A few years ago, I first came across Game Theory at university.
I found the idea fascinating and read a little about it, but I never really went deep into it.
Until recently.
One of the things that brought me back to the topic was watching A House of Dynamite.
Without getting into the story or spoiling anything, one question kept coming back to me:
How do you make a good decision when you don't have complete information, you don't know what the other side is going to do, and making the wrong decision could be extremely costly?
That question sent me back to Game Theory.
But this time, I wasn't particularly interested in the equations or the mathematical models.
I was interested in a much simpler question:
What does a "right decision" even mean when the outcome doesn't depend only on you?
So, What Is Game Theory Really About?
In simple terms, Game Theory studies situations where the outcome of your decision depends not only on what you do, but also on what other people do.
And that last part is what makes things interesting.
A simple decision-making model might look like this:
I have several options → I evaluate the consequences → I choose the best one.
But real life usually isn't that simple.
Imagine you're negotiating with another company.
The outcome doesn't depend only on your offer. It depends on how they respond.
Two engineering teams are working on the same project.
The outcome doesn't depend only on your team's decisions. It also depends on what the other team decides to do.
Two companies are competing.
One country has to decide how to respond to another country.
Even many everyday relationships work this way.
We're not simply making decisions.
We're making decisions inside a game with other players.
And one of the most famous examples for understanding this is the Prisoner's Dilemma.
The Prisoner's Dilemma
Imagine two people are arrested for committing a crime together.
They're placed in separate rooms and can't communicate.
Each person has two choices:
Cooperate — stay silent.
Or:
Defect — testify against the other person.
Now there are several possible outcomes.
If both stay silent, they both receive relatively small punishments.
If one stays silent while the other defects, the person who defects gets a much better outcome while the person who stayed silent gets the worst one.
If both defect, they both receive significant punishments.
At first, the answer seems obvious:
Just cooperate.
Except there's one problem.
You don't know what the other person is going to do.
Suppose I cooperate.
If you cooperate too, great.
But if you defect, I get the worst possible outcome.
So I might start thinking:
"Maybe you're doing exactly the same analysis and you're going to defect. In that case, I should defect first."
But you're probably thinking the same thing about me.
The result?
We may both make decisions that are completely rational from our individual perspectives, yet end up with an outcome that's worse for both of us.
And that's one of the ideas I find most interesting about Game Theory:
A rational decision for an individual doesn't necessarily produce the best outcome for the system.
But What If the Game Doesn't Happen Only Once?
This is where things became much more interesting to me.
The simplest version of the Prisoner's Dilemma is a one-shot interaction.
But most real-world interactions aren't one-shot games.
Tomorrow, you'll see your coworker again.
Next week, your team will work with the same team again.
Next month, you might negotiate with the same company again.
The game repeats.
And once a game is repeated, my decision today doesn't only determine today's outcome.
It can change your behavior tomorrow.
That question led to one of the most famous experiments in Game Theory.
Axelrod's Tournament
In the late 1970s, political scientist Robert Axelrod explored a fascinating question:
What's the best strategy for an iterated Prisoner's Dilemma?
He invited researchers to submit strategies that could compete against each other.
Then those strategies played repeated rounds of the Prisoner's Dilemma.
Some were complicated.
Some tried to predict their opponent's behavior.
Some were aggressive.
Others used more elaborate patterns.
But one of the strategies that performed remarkably well was surprisingly simple:
Tit for Tat
Its logic was basically:
Start by cooperating.
After that:
Do whatever the other player did in the previous round.
If they cooperate, cooperate.
If they defect, defect.
If they return to cooperation, cooperate again.
That's it.
Why Is Tit for Tat So Interesting?
What interests me isn't really the simplicity of the algorithm.
It's the behavior that emerges from it.
Tit for Tat has a few interesting characteristics.
It starts with cooperation
Its initial assumption isn't:
"The other person is probably going to screw me over."
Its first move is cooperative.
It gives the other player an opportunity to cooperate.
But it isn't naive
If the other player defects, Tit for Tat doesn't just continue cooperating forever.
Bad behavior has a cost.
The strategy isn't saying:
Always be nice.
It's closer to:
I'm willing to cooperate as long as you're willing to cooperate too.
It doesn't hold a grudge
This might be the most interesting part.
If the other player defects once, Tit for Tat responds.
But if they cooperate again in the next round, cooperation immediately returns.
One mistake isn't punished forever.
And it's predictable
The other player can quickly understand the pattern:
If I cooperate, I'll probably receive cooperation.
If I defect, I'll probably receive defection.
That predictability is itself part of the strategy.
So... Did We Just Discover the Secret to Life?
Not exactly :)
And I think this is where we need to be careful not to turn Game Theory into a self-help book.
Tit for Tat performed well under specific conditions, and changing those conditions can change the results.
One particularly interesting problem is noise.
Imagine I intended to cooperate, but you misunderstood my action.
You think I defected.
So you defect.
I see your defection and think:
"Okay, so you decided to defect."
Then I retaliate.
Now we may enter a cycle of negative behavior that neither of us originally intended to start.
The real world is full of this kind of noise:
Miscommunication.
Incomplete information.
Misinterpretation.
Different incentives.
Unequal power.
And, of course, humans who aren't perfectly rational players.
So the lesson I take from Tit for Tat isn't:
"This is how you should treat everyone in life."
What I find much more interesting is this:
When interactions repeat, a good strategy can look very different from the optimal strategy in a one-shot interaction.
What Does Any of This Have to Do With Our Everyday Work?
This is where Game Theory stopped feeling like an interesting university topic to me.
Because once you start looking at work through this lens, you can see games everywhere.
Imagine two teams, Frontend and Backend, working on the same product.
Both teams have deadlines.
Both want to deliver quickly.
The Backend team can design an API that solves its immediate requirement as quickly as possible.
Or it can spend a little more time creating a contract that's also easier for the Frontend team to consume and maintain.
Frontend can make similar decisions on its side.
Individually, either team might reasonably say:
"Why should we pay the extra cost? The other team can solve their own problem."
From a local perspective, that might be perfectly rational.
But if both teams keep playing that strategy, a few months later you may have a system where integrations are painful, changes are expensive, and both teams hate working with it.
Each team optimized its own decision.
Nobody optimized the game.
Local Optimum Doesn't Necessarily Mean Global Optimum
You can see the same pattern in many other engineering situations.
Take Code Review.
Imagine your PR comes to me for review.
I'm busy.
I could spend 30 minutes understanding the context and giving you a thoughtful review.
Or I could quickly scan it and write:
LGTM ✅
In the short term, the second strategy is better for me.
I just saved 30 minutes.
Next week, my PR comes to you.
You're busy too.
So you choose the same strategy.
Again, both of us made individually rational decisions.
But what happens after a few months?
Code Review becomes a ceremony.
Quality drops.
Knowledge sharing decreases.
And something that was supposed to act as a safety net gradually loses its value.
"That's Not Our Team's Problem"
There's another sentence most of us have probably heard inside companies:
"Technically, this isn't our team's responsibility."
And very often, that's completely true.
Ownership matters.
Boundaries matter.
A team shouldn't be responsible for every problem in the organization.
But Game Theory gives us another question:
What kind of system do we create if everyone plays this strategy?
Imagine a problem sitting somewhere between two services, teams, or areas of ownership.
Technically, nobody directly owns it.
Every team also has an incentive to focus on its own roadmap.
From each team's perspective:
Not my problem.
From the organization's perspective:
Very much our problem.
This is where the difference between local optimization and systems thinking becomes very visible.
We Almost Never Play Only Once
One thing I've started thinking about more is how many workplace decisions seem to be made as if we're playing a one-shot game.
How do we finish this sprint?
How do we hit this deadline?
How do we improve our team's numbers this quarter?
How do we move this problem outside our scope?
But the game continues.
The team we're disagreeing with today might become our dependency again three months from now.
The engineer we didn't help today might review our PR tomorrow.
The technical debt we postponed today might show up again six months from now.
The trust we spend for a short-term win might take months to rebuild.
When the game is repeated, the definition of the best decision starts to change.
Reputation Becomes Part of the Game
In a repeated interaction, today's decision isn't the only thing that matters.
Other people are gradually building a model of us.
Does this person help when things go wrong?
Does this team respect its contracts?
When something fails, does this manager try to solve the problem or find someone to blame?
Does this engineer continue collaborating when their proposal gets rejected?
Does this company honor its agreements?
Over time, the answers to these questions become reputation.
And reputation changes how other players behave toward us.
So perhaps trust isn't only an ethical or cultural concept.
In repeated games, it can also become part of the strategy.
And That Brought Me Back to A House of Dynamite
What stayed with me after watching A House of Dynamite wasn't simply the crisis or the scale of the decisions.
It was the underlying question:
When information is incomplete, time is limited, and you don't know what the other player is going to do, what does a rational decision actually look like?
If you wait for more information, it might be too late.
If you react immediately, you might be acting on a false assumption.
If you assume hostile intent, your own decision might escalate the game.
If you're too optimistic, you might lose your opportunity to respond.
And what makes this kind of decision especially difficult is that:
Your decision isn't just a response to the game. It changes the game.
The other player observes your move and reacts.
Then you react to their reaction.
And the cycle continues.
Maybe We're Asking the Wrong Question
When we're making a decision, the question we often ask is:
"What's the best decision I can make?"
And that's not a bad question.
But Game Theory made me think more about another one:
"If every player responds to their incentives, what kind of system will this decision eventually create?"
Because what's best for me might not be best for the whole system.
What's optimal today might be a terrible strategy in a repeated game.
And something that has a short-term cost — cooperation, building trust, helping another team — might create a completely different long-term outcome.
That doesn't mean always cooperating.
It doesn't mean ignoring your own interests either.
And it definitely doesn't mean drawing a payoff matrix for every decision you make :)
For me, it's mostly a mental model.
A different way of looking at decisions.
What Kind of Game Does This Decision Create?
The more I read about Game Theory, the less I felt like it was telling me:
"Here's the right decision."
Instead, it helped me ask better questions.
Who are the players?
What are their incentives?
Who knows what?
Is this a one-shot interaction, or are we going to play again?
If I make this move, how might it change the behavior of the other players?
Are we confusing a local optimum with the best outcome for the whole system?
And perhaps most importantly:
What kind of game are we creating for the next round?
Maybe Game Theory won't tell us what the best decision is.
But it reminds us of something important:
Many decisions can't be optimized in isolation.
Because we're almost never playing alone.
Our decisions change other people's behavior.
Their behavior changes our next decision.
And something that looks completely rational today might create a repeated game that none of the players actually wanted.
So these days, alongside:
"What's the best decision for me?"
I try to ask one more question:
Top comments (1)