assignment.1 : Link to the assignment is here . PAPER ONE:  Tie strength in question answer on social network sites How do the authors define strong and weak ties, and how did they measure the strength of a tie in Social Network Sites? Strong ties and weak ties are usually connections that an individual have with people around them, online/offline. The tie strength represents the depth of friendship between individuals. Whether they are close friends or acquaintances. Measurement of tie strength was based on the following variables (which were derived from the study by Gilbert and Karahalios): - Days since last communication - Days since first communication - Words exchanged - Mean tie strength of mutual friends - Positive emotion words - Intimacy words What could be the reasons that more helpful answers come from weak ties? What could be the reasons that more helpful answers come from strong ties? Weak Ties: - Different Soc
1 comment
Read more

Assignment 5 - Data Analysis Reading

DATA ANALYSIS READING ASSIGNMENT The paper studied for the following responses could be found here  and here . §   What were the assumptions of the model and the data analysis the authors used? Assumptions of the model were that the social network adoption spread very similarly to the way diseases spread amongst a finite population. They also assumed that there would be a gradual disuse of social networks as users and their peers begin to lose interest. Hence, it reflected the gradual reduction in number of members, much similar to the cured number of people in epidemiological progression. Authors also assume that web traffic is the general metric to measure the data of social network usage. §   Find three arguments that could call the authors’ conclusions into question. Three arguments I believe that bring the authors’ conclusions into question are: 1) Users shall gradually get disinterested in a social networking platform. – This assumption implies that the inter
Post a Comment
Read more