Realtime user ratings as a strategy for combatting misinformation: an experimental study

Area of Science:

• Social Sciences
• Computer Science
• Information Science

Background:

• Traditional fact-checking is slow, often lagging behind viral misinformation.
• Harnessing collective intelligence via crowd veracity assessments is a proposed solution.
• The goal is to limit the spread of false information by user-generated ratings.

Purpose of the Study:

• To investigate the effectiveness of crowd-based veracity assessments in detecting misinformation.
• To examine how community structure (ideological segregation vs. mixing) impacts the accuracy of crowd ratings.
• To understand the conditions under which crowd intelligence for misinformation detection succeeds or fails.

Main Methods:

• Experiment involving 4000 participants across 80 bipartisan online communities.
• Sequential rating of informational messages by community members.
• Analysis of how displayed prior ratings influence subsequent users' veracity judgments.

Main Results:

• In well-mixed communities, public display of veracity ratings improved correct classification of true and false messages.
• In ideologically segregated communities, crowd intelligence backfired for false information.
• Ideological bias of early raters in segregated communities skewed subsequent assessments away from accuracy.

Conclusions:

• Crowd-based misinformation detection is effective in ideologically diverse online communities.
• Network segregation presents a significant challenge, potentially amplifying misinformation.
• Future community misinformation detection systems must address and mitigate the effects of network segregation and ideological bias.