Hacker News new | past | comments | ask | show | jobs | submit login
Game Theory (2007) (yale.edu)
285 points by bezelbuttons 4 months ago | hide | past | favorite | 43 comments

To see Game Theory from a 'cooperation' point of view, check out the landmark book, The Evolution of Cooperation[1] by Robert Axelrod.

Axelrod's work was also summarized in The Selfish Gene. (And more recently, in Behave, by Sapolsky.)

From the abstract:

"We assume that, in a world ruled by natural selection, selfishness pays. So why cooperate? In The Evolution of Cooperation, political scientist Robert Axelrod seeks to answer this question. In 1980, he organized the famed Computer Prisoners Dilemma Tournament, which sought to find the optimal strategy for survival in a particular game. Over and over, the simplest strategy, a cooperative program called Tit for Tat, shut out the competition. In other words, cooperation, not unfettered competition, turns out to be our best chance for survival."

[1] https://www.basicbooks.com/titles/robert-axelrod/the-evoluti...

I would highly recommend watching the mathematician Hannah Frys documentary called the joy of winning (link below). It has interviews from a wide variety of academics and overall a fun way to understand game theory. The program you mentioned tit for tat is also mentioned in there as the optimal way to get what you want and why pure selfishness dosnt work. It’s a great investment of time. For less than a hour you get a lot and besides test the waters if you want to dive deeper.


>The program you mentioned tit for tat is also mentioned in there as the optimal way to get what you want and why pure selfishness dosnt work.

I've never been a fan of this framing. Cooperation can be viewed selfishly when the cooperation occurs because of well informed and strategic parties in the cooperation. A great example of this in real life are cartels. Cartels are only cooperative to the point where it optimizes a stable level/equilibrium of selfishness. Most cartels would happily disband and pull all the gains but the participants realize their minimal cooperation results in the best outcomes for themselves.

I guess the main issue I have with this boils down to the underlying intent and framing of intent. Ultimately, the intent is still driven entirely by selfishness, it's just selfishness with strategy and realizing that sometimes taking everything you can from everyone won't always navigate you to your selfish goal-state.

Perhaps that's what you mean here by "pure" selfishness not working. I'd argue that pure selfishness is the intent to only do what's minimally required for others for self-gain. You can be strategic or instrategic about it.

Something like "altruistic" cooperation on the other hand creates cooperation for the sake of cooperation alone (building trust, social cohesion, stability for more people). There may be little or nothing to gain for the cooperation for some participants, perhaps even some lose by participating (knowingly or unknowingly), but the cooperation as whole benefits and those benefits are worth the costs to the participants. The cooperation may exist to help some other party with no expectation of reciprocation. I think of it like a potluck dinner: everyone brings a dish and everyone gets to eat and have a good time. Some may not bring a dish and that's fine, they still eat and gain and others know that's the case, meanwhile some may fix elaborate dishes, expensive in terms of preparation time/complexity and/or ingredients and they do so knowing they're going over and above and won't necessarily be rewarded (perhaps a compliment) but they do so anyways because they wish to share the dish or experience with others.

Is this related in any way to the ideas Ayn Rand laid out with Objectivism?

Not really. Perhaps if you squint really, really hard?

No, not at all. It's about how it is strategically better for your own goals to work with others.

Though I got downvoted quite a bit I thought it was an interesting question about the relationship between Objectivism and Game Theory and found a couple links. Keep in mind that these are from the Atlas institute so they are surely very biased opinions, but I thought it was interesting to see the defense of Objectivism in regards to game theory questions.

One about game theory and its relation to Objectivism: https://www.atlassociety.org/post/game-theory-and-objectivis...

And another about the Prisoners Dilemma specifically and which touches on the Nash equilibria: https://www.atlassociety.org/post/prisoners-dilemma

And for people like me who are interested in the relationships between the two, here are some opinions of people from quora: https://www.quora.com/How-does-Objectivism-view-the-Nash-equ...

It's a fair question, but the whole point of Axelrod /The Selfish Gene / Cooperative Games is to call out how much a game theory approach would support strategies which are anti-Objectivist. Where contributing to the commons, investing in public good, cooperating, radically grows the pie to the point where it's in everyone's self interest to cooperate.

Basically, times in life where it's a 'positive sum' as opposed to 'zero-sum' game.

Non-zero by Robert Wright, is probably the best synthesis on this topic.


How is cooperation anti-Objectivist? If the cooperation is voluntary and within your self-interest then it isn't incompatible.

Sorry for the downvote. It was a good question, but I did not want to encourage a link between game theory and objectivism to the casual observer because I seem them as complete opposites. Objectivism is, at best, short-sighted - screwing over the people around you because it is convenient in the moment and becoming indignant when the folks that were mistreated no longer want to be your friend. (At worst it is justification for people that enjoy inflicting cruelty on others.) Game theory can demonstrate how to be truly selfish, it is in our best interest to be kind, thoughtful, and good communicators and to cultivate strong and healthy relationships.

Objectivism is just toxic foolishness.

I actually think there are some parallels when you listen to Rand herself talk about cooperation in a selfish framework: https://youtu.be/CoAKer8lfds

I think she outlines cooperation as an intrinsic result when people are acting within their own self interests which I think was why I originally saw the relationship.

The URL to download the course is on HTTP and not HTTPS. Quick link - https://openmedia.yale.edu/cgi-bin/open_yale/media_downloade...

Thank you!

Off topic but there is also a course there on Financial Markets by Robert Schiller of the Case Schiller Index that I found very interesting.


Shiller does such a great job explaining finance theory in simple terms! Would recommend

Schiller postponed his Narrative Economics mooc last year, regretfully

Wondering if it's better than the Coursera class with Stanford and University of British Columbia professors.


I just started taking that course last night from Coursera.

Please take both and report back!

Is there any real world usage of game theory for ordinary people?

Yes! Very much! I took a course of evolutionary game theory, which opened up my eyes and changed the way I think. So much that I'd recommend (evolutionary) game theory for everyone.

It's relevant in every day life: human interaction often fits the schema of the prisoner's dilemma, the signalling game or the chicken game. That tit-for-tat is the optimal strategy in iterated prisoner's dilemmata matches intuition about human cooperation.

The theory is beautiful, easy to understand and leads to verifiable predictions, especially in the context of evolution, such as the competition between genes, the behaviour of parasites, allocation of resources in a tribe, mating rituals, etc.

> That tit-for-tat is the optimal strategy in iterated prisoner's dilemmata matches intuition about human cooperation.

It certainly doesn't match my intuition because it's a quite mechanistic view of things. But speaking in mathematical terms, I have the suspicion that it's a solution for a very specific situation that generalizes with negative side-effects to social life. In any case, I've read that governments use this strategy when it comes to foreign affairs.

Well, intuitions can sometimes be completely off. But of course, tit-for-tat is just a general strategy it doesn't mean that you have to apply it blindly, without e.g. tuning the parameters. And, as far as I can remember, there are slightly modified formalized versions, where you allow for the other party cheating/competing without immediate retortion.

The idea is, that of course, in real life people are not 100% uniform, miscommunication and misunderstanding happens, etc. And you'll want to tune your parameters to the situation at hand. E.g. if you have a long standing relationship with the other actor and they've always been cooperating then you may not want to compete/retort when you see then doing it for the first time. (OTOH, what I personally found important is that in a new relationship, you should be very strict about this. Even if not immediately compete, shoot a very directed and serious warning.)

The general idea is that you don't let the other party to take advantage of you in the long run , but you always allow for de-escalation and don't try to take advantage of them in the long run either. (You will get taken advantage of, of course, because you'll always start cooperating, so if every other actor/player is competing then you'll lose on average. But not by much. However, if there are just a few one who are willing to cooperate most of the time, then you'll win big time.)

I agree, there are plenty of social scenarios that aren't congruent to iterated prisoner's dilemma.

E.g. if you implement tit-for-tat in your marriage, you're gonna have a bad time.

I'd say it's extremely important as a foundation to understand and better influence all kinds of politics, whether it's governmental, workplace, family, or any other relationships.

Most people have a very simple view of the nature of politics, and therefore, tend to think the people charge are out to get them in some way, when really, they're just operating under some kind of game that forces their hand to make difficult decisions. By understanding a little game theory, a person might at least be able to better understand what's going on, if not even influence the odds in their favor.

Obviously, ordinary people might not want to play this game, but the game will still play with you.

This video is a good rundown of that for government, https://www.youtube.com/watch?v=rStL7niR7gs

I’m a natural optimist, and game theory gave me a better grounding in thinking through and internalizing why things ARE, not just what they can or should be.

One analogy is shifting from a blame culture — why did that one person do that thing? — to a blameless postmortem: what is it about this system of individual actions that leads to this outcome? How did we get here? What mechanisms and incentives would repeat it? What would change it?

Lots of specific comments below. My experience was that game theory changed the way I think about almost any aspect of life, by giving me a set of simple tools. It's not the only toolkit you should have, but it's very powerful. "This is a coordination game; this is signalling; that is a prisoner's dilemma..." turns out to be a very useful vocabulary for thinking about situations, from everyday life up to superpower conflict.

For a great introduction from this point of view, check out Thomas Schelling's Strategy of Conflict. No maths, just a great read from a Nobel prize winner.

You may consider the Cold War a real world usage for ordinary people:


Optimal auctions, and mechanism design in general (ie screening problems) is a very useful application of game theory.

But caution: game theory is not about winning games, but finding equilibria. You might be more interested in behavioral game theory (how people learn the games, k-level thinking etc) if you want to get a leg up.

"Game theory is not about winning games, but finding equilibria."

This. However, if you understand the equilibrium for some behaviour, you can understand what would it take to change the behaviour.

I use this in my startup where we create an equilibrium for dentists to provide reliable dental care. And, it actually works.

Many real life scenarios involving co-ordination / co-operative action between people can be “boiled down” to a game theory problem. The hard part is generating an estimate of the payoffs and probabilities of different outcomes, that are normally just supplied to you in most theoretical problems.

What is "original people"? Did you mean "ordinary" people? In that case almost every branch of mathematics is useful for "ordinary" people if they have the skills or inclination to make use of it.

I presume ordinary in simply doing ordinary things like: doing the groceries, paying bills, doing a simple desk job or construction job, having one or two hobbies like playing a musical instrument or a sport.

Something you'd describe as being ordinary.

Though, I could be very wrong here, since it is definitely open for multiple interpretations.

> doing the groceries

Is there a shortage of groceries? Are people hoarding? Are they just 'bad people' or does a rational person change what they buy during a shortage? Why? Would you expect a Government saying "don't hoard" to have any effect?

> playing a musical instrument

Is there an orchestra you want to join? Will a particular instrument get you in the door easier than any other instrument?

> sport

Soccer. Why do you see so many obvious dives in professional soccer? Why do grown men hold their shins and cry when the replay shows they weren't hit?

Sumo. When a sumo wrestler is one win away from a promotion in a tournament, and he's facing an opponent who has already guaranteed his own promotion, what are the odds the first sumo wrestler will win?

It helps you understand the meta in Starcraft tournaments, but if you're interested in that - perhaps you're not ordinary...?

It's like a lot of things, you need to know enough to know why its useful, so its a bit of a catch-22. And then there's when you should apply it - a lot of the times just because you can doesn't mean you should.

It’s big in the crypto/DeFi space. Lots of thinking around “mechanism design” - how to engineer a system of incentives where some desired outcome is the stable equilibrium.

Adam Grant's Give and Take is a good layman application of game theory to interpersonal relationships. Tit for tat would be called the 'matchers'

can you be more specific about what exactly you're asking here?

are you asking "would an ordinary person encounter events in which game theory would be applicable on a normal, day-to-day basis?"

or are you asking "would an ordinary person who has studied game theory (but is not especially well versed or practiced) be capable of using it when it comes to making decisions?"

also can you please specify what you mean by an "ordinary person". how good are they at linear algebra and probability?

All of above.

And assume they are good at maths.

Various aspects of the stock market or, more generally, financial markets.

This course helped me when I was studying Mathematical Economics. Gem!

Such an amazing course it is. Loved it.

Yo that's my hacker news screen name :D

Applications are open for YC Winter 2022

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact