Technology as Enabler of
Fake News and a Potential
Tool to Combat it
[Link]. Žiga Turk
[Link]@[Link]
Research commissioned by Policy Department A on
the request of IMCO Committee
Content
news circulation process has changed
• in the past it was edited by professionals
• now it is edited by everyone
“fake news” circulates largely due to social
networks
• 30-40%
technically it is possible to interfere with
circulation of “fake news”
• but cannot be stopped entirely
conclusion
• yes we can but should we
The news process
Create news
Quality control of news
Publication of news
Amplification of news
Consumption of news
The news process
schematically
Traditional news process
control control
create consume
edit publish amplify
create consume
reasonable, responsible, liable publisher
create consume
professionals editorial state
consume
policy censorship
consume
self
censorship
consume
Enters the Internet
anyone can
publish
anyone can
amplify
anyone can edit
editors are gone
no trashcan
source: wikipedia
Internet-age news process
? ?
amateurs and
professionals
create consume
create publish consume
create consume
edit amplify
consume
consume
self
censorship consume
Technology for “fake
news” dissemination
Social media is main driver for “fake
news”
10% of readers of top news come via social media
40% of readers of “fake news” come via social media
Fake and real news have about the
same engagement on social media
this is 10% for
real news
but 40% for
“fake news”
In Europe “real” beats “fake” 6:1,
in US it is 1:1
6 1
1 1
“Fake news” (orange) audience is
10% of real news (green)
For „fake news“ main source is
Facebook
On Facebook users see 10% of what
they „subscribe” to
Facebook decides
which 10%
Facebook chose to
remove news from
feed
• hurting 10% of real
news traffic
• hurting 40% of fake
news traffic
Both unhappy!
How they choose the 10% is
key issue for stopping “fake news”!
Facebook’s
goal: „keep
audience at site
and engaged“
Algorithm is
opaque
Algorithm is
changing
Algorithm is
unfair?
source: The Western Journal
Bots spread fake news on Twitter.
And real news.
“We find evidence that
social bots play
a disproportionate role in
spreading and repeating
misinformation.” (source).
“Contrary to conventional
wisdom, robots
accelerated the spread of
true and false news at the
same rate, implying that
false news spreads more
than the truth because
humans, not robots, are
more likely to spread it”
(source)
Propagation of fake news: Nodes and links represent Twitter accounts and
retweets of the claim, respectively. Node size indicates account influence,
measured by the number of times an account is retweeted. Node color
represents bot score, from blue (likely human) to red (likely bot).
Can we stop
“fake news”?
Yes we can!
with author
with publisher
with platform
with amplifier
with reader
Stopping authors to publish “fake
news”
demonetize
• deny
advertising at
fake news sites
threaten prison
• e.g. Polish anti-
holocaust law
lock them up
• e.g. Turkey
Stopping “fake news” when published
deny hosting
deny listing in
DNS directory
block internet
traffic to “fake
news” site
Stopping fake news at platforms
public platforms
• Facebook, Google, Twitter, Baidu …
• label, augment, hide or remove “fake news”
private messaging platforms
• Viber, WeChat, Snapchat
• listen to private conversations?!
needs collaboration of platforms
Label „fake news“ on Facebook
Label “fake news” on Google
Label “fake news” on Twitter browser
Hide “fake news” by downranking –
pushing down on search results
Augment “fake news” with links to
real news, warn before sharing
Remove “fake news” - deplatform
Stop bots!
Stopping “fake news” in private chat,
email – the next frontier
What’s Up
Facebook Chat
Snapchat
Gmail
Signal
Telegram
Viber
can be done, but!?
Key element: how to
identify “fake news”?
Fake news can be identified by
humans, machines or combination
Human
• human work by fact checkers
• fact-checking industry appearing, looking for
business model
Human-machine teamwork
• machine identifies potentially fake stories
• double-checked by human
Machine only
• artificial intelligence
Two strategies for automatic
identification of “fake news”
From content
• facts
• style
From social
• stance
• propagation
Shu, Kai, Amy Sliva, Suhang Wang, Jiliang Tang, and Huan Liu.
"Fake news detection on social media: A data mining
perspective." ACM SIGKDD Explorations Newsletter 19, no. 1
(2017): 22-36.
Content analysis is hard because …
… needs to
understand content
• few news pieces are
blatantly false
style?
• authors will adapt
Artificial Intelligence can help.
But “Computer did it” is a fallacy
New source is an easy identifier of
“fake news”
what site is the
origin of news
shortlist of “fake
news” sites
Social context tells a lot about
trustworthiness of news
“influencers”
propagation
patterns
deployment
of bots
need for
speed
Propagation of Fake climate change story
Stop at consumption
Media literate user is the last and
best line of defense
people
remember and
believe “fake
news” about as
much as
placebo news
(non existent
news)
“Available
evidence
suggests that
for now the
influence of
fake news is
limited”.
Hunt Allcott & Matthew Gentzkow, 2017. "Social Media and Fake News in the 2016 Election," Journal of
Economic Perspectives, vol 31(2), pages 211-236.
Conclusions
Yes we can. But may we?
gatekeepers are gone
• only guerrilla war on fake news is possible
technology to disturb circulation of fake news
exists
• most efficient if platforms cooperate
• some approaches are clearly interfering with free
speech
users seem to be quite capable at identifying
fake news
• to early to blame election results on fake news
The End
For more information:
[Link]@[Link]
@zigaTurkEU