Facebook Study Says Users Control What They See https://research.fb.com/exposure-to- diverse-information-on-facebook-2/
Facebook Study Says
Users Control What
They See
https://research.fb.com/exposure-to-
diverse-information-on-facebook-2/
EXPOSURE TO IDEOLOGICALLY DIVERSE
NEWS AND OPINION ON FACEBOOK
- Eytan Bakshy , Solomon Messing , Lada Adamic
Presenter: Harshitha Chidananda
CONTENT
• Introduction
• Data
• Process
• Exposure
• Alignment
• Results
• Limitations
• Conclusion
DO YOU CONTROL WHAT YOU SEE ON
FACEBOOK?
INTRODUCTION
• Exposure to news, opinion and civic information through social media
• How do these online networks influence exposure to perspectives that cut across ideological lines?
DATA
• De-identified data
• 10.1 Million accounts
• Active US users
• Self-reported their ideological affiliation
• 7 million weblinks shared over 6-month period
• July 7, 2014 and January 7, 2015
• - Their interaction with socially shared news
• Information Abundance
• While these technologies have the potential to expose individuals to more diverse viewpoints, they also have the
potential to limit exposure to attitude-changing information
• Who they friend and what content they click on are more consequential than the News Feed ranking in terms of
how much diverse content they encounter
FACEBOOK’S CREATION
Echo Chambers
Individuals are exposed only to information from
like-minded individual
FACEBOOK’S CREATION
Echo Chambers
Individuals are exposed only to information from
like-minded individual
Filter Bubbles
Content is selected by algorithms based on a
viewer’s previous behaviors
PROCESS
• Compare ideological diversity of the broad set of
news and opinion shared on Facebook with that
shared by individuals friends network
• Compare this to the subset of stories that appear in
individuals algorithmically-ranked News Feed
• Observe what information individuals choose to
consume, given exposure on News Feed
PROCESS
• Compare ideological diversity of the broad set of
news and opinion shared on Facebook with that
shared by individuals friends network
• Compare this to the subset of stories that appear in
individuals algorithmically-ranked News Feed
• Observe what information individuals choose to
consume, given exposure on News Feed
PROCESS
• Compare ideological diversity of the broad set of
news and opinion shared on Facebook with that
shared by individuals friends network
• Compare this to the subset of stories that appear in
individuals algorithmically-ranked News Feed
• Observe what information individuals choose to
consume, given exposure on News Feed
• Stories classification
• Hard (National news, politics, world affairs)
• Soft ( Sports, entertainment, travel)
• Training a SVM on text features
• 13% Hard content
EXPOSURE
• 226 Thousand distinct hard content URLs
• Shared by at least 20 users
• 3.8 Billion unique potential exposures:
• Cases in which an individual’s friend shared hard content, regardless
of whether it appeared in her News Feed
• 903 million unique exposures
• cases in which a link to the content appears on screen in an
individual’s News Feed
• 59 Million unique clicks
3.8 Billion unique potential
exposures
EXPOSURE
• 226 Thousand distinct hard content URLs
• Shared by at least 20 users
• 3.8 Billion unique potential exposures:
• Cases in which an individual’s friend shared hard content, regardless
of whether it appeared in her News Feed
• 903 million unique exposures
• cases in which a link to the content appears on screen in an
individual’s News Feed
• 59 Million unique clicks
903 million
unique
exposures
EXPOSURE
• 226 Thousand distinct hard content URLs
• Shared by at least 20 users
• 3.8 Billion unique potential exposures:
• Cases in which an individual’s friend shared hard content, regardless
of whether it appeared in her News Feed
• 903 million unique exposures
• cases in which a link to the content appears on screen in an
individual’s News Feed
• 59 million unique clicks
59
million
unique
clicks
ALIGNMENT
• We then obtain a measure of content alignment(A) for each hard story by averaging the ideological affiliation of each user who shared the article.
ALIGNMENT
• We then obtain a measure of content alignment(A) for each hard story by averaging the ideological affiliation of each user who shared the article.
As = +0.80 As = -0.65
“On average, 23 percent of people’s friends claim an opposing political ideology.
Half of users have between 9 and 33
percent of friends from opposing
ideologies
25 percent have less than 9 percent
Remaining 25 percent have more than 33
percent
Among friendships with individuals who
report their ideological affiliation in their
profile
Media proportion of friendships liberals
maintain with conservatives is 0.20,
interquartile range (IQR) [0.09, 0.36].
Similarly, the median proportion of
friendships that conservatives maintain with
liberals is 0.18, IQR [0.09, 0.30]
WHAT APPEARS TO USERS?
Facebook Feed
What their friends share
News Feed ranking algorithm
What individuals choose to read
Facebook feed order
Frequency of visit to Facebook
How much they interact with certain friends?
How often they have clicked on links to certain website in News Feed in the past
Of the hard news content that people’s friends share, 29.5 percent of it cuts across ideological
lines.
When it comes to what people see in the News Feed, 28.5 percent of the hard news encountered
cuts across ideological lines, on average.
24.9 percent of the hard news content that people actually clicked on was cross-cutting.”
How much cross-cutting content
individuals encounter depends on
who their friends are and what
information those friends share.
Liberals tend to be connected to
fewer friends who share information
from the other side, compared to
their conservative counterparts
After algorithmic ranking,
conservatives see 5% less cross-
cutting content compared to what
friends share
Liberals see 8% less ideologically
diverse content
54 percent—more than half—clicked on ideologically cross-cutting content, although less
than the 87 percent who clicked on ideologically aligned content.
LIMITATIONS
Study on facebook
Facebook’s users tend to be younger, more educated, and female, compared to the U.S. population as a whole
Blogs and Twitter, have been shown to exhibit different patterns
Distinction between exposure and consumption is imperfect
Individuals may read the summaries of articles that appear in the News Feed, and therefore be exposed to some
of the articles’ content without clicking through
“FACEBOOK FEED IS DIVERSE”
RETHINKING INFORMATION DIVERSITY IN NETWORK
Information in social media may come from weak
ties as well
Suggests that social media may actually increase the
spread of novel information and diverse viewpoints
CONCLUSION
Facebook Study Says
Users Control What
They See
https://research.fb.com/exposure-to-
diverse-information-on-facebook-2/
THANK YOUQUESTIONS?
We classified stories as either “hard” (e.g., national news, politics,
world affairs), or “soft” content (e.g., sports, entertainment, travel)
by training a support vector machine on unigram, bigram, and
trigram text features