Strategic Misinformation Generation and Detection

52 Pages Posted: 31 Jul 2024

See all articles by Wenxiao Yang

Wenxiao Yang

University of California, Berkeley - Haas School of Business

Yunfei (Jesse) Yao

The Chinese University of Hong Kong (CUHK)

Pengxiang Zhou

Hong Kong University of Science and Technology

Date Written: July 22, 2024

Abstract

Misinformation detection is becoming increasingly important and relevant because it is easier than ever to create and disseminate misinformation. How does detection ability affect the incentive to generate misinformation? Given the practical constraints of classification technology, how should a detector be designed? To answer these questions, this paper studies the problem where a sender strategically communicates his type (high or low) to a receiver, and a lie detector generates a noisy signal on the truthfulness of the sender's message. The receiver then infers the sender's type both through the message from the sender and through the signal from the detector. We find a non-monotonic relationship between the probability that the low-type sender is lying and the accuracy of detection. More accurate detection (a higher true-positive rate and a lower false-positive rate) increases the probability of lying when the true-positive rate is low, because of a persuasive effect. By contrast, more accurate detection decreases the probability of lying when the true-positive rate is high, due to an alarm effect. We also characterize the optimal detector design. The designer always chooses the lowest feasible false-positive rate for any true-positive rate. The possibility of false-positive alarms implies that the designer chooses an intermediate true-positive rate rather than the highest true-positive rate. Counter-intuitively, the optimal detector may raise an alarm about a smaller percentage of misinformation when its underlying classifier is better at distinguishing the sender's type.

Suggested Citation

Yang, Wenxiao and Yao, Yunfei (Jesse) and Zhou, Pengxiang, Strategic Misinformation Generation and Detection (July 22, 2024). Available at SSRN: https://ssrn.com/abstract=4901012.

Wenxiao Yang

University of California, Berkeley - Haas School of Business ( email )

545 Student Services Building, #1900
2220 Piedmont Avenue
Berkeley, CA 94720
United States

Yunfei (Jesse) Yao

The Chinese University of Hong Kong (CUHK) ( email )

Shatin, N.T.
Hong Kong
Hong Kong

Pengxiang Zhou (Contact Author)

Hong Kong University of Science and Technology ( email )

HKUST Business School
Clear Water Bay
Hong Kong

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
178
Abstract Views
728
Rank
347,396
PlumX Metrics