Game Theory and Management

By Michael Hatfield

Rating: 7/10

Best Line:  The very best manager with 20% of the information needed to reach the best decision will be bettered by the poorest of managers who have access to 80% of the information needed.

Funniest Line: It didn’t take me long to realize that the accountants and I belonged to two different universes and theirs was the one where Spock has a beard.

I’ve always known that videogames helped me think more clearly. Turns out I wasn’t alone. But what really fascinates me is the way that gaming, as a whole, can help anyone. Games are a dance with uncertainty under fixed rules. Yes, the game of Life isn’t remotely as complex as life itself but features many things that rhyme with the bulk of what we actually experience.1

The more I’ve read across all literature, the more I see patterns emerge. From patterns come classic examples of cause and effect. From cause and effect, one starts to identify players, conditions, and models of behavior. Anyone who experiences the same things in enough different places develops their own variations of what can only be known as games.

Game theory then emerges from the question of how to best play these games. As the author puts it,

Game theory provides a structure—a concept, it you will—that allows a more prescient analysis of things that have occurred within the organization. [In] those instances where the payoff matrix can be accurately known and a set of strategies available to the players can also be known in its entirety, Game Theory may also have the capacity for predicting a probable future.   

Prescient analysis. Prediction of probable futures. Understanding what’s going on. The book helps us do all three while maintaining a proper skepticism on over-relying on any single technique. Three aspects stand out as the most valuable:

Game Types

It’s really important to understand the games in which you find yourself participating. It’s also really hard. In the vast sea of literature on game theory, you can get lost really quick in simply sussing out the nuances of each type. Differential games? I kinda get it but just barely. Anyway, critical components that I think everyone can grasp include the following:

  1. Zero-sum vs non-zero-sum (one’s gain is another’s loss vs both can win something)
  2. Simultaneous vs sequential (basketball vs chess)
  3. Perfect vs imperfect information (checkers vs backgammon)
  4. Combinatorial games (any game with no proven optimal strategy)
  5. Symmetric vs asymmetric (your strategy depends on their strategy, like tic-tac-toe, vs games offering different strategies to each player, like much of life)

I don’t think you have to be a game theory expert to get tremendous value from these concepts. If you’re able to simply identify that your endeavor has the quality of a game, with a goal and set of parameters, you’re already ahead of most people. To then relate the parameters to these game types means that you’re now light years ahead of most people in terms of thinking clearly.

For example, a game where you know there is imperfect information (most games in real life) will put you at a massive advantage to every rookie who thinks there is perfect information. A fine example is the poor sucker at the poker table who doesn’t understand the concept of a bluff. These people actually exist and actually go to poker tables and actually lose their money. An even better example in management is the poor guy who suffers “analysis paralysis” on his quest for perfect information. Even as Colin Powell would tell you to act after you have 40-70% of the information. Mr. Powell is 100% correct.

I could write more on the broader applications of these game types but we would be here for days. The fact is that these game types don’t simplify as much as you might think. They actually show you the true complexity and how much of it is beyond anyone’s control. Which, coincidentally, is a major lesson for any manager to firmly understand.

Payoffs, Decision Trees, and Available Strategies

I covered the payoff matrix in a previous article but it’s worth revisiting. Imagine the standard office meeting as a simultaneous game. Everyone participates in the meeting at the same time, presumably with the same goal of getting out of the meeting as soon as possible. When that is the goal, you and your colleagues consider your actions at any given turn through a basic payoff matrix of three outcomes:

  • You talk, they talk = good meeting with real conversation.
  • You talk, they don’t talk = you’re engaged, they are bored.
  • You don’t talk, they talk = you’re bored, they’re engaged.
  • You don’t talk, they don’t talk = good meeting because no one commits and it ends quickly.
Payoff Matrix Colleagues Talk Colleagues Don’t Talk
You Talk Good Meeting Engaged, Bored
You Don’t Talk Bored, Engaged Good Meeting

Jokes aside, it’s really important to see simultaneous games as a combination of several possible outcomes and finding the one that carries the best outcome. But lest we think payoff matrices are great for everything, it’s important to understand the real distinction here between simultaneous games, such as a meeting, and sequential games like a negotiation.   

Without getting into the details, the sequential game requires a different analysis of payoffs. This is where decision trees come in. These are my favorite mental technology. In short order, these trees give you a set of what you consider to be the total available strategies and the commensurate probabilities of cause and effect so that you know what will likely happen from each option. None of it is predictive, despite people’s efforts to make it so, but it really opens your mind to formally show others what you think could happen.

The point? When in a sequential game like negotiations, debates, board meetings, traffic management, and anything else where a player contributes, awaits reaction, and contributes again, a decision tree is a technique we all use. It’s so important that I’ll eventually give it a full week to discuss on its own.

Information Streams

Every game is fueled by information. In the office, a manager is only as good as the information they receive. Of course, there’s so much these days that it’s hard to know where to look. A helpful part of Mr. Hatfield’s book discusses the important of information that is timely, relevant, and accurate. Or TAR for short. Which, again, is very different from “complete” information.

In his words: Managers advance their competency by coming to appropriate conclusions and making the right decisions with less and less “complete” information.

TAR instead of complete information is what matters. After all, what is “complete” information anyway? It differs from one person to the next and is fixed on a comfort level instead of an empirical definition.

Why does this matter? Because every line of work spends vast amounts of time and money gathering information. Again, every game is fueled by information. So, too, is ever organization. But if we reviewed past decisions, we would find a whole lot (90% in many cases!) of the information we crammed in a staff report failed to be either timely, relevant, or accurate.

We’re playing the game all wrong in many cases. And while this doesn’t feel like a conversation on game theory, per se, it was a valuable insight.

Conclusion

Game theory is a vast topic. Management is a larger topic. Together, there’s simply no way that a single book and a review of said book can cover all the necessary ground. But the combination of the two, applied in specific circumstance and frameworks, is of great value. I can’t do justice to this other than to strongly recommend that everyone read this book. It won’t make you a better leader. It won’t make you a better manager. It will make you a better strategist and an immensely better observer.

Kudos to Mr. Hatfield for finding the intersection of these fascinating themes and bringing something to light. The writing is very funny at times and the examples are practically custom-made for someone like me. One chapter covers the stories of World of Warcraft as a means of explaining network theory. The next chapter features the Battle of Midway as a means of understanding the power of information. And the combination of Maccoby’s Gamesman archetypes with broader organizational situations is just fabulous.

And I can’t help but mention that his analysis of the Enron scandal in light of the State of California’s regulatory scheme completely expanded my mind. I had always felt something similar to what he wrote but he really crystallized it for me. Just fantastic.

I learned a great deal from this book. Strangely, though, much of what I learned is not a direct and obvious application of game theory to management. Some is a treatise on the value of knowledge itself. Some is about detailed techniques in project management (Hatfield is a regular, well-known contributor in the project management industry). Actually, a lot of the book is about project management.

That’s fine by me. But it’s not what a reader would suspect when reading the title and subtitle. I sought this book as a more nuts-and-bolts, technical and applied way of understanding Game Theory in my line of work. It gave me that in about half the pages; interspersed were brilliant tangents on many other topics and even a few stirring rants.

One in particular takes a whole chapter. In discussing risk management, Mr. Hatfield tees off with some great insights and then veers into citations of the Project Management Body of Knowledge (PMBOK) that feel wildly like inside baseball. It’s a Nassim Taleb-ian style of tangential ranting that could have been just as fiery, just as fun, had it been pointed in a better direction for readers like me.

So I take three arbitrary points off for those digressions. I do it solely because I think Mr. Hatfield could have benefitted from a better editor working with him. Not to make this book but to instead make two books. His voice is strong, his grasp of the information solid, and his humor is terrific. Take Game Theory and Management, split it in half with the titular items on one side, the tangents on the other book—titled Hatfield’s Musings—and expand both into their own themes. Everyone wins. We would have two books that score a 9/10 on the Norm-o-Meter rather than one small, slightly jangled volume that gets a 7/10.

I hope Mr. Hatfield will consider it. And continue his work. It helps us all.

The content is far richer than what has been portrayed here. You can buy the book on Amazon. It’s hard to find elsewhere.  

Mental Models and Principles

  • Variable rates of reinforcement guarantee behavior
  • Work Breakdown Structure
  • Payoff matrix for simultaneous games
  • Decision Tree analysis for sequential games
  • Nash Equilibrium – something everyone seeks in most office games we play
  • Risk categories – Avoid, Control, Accept, Transfer
    • Transferring risk is a fascinating idea
  • Monte Carlo simulation of project performance to establish contingency budgets
  • Reduce risk by dividing the project lifecycle into phases, separated by Critical Decision Points.
  • Intuition with broader bounds is better than rationality in narrow scope (a gambler’s intuition to never draw on an inside straight)
  • Game types within game theory
  • Cartage schemes – a set of canned strategies designed to be invoked when certain parameters are presented.
    • An If-This-Then-That for strategy
  • Maccoby player types – Gamesman, Jungle Fighter, Company Man, Craftsman
  • Metcalf’s Law and network effects
  • Personality tests are effectively an inside view at the preferred mixed strategies for a person in an organization
  • When the organization is losing, people act different. They play to maximize personal payoff.
  • Information is the lifeblood of the organization.
  • Information streams must be designed for timely, accurate, and relevant.
  • Management Information Structures – collect data, turn it into information, deliver it to decision-makers.
  • Tit-for-Tat strategies and the power of immediate retaliation and immediate forgiveness.
  • The Corner Cube Model (3-dimensional, three variable graph) of strategic, project, and asset management KPIs and cartage schemes on high and low KPI value.