Seth Frey - Dartmouth College
Dec. 4, 2015, 2:30 p.m. - Dec. 4, 2015, 3:30 p.m.
Players in incomplete information games have an equivocal relationship with uncertainty: they can use it to stay unpredictable to others, but it also keeps them from accurately predicting others. Poker experts, for example, must extract information from opponents without revealing any about themselves, even though integrating more information from an opponent’s behavior must mean increasing one’s statistical coherence with it. How do strategic game experts stay unpredictable without decoupling themselves from the valuable information contained in an opponent’s signals?
Comparing expert and novice poker players, "sharks" and "fish" in online No-Limit Texas Hold’em (NLHE), we find that poker experts can have their uncertainty cake and eat it to, thanks to how well they integrate public information from others with private information in their cards. It should be no surprise that the integrative information processing of experts leads them to make more profitable decisions. But information processing can play the additional strategic role of making you harder to read. When you integrate two information sources, one public and one private, the private information obfuscates how the public information influenced you: you perform encryption. Using a large poker corpus and data-driven statistics — specifically a multivariate generalization of information theory — we show that poker experts functionally encrypt public information with the "private key" information hidden in their own private cards. This enables them to profit from others opaquely and, therefore, unreservedly. We call it
strategic information encryption.
Seth Frey studies human decision behavior in complex social environments. He is a postdoctoral fellow at Dartmouth's interdisciplinary Neukom Institute for Computational Science. He was a postdoctoral researcher at Disney Research, a part of Walt Disney Imagineering. In 2013, he earned a Ph.D. in Cognitive Science and Informatics at Indiana University, with Robert L. Goldstone. He earned a B.A. in Cognitive Science from the University of California at Berkeley in 2004.