Name: Justin Sukernek

Ph.D. Dissertation Proposal Meeting

Date: Monday, Oct. 24, 2022  

Time: 3:30pm EST

Location:  Zoom click here

Meeting ID: 374 319 4773

Passcode: 503474

Advisor: Rick Thomas, Ph.D. (Georgia Tech)

Dissertation Committee Members:

Dobromir Rahnev, Ph.D. (Georgia Tech)

Sashank Varma, Ph.D. (Georgia Tech)

Jamie Gorman, Ph.D. (Arizona State)

Michael Dougherty, Ph.D. (University of Maryland)

Title: Exploring the Robustness of the Surprisingly Popular Signal

Abstract:  For years, the wisdom of the crowd (WOC) has been employed and investigated to answer questions in a myriad of applications, including forecasting and general trivia. A new innovation in this space, the Surprisingly Popular algorithm (SP), leverages the 'surprisingly popular signal,' originally introduced in 2004 with the Bayesian Truth Serum (BTS). Recent literature has found that when using only the best experts in a sample, SP can outperform WOC and other methodologies, including the best individual expert. I propose three experiments that would further our knowledge of SP and BTS, two relatively understudied methodologies with highly promising applications in forecasting. Two of the experiments serve to confirm the replicability of SP and BTS's effectiveness in previously researched tasks with added novel layers that continue to investigate expertise's role in boosting SP's accuracy. The third experiment explores the same concepts in a new consumer decision-making task which could capitalize on SP and BTS's social sensing signals to provide valuable insight into consumer preferences; social influence is also introduced to measure the effect it may have on answer preference, SP answer selections, and BTS scores. In all three experiments, the answer accuracy of SP, BTS, and WOC are compared. As a whole, the study will provide more clarity on the advantages and disadvantages of each methodology across three different task contexts.