A great way to look at how value is created in interactions between autonomous actors is game theory. How do autonomous actors respond to other actors? This is the core question here. How do you respond to an offer from a service provider? But also, how do they respond to you?

I have found game theory to be an effective model to think about the outcomes of sequences of interactions. This is because it clearly identifies the actors (or players) and their interests.

I use this all the time to think about how I relate to other parties that I choose to work with, or just have to work with. It also helps me to understand how parties and stakeholders act upon one another.

A game, in this view, has a number of players, who make moves that result in payoffs. The payoffs are defined by the rules of the game.

A simple example is tic-tac-toe, where players take turns (make moves) in marking an X or O on a 3x3 board. Whichever player first aligns three in a row wins. The win is called the payoff.

A strategy is a plan for the moves that you are going to make. A winning strategy is one that always wins, despite the moves of opponents. By the way, there is no winning strategy for tic-tac-toe. The best you can hope for is an opponent’s mistake.

The digital infrastructures world has many examples and applications of this. One example can be found by looking at a simple provider-consumer relationship.

In this relationship, the provider pools resources such as bandwidth and servers for multiple consumers, and charges a certain price to them. The moves they make are setting a price and setting the size of the pool of resources.

The consumers’ moves, simply stated, are to buy the service from this provider, or not, and to consume the service, or not.

The rules of the game are that the more consumers use the same pool, the less performance they get, which reduces their payoff. Think of the cake at a birthday party: the more guests there are, the smaller the slice each gets. Note: for the consumer this is an example of an externality (an outside factor that influences the value), which we discuss elsewhere.

The payoff for the provider is the amount of revenue they get. Having more users results in more revenue. The payoff for the consumers is the performance they get.

As a provider, you are probably looking for a strategy that maximizes your payoff, by balancing between having too much expensive capacity, or by alienating consumers by not having enough capacity. I once consulted an Internet Services Provider on how to play that game.

Contracts are games

Whenever you enter into a contract, you are also entering a game. The contract specifies the rules of the game, the allowable moves, and their payoffs.

When you sign up for a mobile phone subscription, your allowable moves include making calls and sending text messages. Each of these moves has a payoff. Part of the payoff is the charge from the mobile operator, and that is part of the contract. Terminating your contract is also a move, and the contract specifies your notice period.

Contracts between business entities often get a lot of scrutiny from lawyers. An important part of the work of those lawyers is figuring out what the game is.

  • What are the moves?
  • What are the payoffs?
  • Do other parties have a strategy that gives us negative payoffs?

Negotiating a contract means trying to establish the rules of the game. These rules will apply when the contract goes into operation.

Interestingly, and confusing to some, is that negotiating a contract is a game in itself. The moves are the various proposals, clauses, and amendments that are made. The payoff is a contract, which hopefully has value. But abandoning the negotiations is also a move. Negotiation experts therefore urge you to consider the payoff beforehand. BATNA, or Best Alternative To Negotiated Agreement, represents your moves if a negotiated agreement fails. Understanding it gives you power in a negotiation.

Taking a risk is a game

Whenever you make a bet, or take a chance, you are playing a game. In roulette for example, you are playing against chance.

You make a move, put your money on a color or a number, the table turns, and your payoff is either negative, you lose your money (likely), or you win and you get X times your bet (less likely).

Statistics helps you figure out what you can expect to win (or lose, as the case may be) based on your strategy. If you play in a casino, you can assume that the casino has a winning strategy, and you don’t.

In cybersecurity we are playing games all the time. For example: your software has a vulnerability. Vulnerabilities may lead to hacks, where your data gets stolen, or other bad things happen. Your move is to either fix it (negative payoff, as it is an effort) or not (zero payoff). Then you either don’t get hacked (zero payoff) or you are (big negative payoff). Statistics won’t help you much here. If a vulnerability is easy to exploit, it will be exploited eventually.

Cheating is a game

Cheating is a game too.

The cheater is taking a risk. If they are not found out, they win. If they get caught, they lose.

The move is clear, as is the payoff.

But it gets more interesting when two cheaters play together. One example of that is the game called the “prisoner’s dilemma”, which we will discuss in another unit.

Change the game

Automation often allows us to change the game, or at least change its rules and payoffs.

Consider the following practice. When software is changed, it may have new bugs. That is why software is tested before it goes to production. But testing costs money, so the game is the following.

Consider a test, now decide whether to run it, which is a move that brings cost, or don’t run it, and risk that it goes bad in production, which may bring a lot more cost. That is the game as it was customarily played before around 2010.

Enter: automatic testing.

If you automate the tests, you will significantly reduce the cost of running them. This means you can afford to do them more often, thus reducing the risk of bugs getting into production.

Now you are playing a different game. What are the tests that you are going to automate first? Automating tests costs money. The payoff is finding important bugs earlier. The move is to automate, or not.

This is a classic investment game. Do I invest in something, such as buying a server, and getting the payoff of low usage costs, or rent a cloud server, and have the payoff of not spending a lot of money upfront?

Understanding the games being played also helps understanding how power flows, because it shows how decisions by actors (moves by players) lead to outcomes such as benefits and costs. Game theory shows that benefits are not always reaped by those who can influence them.

Moves by one player can influence the moves by other players, because it influences their payoffs. In tic-tac-toe there are often forcing moves that leave the opponent only one choice except to lose. That also is power of one player over the moves of another player.

In a similar way, tech providers influence the choices of their consumers. A game a tech provider likes to play is ‘vendor lock-in’.

Once a consumer has started using a particular service, there will be a cost associated with switching to different provider. This is known as the switching cost. Providers can make several moves:

  • keep switching costs as low as possible, which will make it easy for customers to switch
  • drive up switching costs, of which the payoff is higher revenue
  • drive up switching costs to a really high level, of which the payoff is that they won’t be getting many new customers anymore.

Customers can make defensive moves, if they plan ahead.

And often, regulators step in to keep switching costs down, in order to protect customers. Phone number portability is an example regulatory move.