More Related Content Similar to Trust and Information Disorders - a Dispute of Narratives (20) Trust and Information Disorders - a Dispute of Narratives3. Information disorders (Wardle)
• Misinformation –
• “Information whose inaccuracy is unintentional, and spread
unknowingly.” (Wardle, Donovan, …)
• “cases in which people’s beliefs about factual matters are not supported
by clear evidence and expert opinion (consensus)” (Nyhan & Reifler)
• “beliefs about politics that are inconsistent with the best available
evidence” (Garret et al)
• Disinformation – “Information that is deliberately false or
misleading, often spread for political gain or profit, or to
discredit a target individual, group, movement, or political
party.” (Wardle, Donovan)
© Daniel Schwabe 2021
Emily K. Vraga & Leticia Bode (2020) Defining Misinformation and Understanding its Bounded
Nature: Using Expertise and Evidence for Describing Misinformation, Political Communication,
37:1, 136-144
4. Information disorders++
• Conspiracy theories – “attempts to explain the ultimate causes of
significant social and political events and circumstances with claims of
secret plots by two or more powerful actors” (Douglas et al)
• Propaganda
• “the spreading of information whether it be true or false, good or bad - literally
‘spreading the faith’” (Vernon McKenzie)
• “the attempt to influence the public opinions of an audience through the
transmission of ideas and values” (Richard Taylor)
• “Propaganda is the deliberate and systematic attempt to shape perceptions,
manipulate cognitions, and direct behavior to achieve a response that furthers the
desired intent of the propagandist.” (Jowett & O’Donnel)
• Gaslighting - “where one agent seeks to gain control over another by
undermining the other’s conception of herself as an independent locus
of judgment and deliberation” (Spear)
© Daniel Schwabe 2021
Sinha, G. Alex, Lies, Gaslighting and Propaganda (March 7, 2020). 68 Buffalo Law Review 1037 (2020),
SSRN: https://ssrn.com/abstract=3550591
All these phenomena implicitly assume there is an absolute “Truth”
5. Shades of misinformation (Vraga+Bode)
© Daniel Schwabe 2021
Emily K. Vraga & Leticia Bode (2020) Defining Misinformation and Understanding its Bounded Nature: Using Expertise and
Evidence for Describing Misinformation, Political Communication, 37:1, 136-144, DOI: 10.1080/10584609.2020.1716500
6. Pitfalls in consuming information
• Preference for higher personal value over evidence
• Confirmation bias
• Cognitive Dissonance minimization strategy preference
• Pre-disposition towards conspiracy theories
• Inappropriate heuristics
• Motivated reasoning x “Cognitive Economy”/insufficient
analytical thinking
• Misperception about existing consensus
© Daniel Schwabe 2021
7. Consuming information
• It’s only relevant when connected to some action – now or in
the future
© Daniel Schwabe 2021
Claim A
Agent
Trusted claims
KG (“Facts”)
Claim A
X
Claim B
supports
Claim N
supports
***
8. Anchoring the Trust Lattice
© Daniel Schwabe 2021
Claim A
Agent
Claim B
supports
Claim N
supports
***
Claim B1
supports
Claim BN
supports
*** Claim B1
supports
Claim BN
supports
***
Beliefs!
9. Beliefs
• Linked to brain functions
• Own experience/observation (but fallible!)
• Selective attention - “Gorilla experiment”
• Deepfakes
• External authority
• Social norms
• E.g., birth registry determine age
• Religious norms
• Moral norms
© Daniel Schwabe 2021
Angel, H.-F., & Seitz, R. J. (2016). Process of believing as fundamental brain function: The
concept of credition. SFU Research Bulletin, 3, 1–20.
http://journals.sfu.ac.at/index.php/sfufb/article/view/91/99
http://www.theinvisiblegorilla.com/videos.html
10. Countering Information Disorders
• What is the desired effect?
• Replacing Mental Models
• Corrections
• How – direct rebuttal, alternative related facts
• How much – frequency (once, repeatedly), extent (detailed x
simple)
• When – before, simultaneously, after
• Effectiveness in changing “beliefs”?
• Side-effects
• Backfire? Overkill, Personal Value Dissonance, Familiarity
• Accuracy perception of non-tagged information
• “Spillover Effect”
© Daniel Schwabe 2021
11. Fact check x “Political Check”
• There are fewer objective facts
• Fact checks also express opinions
© Daniel Schwabe 2021
https://newrepublic.com/amp/article/156039/political-fact-checkers-
distort-%20truth
https://newrepublic.com/amp/article/156039/political-fact-checkers-distort-
%20truth
https://www.politifact.com/factchecks/2018/jun/
07/blog-posting/it-illegal-shower-and-do-laundry-
same-day-
californ/#annotations:0Hb9kmcBEeuuHder69vXsQ
13. Issues in countering information disorders
• It’s not about truth!
• A dispute of narratives
© Daniel Schwabe 2021
14. Trust Process
© Daniel Schwabe 2021
Data/Information
Metada
Context
Policies
Trust Process Trusted Data/Information
Action
Agent
Daniel Schwabe. 2019. Trust and Privacy in Knowledge Graphs. In Companion Proceedings
of WWW '19. ACM, New York, NY, USA, 722–728.
DOI:https://doi.org/10.1145/3308560.3317705
15. What is trust (in information)?
• For our purposes, trust = belief
• Trust as a social process.
• Some properties of trust
• It only matters when some action is involved (Castelfranchi’s delegation)
• It is held towards information provided by some agent regarding some
“matter”.
• It is binary! You can’t “half trust”, because you can’t “half act”
• Models of Trust x Trust Models (trustworthiness)
© Daniel Schwabe 2021 15
❝ Trust is knowledge-based reliance on received information❞. (Gerck)
16. All three studies find statistically significant evidence that the “Great Replacement” –
the idea that minorities will have more rights than whites – is a key driver.
Report on Capito Hill Insurrectionists - U. of Chicago CPOST - https://cpost.uchicago.edu/research/domestic_extremism/
Values - Example
© Daniel Schwabe 2021
“Storming the US Capitol was an act of collective political violence, inspired by a
leader, President Trump, and not merely vandalism or trespassing for other
purposes.”
17. CONSISTENT FACTOR ACROSS STUDIES
• STUDY 1: Odds of sending an insurrectionist is six times higher in counties where %
non-Hispanic whites declined.
• STUDY 2: CPOST/NORC Survey: Among Americans, believing that blacks and
Hispanics are overtaking Whites increases odds of being in the insurrectionist
movement three-fold.
• STUDY 3: CPOST/MTurk → Among conservative Americans, fear that blacks and
Hispanics will have more rights than whites increases odds of being in the
insurrectionist movement two-fold.
Report on Capito Hill Insurrectionists - U. of Chicago CPOST - https://cpost.uchicago.edu/research/domestic_extremism/
Values - Example
18. Donald Trump tweet
Whites have more
rights than Blacks
or Hispanics
claims
author
Massive
Electoral
Fraud
2020 Election
Results
claims author
Joe Biden is 46th
US President
Joe Biden will
implement
Democratic Values
contradicts
US Congress
Whites, Blacks
and Hispanics
have equal rights
imply
Narrative B
Joe Biden is a
Member of the
Democratic Party
Democratic
Party
Joe Biden's
Registration
author
imply
contradicts
Narrative A
tweet
author
Make
America
Great Again claims
imply
claims
© Daniel Schwabe 2021
19. Conspiracy Theory - Example
© Daniel Schwabe 2021
https://www.reuters.com/article/uk-factcheck-israel-temperature/fact-check-video-shows-
scanners-checking-temperature-not-vaccination-mark-or-microchip-
idUSKCN2AU29I?mc_cid=4f35f35d5d&mc_eid=8b93f52094
20. People will
be checked
for the Mark
of the Beast
John
Revelation
13:15-17
author claims
Vaccines are
a Mark of
the Beast
Person X FB Post
author claims
People are
being
scanned for
vaccinations
Reuters Video
author claims
imply
People will be
scanned for
temperature in
public events
Government
Regulation
author
claims
People are
attending
public event
Reuters
Video
author
claims
People are
being scanned
for temperature
imply
contradicts
Narrative B
Narrative A
© Daniel Schwabe 2021
21. Implications for Knowledge Graphs
• If KGs provide “ground truth”, how do we control quality?
• KGs must represent “points of view” explicitly!
• Simple node and link is too low level – Nanopublications?
• Provenance
• Wikidata data model? – No authorship information (as data)!
• Attribution
• Fine-grained argumentation relations
• Human values?
© Daniel Schwabe 2021