Friday, September 7, 2018

To Bluff or Not to Bluff

02 March 2015
Author: Drew Calvert
Featured Faculty: Ehud Kalai, Professor of Managerial Economics & Decision Sciences at Northwestern University's Kellogg School of Management & Kent Grayson, Associate Professor of Marketing at Northwestern University's Kellogg School of Management

The bluff is a common strategic move that is often thought of as a clever psychological ploy when the odds are stacked against us. However, game theorists often view bluffing as primarily computationally, not psychologically. Professor Ehud Kalai argues that to win in any strategic game one should be unpredictable. If someone bluffs all the time bluffing is no longer an effective method because the opponent knows that they are likely bluffing. The same can be said when someone never bluffs. This article is set up into three sections, Mixing It Up, Knowing the Game, Knowing Yourself and High-Stakes Math.

Kalai defines strategic moves as maximizing one’s ability to be unpredictable. In court, Kalai had to show his ability to play poker and blackjack machines in both strategic and nonstrategic ways. The machines were banned from a local bar in Chicago until the he was able to prove that the machines could be played as a game of skill and not just a game of luck. Zero-sum games uses mixed strategy due to the balancing act of bluffing. This based on the assumption that the opponent is playing just as strategically.  

In non-antagonistic games it does not always pay to be unpredictable due to the requirement of full or partial cooperation. Kent Grayson, studied trust and deception.  Trust is comprised of three components: competence, honesty, and benevolence. He argues “bluffing is only effective when it is done with a measure of self-awareness.” How the company is perceived determines whether or not they can get away with bluffing without backlash.  In games bluffing can be harmless. In businesses bluffing comes at a higher risk.

Game theorists have proven mixed strategy to produce the best result. However, when the stakes are higher, some are not willing to take the risk on a bluff. In the last section, Kalai argues how randomization is tough when it comes to warfare and politics. The example he gives is during the Six-Day War between Israel and its Arab neighbors. Egyptian convoys used Israeli symbols on the roofs of their trucks to fool the Israeli bombers. The bombers were not willing to flip a coin on if the were taking out the right truck or not. The higher the pressure, the more risk-averse people become when it comes to randomizations.

The article To Bluff or Not to Bluff was well written overall. Even though game theorist argues that bluffing is computationally, not psychologically, they still take in to account the stakes of the bluff. I think the last section on High-Stakes Math was a very strong section. Even though the math is there to support the bluff, the psychologically doubt still there. I’m not sure that will ever gone away completely because it human emotions and error. 


  1. The randomizing example about the Egyptian trucks was interesting. It's a great example of changing the game. With game theory, players have finite choices. But in real life, choices are often hybridized or altered entirely. If you don't like your options in the game, change the game. I think the example was speaking more to advantage of making your opponent more risk-averse, but I would interpret it further to the point where you're presenting the opponent with so many options, they can't reliably identify and choose their optimal approach.
    I wonder, if you turn that around and think of it like, we want to identify as many possible scenarios as we can so we don't miss anything, do you think that can work against us as analysts? We want to give our DMs the most important information and the most inclusive, nuanced estimates. But by identifying every possible outcome, scenario, and method, do we run the risk of making a decision impossible?

  2. Alyssa,
    I find your post to relate to Bryant's particularly well in regards to the overlap of game theory and international relations. Whereas Bryant discusses the importance of being predictable to encourage cooperation from allies and adversaries, you illustrate international situations in which bluffing is advantageous. This contradiction about whether to be predictable or not in international relations prompts me to think of a bluffing strategy as conditional: only bluff against adversaries and in even then only in tactical decisions (like combat). In contrast, I think we ought to be predictable at the strategic/policy agenda level to encourage cooperation from both allies and adversaries. This however, poses a Catch-22 in that adversaries may recognize the conditions in which we tend to bluff (only against adversaries and in tactical decisions); thus, nullifying the advantage of bluffing. The role of bluffing is an interesting discussion point and perhaps supports the general consensus that a mixed strategy of predictable/unpredictable is the most effective strategy. Do you agree that a mixed strategy is "produces the best result"? Great post!

    1. I agree with Tom here and it goes back to the idea of really researching the competitor on what is the most effective strategy. I think the of range of models in game theory, some are able to map out this idea better than others. One model that specifically comes to mind for "bluffing" is the zero sum model. However, I don't think zero sum is as transferable to real-life situations.

  3. I enjoyed reading this summary of the article. I definitely think more is at stake, including lives, when it comes to making decisions in politics and warfare. It is only conventional that a high-risk situation makes one more hesitant in making a decision. Randomization may not be ideal, maybe pure strategy where decisions are not only made from your own preferences, is the approach when dealing with warfare or politics.