You Are Not Arguing About What You Think You Are Arguing About
Most of the arguments you have this week won’t be about what you think they are about. They will be about which game you are in. One person is working on a joint project and the other is in a contest, and neither has noticed.
Dr Daniel Bonevac’s example, in his Game Theory course on Peterson Academy, is two junior colleagues up for the same promotion. One believes the group does best if everyone shares credit. The other believes only one of them can win, so every conversation is a fight.
Neither is stupid. They are playing different games in the same room, and that mismatch, more than dishonesty or bad intent, is often what breaks a relationship or a team.
You don’t need to draw a single grid to use the course. Here are five questions from it, in the order I would ask them. The numbering and the wording are mine, not Bonevac’s.
1. Which game is this, and does the other person agree?
Bonevac sorts situations into three families.
In a coordination game you win by matching, like driving on the right.
In a cooperation game everyone gains if they coordinate, and each is tempted to defect.
In a competitive game one side’s gain is the other’s loss.
The right strategy in one family is a disaster in another. In coordination you want the other side to know exactly what you will do. In competition you want them unable to guess.
So before you choose a move, ask yourself what the other person wants out of this. If you cannot say, you do not yet know the game.
As I’m writing this, I have contractors on site dealing with significant damage in one of the flats, caused by a leak that is unusually hard to track.
The contractors want to get paid, so it’s in their interest to find the leak and stop it.
The affected residents, as patient with me as they are, obviously want the problem resolved, so they are pressuring me to work quicker and more effectively as the building manager. I want this pressure to stop, and that will only happen when the leak is found and stopped.
Even if we might have different reasons, the contractors and I have the same goal: finding and fixing the damn leak. We are in a coordination game. They know exactly what I have already done and what I am planning to do about it, and they tell me their plan of action too.
2. Is this choice mine, or is it the menu’s?
In Lecture 1 Bonevac gives Paul Samuelson’s test: if you choose X over Y once, you should not choose Y over X later just because the surrounding options changed. A real preference stays put.
His example is Nassim Taleb’s joke about ordering whatever the fattest person at the table orders, as they must know a lot about food. Bring in someone heavier and your “choice” flips, though nothing about the food changed. The same failure shows up in always taking the middle option, which moves whenever the other options move.
Try it on one choice you made recently.
Would it flip if one option was added or removed? If so, you reacted, you didn’t choose.
3. Is this bad outcome stable, and what would change the payoff?
In the prisoner’s dilemma each prisoner does better by testifying whatever the other does, so both testify and both do worse than if they had stayed silent.
I used to drive to work along tiny English country lanes, and almost every day I’d meet another car where the road was too narrow for us to pass.
There was a little negotiation.
If we both carried on, we’d end up bonnet to bonnet, blocking each other.
If one of us held our ground, the other had to give way.
The sensible answer was usually for one person to slow down and find somewhere to pull in, even if it meant a small delay.
It looked like a prisoner’s dilemma: each of us would prefer to be the one who got through without yielding. But it wasn’t quite one.
If the other driver kept coming, pressing on wasn’t a winning move. It was how we got stuck.
This was closer to a game of chicken. The question wasn’t whether to trust the other driver to cooperate. It was who would give way first, and whether either of us was willing to risk the worse outcome.
Wanting what is best for you whatever they do is ordinary rationality. It is also exactly what produces the worst result when both sides do it.
The fix is not to lecture people. It’s to change the payoffs: a rule, a fine, a price, or a reputation that makes cooperating the sensible move. Nobody gets out of a bad equilibrium by trying harder, because trying harder is not what is missing.
4. Will we meet again?
Against a stranger you will never see again, being difficult costs you nothing.
Against someone you will deal with for years it costs a lot.
Bonevac’s name for this is the footprint of the future: how large and how certain the expectation of dealing with each other again is. Treat any relationship you care about as a repeated game and enlarge its footprint on purpose.
He gives a short list of what separates strategies that do well over time.
Never defect first.
Be legible, because behaviour the other person cannot read looks like noise.
Be willing to retaliate, or you are just cooperating with whoever will take advantage.
Be willing to forgive, because one misread signal can lock both sides into permanent punishment.
And do not be envious of whoever is currently ahead. Stop scoring each exchange and watch whether the relationship as a whole is producing value.
Watch what happens when a commitment loosens.
A job taken as a stepping stone or a resignation already handed in shrinks the footprint, and behaviour changes, often before anything else has.
I used to work with someone who had a lot of self-esteem and confidence. Without sugar coating it, I thought he was arrogant. He thought he was better than everyone else and had higher expectations than everyone else about what he deserved. He treated all interactions as a game that must be won at all costs. He would be rude and demanding, and didn’t care about upsetting people with his comments. I watched him for four years seemingly getting more than the rest of us on the team. But as time progressed, I noticed that slowly but surely people including external contractors would refuse to work with him, and people would complain about him to higher management, until he had to go and start from scratch somewhere else.
I only had the language for that later, from Will Storr’s The Status Game, which I wrote about in The Most Dangerous People Are Not the Angry Ones. He was winning every exchange and losing the longer game that runs underneath them.
Luckily for me, fairly early in my life I bumped into someone I had dealings with some years before, and to cut a long story short I learned that I have only one face that I have to protect, even if it means losing a possible short-term gain. That lesson made me think about my interactions in the long term, and how my performance today can affect my life five years down the line.
5. What will they do in reply, and would they carry out the threat?
Bonevac’s lecture on games played in turns gives one habit worth taking away: start from the ending. Write the numbers one to twenty-one in a row and cross out one, two or three on your turn, and whoever is forced to cross out twenty-one loses. Most people play by feel and lose. The winner always leaves the other player facing a position where the last number crossed out is a multiple of four, which only becomes obvious when you work back from the end.
Ask what a good ending looks like, ask what position would force it, and keep asking one step earlier until you reach the present.
The same habit tests threats. A threat is credible only if carrying it out would serve the person making it. His comparison is telling a child to study or you will cut off your own nose: she does not believe you and she is right.
But keep the test honest. Bonevac also notes that people keep promises that cost them, to uphold norms they want to live under, and that an incumbent firm may fight a price war at a loss because driving out a competitor can protect profits for decades.
When the model gets it wrong
Do not take any of this as a prediction of how people behave. In the ultimatum game one player proposes a split of 100 dollars and the other can accept or reject, with both getting nothing on rejection. The model says offer almost nothing and accept anything.
Bonevac reports that proposers offer forty to fifty percent and responders usually reject anything much below a quarter, giving up real money to punish an unfair offer.
He reads this through David Hume: shared conventions about fair treatment matter more to people than the immediate gain.
So do not build a plan on the assumption that the other person will accept a lowball because it is rational. The questions above tell you the shape of the situation. They do not tell you what a real person will do.
The seven-day test I am running after the course
One small test a day, a few lines each.
Day 1: name the game in three interactions and ask whether the other person is playing the same one.
Day 2: take one recent choice and check whether it would flip if the menu changed.
Day 3: find a stable bad outcome and write one change in rules, price or reputation that would shift the payoff.
Day 4: pick one relationship and say plainly what you are doing and why, to enlarge its footprint.
Day 5: write the ending you want, then ask what the other person does in reply, and test any threat you are given.
Day 6: in one group decision, notice who sets the order of the questions.
Day 7: write one paragraph on which interaction turned out to be a different game from the one you assumed.
The question the whole test is built on is this: which game am I actually in, and does the other person know it?
I have run the test for three days so far, and I noticed that:
Even in everyday interactions with people closest to me, I play a different game than they do without realising it.
I was planning to buy myself something and I had two choices. I did my research and decided that one item beat the other hands down for what I needed. Both items were similarly priced. When the time came to pull the trigger, the losing item was on sale and I bought it instead of my preferred one.
On my old commute along the tiny country lanes that I mentioned earlier, sometimes we would both keep edging forward, each hoping the other would reverse. A simple rule could change that calculation: the driver nearest a safe passing place backs up. Then yielding is no longer an individual defeat or a guess about who should move. Both drivers know what to expect, and the road clears.
What do I think will stay with me after going through the course? The things I wasn’t aware of and that have real consequences.
I can see how knowing which type of game you’re playing is useful, and trying to answer that question is already becoming part of my life and has helped me avoid at least one unnecessary conflict.
Knowing if I’m choosing or reacting to surrounding factors is also useful. From my example above, I returned the discounted item after two days of using it and bought my first choice at full price, and I don’t regret doing it.
Point 4 above is the big one for me. As I said above, life taught me that lesson, and after twenty years I can see applying it to all my interactions paying off.
Question 5 is the one I want to practise, because strategic thinking that way is not my strongest trait.
What I am less convinced by
The five questions as a method. They are my summary of eight lectures, not Bonevac’s list, and Bonevac himself says the theory is often wrong as a predictor. I don’t have much evidence yet that asking them improves anything.
The advice to forgive and never defect first. It rests on tournaments and simulations Bonevac describes, where everyone sees the last move. Real people misread each other, and I do not know how well the advice survives that.
Whether any of this survives contact with a real argument, I don’t know. I haven’t tested it. That is the point of the seven days.
I wrote about the course itself, and why I finished it without getting excited, in Nobody Is Being Irrational and Everyone Is Worse Off.
The study pack for this course
Everything above is the short version. The full seven-day test, with the writing prompts for each day, is in the pack, along with the fourteen ideas the five questions came out of and 109 cards laid out to print and cut.
Thirty-four pages, free, and nothing held back. No email, no sign-up, no catch. It is below.
I build one of these for every book and course I work through. The ones for books are on the site and most of them are paid. They are built to exactly this standard, so if this one is worth your time, you already know what those are.
Sources & Disclaimer: Game Theory by Dr Daniel Bonevac, Peterson Academy; The Status Game by Will Storr.
This post contains affiliate links. If you buy through them I may earn a small commission, at no cost to you.
The reading and the arguments are mine. I write here to find out whether I have understood what I read, because explaining something is the cheapest way to discover whether you actually understood it or only recognised it.
AI is the other half of that. I argue with it, it pushes back, and what you are reading is what survived. English is not my first language, so it also helps me produce the clean version.
Nothing goes out that I could not defend if you asked.