Game Theory?

Ed Thompson's picture
Submitted by Ed Thompson on Wed, 2016-02-10 02:39

Economic, agent-based, game theory research is something which could be ignored by Objectivists based on a first-principle analysis, but that may also be a reactionary mistake. After all, in On Liberty, Mill said:

"... but we have done the best that the existing state of human reason admits of; we have neglected nothing that could give the truth a chance of reaching us: if the lists are kept open, we may hope that if there be a better truth, it will be found when the human mind is capable of receiving it ... This is the amount of certainty attainable by a fallible being, and this is the sole way of attaining it."

It's a good idea to do whatever it is that you can in order to successfully refrain from neglecting anything "that could give the truth a chance of reaching us." On that maxim, a cursory glance at game theory research and results can be something which fits somewhere into a growing body of knowledge.

That said, where are the "problems" with looking into game theory research and results?

Well, you can't fall into the same philosophical traps of the university researchers who are running most of the prevailing research. For instance, homo economicus--or rational, expected-utility-maximizing man--is fine as an abstract concept, but you have to leave it there (in the abstract only). Mises talked about 'perfect competition' as being something that is fine to look at, but not fine to reify.

An example of university scientists "getting the philosophy wrong" is when they claimed that cut-throat behavior is most rational (because it is mathematically maximizing of an interacting agent's "income" or "payoff" in the short term, or for the "current round" of play).

It's as if they do not yet realize that rational action for humans is action taken with the long-term in mind--and with the inductive knowledge that reputations follow you. If these university elites "polish-up" their philosophy (or take tips from philosophical experts), then they will start to design games better--and scientific progress with accelerate.

The Machiguenga of the Peruvian Amazon were studied with an ultimatum bargaining game and it was discovered that they are stingy (they make some pretty tough ultimatums with each other).

A mistake would be to say that, because these primitive peoples act like they understand the game theorists' theory about rational behavior, that they are--for that fact and nothing else--more rational than civilized man. There are other things in play, and there is a philosophical potential to discover most and perhaps someday even all of them.

When everything peripheral-but-still-sufficiently-relevant is known, the results of game theory research will illuminate key truths to people who only think in concrete-bound terms (because the bankruptcy and extinction which can be shown to follow from certain behaviors and strategies in game theory, are things you cannot deny, even if you are anti-conceptual).

There are diffferent ways to reach people with the same underlying truth. Averroes said you should try 3 ways with people--depending on the listeners' philosophical acumen. Game theory research represents one of the ways to dramatically increase the scope of recognized truth regarding social systems, because its results are "in your face" moreso than abstract reasoning based on syllogistic accounts of basic truth and axiomatic concepts. It's really important not to personally neglect anything that could give the truth a chance of reaching you.

It is almost as important to become aware of the things which actively entail neglection on the part of others--such as concrete-bound, hyper-empirical folks--which preempt the truth from ever reaching them.

They vote, for instance.

( categories: )


Kyrel Zantonavitch's picture

Good observations, Ed. In my view, some of these bizarre questions do have applications to real life. They can also challenge and expand you mind. But most are a waste of time and effort, taking you away from important, difficult, and real philosophical problems that truly need to be solved.

For instance ...

Ed Thompson's picture

... let's say you were a country. Well, not really a country, but the head of a country--the head administrator of the government of a country. And, let's say that you were interested in jacking with the money supply, in order to fund grand infrastructure projects such as 80-ft tall statues of yourself. But the only thing that you are worried about is that if you print money out of thin air--in order to fund the statues and self-congratulatory buildings and whatnot--then the extra money in your economy may have an effect on the prices that your citizens have to pay for things.

Now you are in a pickle, because you are unsure of how much harm you will cause. Enter Game Theory. With game theory, you can estimate--in a rough-but-still-informative sense--how much harm you will cause to your people. That's because game theory research shortens time spans and allows you to narrow in on specific outcomes from specific changes in things. You have 3 places to go for answers:

1) a priori pronouncements
2) trying it out and perhaps harming millions of people in the process
3) checking it out with game theory first, before trying it out in real life

You can test whether a government (unproductive bureaucratic management of resources) can be 67% of an economy without effect--or whether it will cause double-digit deficits even in the first year, and increasing debt and deficits after that.

You can test whether having a gulag increases your chances of staying in power, or whether it backfires by accidentally making martyrs out of the political dissenters that you send there.

You can test whether you can get away with borrowing 42 cents on every dollar you spend from foreign entities--or whether that will backfire.

You can test whether respect of property rights and rule of law is necessary for expanding a country's wealth.

You can test whether heavy bureaucracy leads to corrupt bureaucrats holding up the paperwork and permitting of citizens--in order to shake them down for a bribe. You can even measure how fast it happens.

You can see the different results of different management styles. Etc., etc. The list is endless.



Ed Thompson's picture

You said life may be too short for pointless pseudo-philosophical speculation, and I completely agree with that. But check out this reasoning for a second:

There are 2 ways to know things (1) self-evident fluctuations in the ambient stimulus array (perceiving concrete changes and contrasts in your world) and (2) reasoning from the perceived changes to other knowledge inferred (contemplating the stakes of what you see and hear--how these things integrate with everything else you know).

Pseudo-speculation is a third way to go but--as you said--it is an illegitimate use of one's time (under the standard of growth in knowledge, or improvement in action). Looking into empirical investigations regarding agent interactions--ie., integrating game theory research--is, in my mind, not pseudo-speculation, but instead, grounding inferences in facts. Let's say that you wanted to determine whether my game--Two-headed Monster*--would be profitable for you to play. How can you find out?

Well, you can play it--but you will lose money to me doing that, because I know how the game turns out in the long run (I have knowledge of what it does even without personally engaging in it). Alternatively, you could rely on game theory research--which tells you how things play out even before you try them.

And, if you did that, you would not be "pseudo-speculating"--you would be saving your hard-earned money, and saving money is never a waste of time. There are things that can be picked up from such research which help real people with real problems. Gambling games are the easiest (every casino game can be modelled by game theory research and you will know all the results of playing them)--but other interactions with people based on sets of parameters are not different in kind, but merely in degree.

What this means is that their long-run consequences can be illuminated in the same manner that gambling games can, it just takes more attention-to-detail. The benefit from the research is that it shortens time-spans and allows you to focus on specifics in a manner impossible to someone just bumping along in life. Here are the options again:

1) just sit and speculate, like a fool, without ever checking assumptions against real interactions in the world
2) go out and get some personal experience of the world (try sucking on a helium balloon, try eating ice cream really fast, try dodging cars on the freeway, etc.)
3) utilize your capacity for integration of evidence, both from real life (unorganized, uncontrolled concrete experience) and from scientific investigations into matters (organized, disciplined, controlled investigation into interactions of concretes0.

You seem to think that game theory research is an example of (1), but I say it is a (3).


*Flip 2 coins. If they are both heads, I double your bet. If not, you lose your bet.


Mr_Lineberry's picture

said Kyrel - think I will stick to golf haha!

Game Theory

Kyrel Zantonavitch's picture

Does "game theory" relate to real life and have applications thereof? If so, what are they?

Or is it like: 1) sound of one hand clapping; 2) sound of tree falling when no-one hears; 3) two men in a one-man lifeboat; 4) life worthwhile standing in one yard of space forever; 5) death by drowning or suffocation; 6) sex with a cow or your mom; etc.? Life may be too short for ridiculous, moot, useless, pointless pseudo-philosophical speculation.

One more ...

Ed Thompson's picture


Who did it?
Samuel Bowles

What did he do?
He discovered that, in a peaceful society, you cannot sustain more than about 3% altruism evolutionarily. If you want more altruism than that--you have got to get your country into wars.

Why does it matter?
A lot of people think that there should be a whole bunch of altruism in the world. They think that it would make the world a better place. They want to kill the world with kindness (they think that that will actually work--when tried out in reality). But behaviors have got to be measured by at least 2 standards:

1. needs served or met by the new behavior
2. costs of adopting the new behavior

You have to balance these, or you are unrealistic--and being unrealistic is morally wrong (even if it feels oh so right in your heart of hearts, giving you lots and lots of warm-and-fuzzies).

You could, for instance, tell a lover you want to give her the moon--but if you actually did give someone the moon, then you would throw off gravitational trajectories in our solar system: endangering us all. Not to mention the fact that you would more than likely crush her (and yourself) with it during the "hand-off" (that thing has got to be pretty heavy, I'm sure of it).

It is not moral to be unrealistic and it is potentially very, very immoral to do so.

Some relevant research ...

Ed Thompson's picture

Who did it?
Gilbert Roberts

What did he do?
Discovered that justice (treating others as they objectively deserve, with reward and punishment) pays off with a net gain in value to humans, as long as the cost to punish others who acted in the wrong is only a third of the cost imposed on the guilty party--publishing his results in PLoS One under the title: "When Punishment Pays."

Why does it matter?
Some New-Age folks (Linz calls them "PoMo-Wankers") are currently of the position that you should reserve punishment of crimes for those times when it will help the most--making the punishment fit the criminal, rather than fitting the crime. They accidentally believe in non-retributive justice, tossing it aside with snide remarks to the effect that it is antiquated "an eye for an eye" draconianism. They think that they have arrived at a new version of justice--a "higher" view of it.

These cater-waulers think they've got the best view of justice--the one that "benefits society" and leads to peace and hugs by campfires and whatnot (with pretty birds singing in the cool night air). But if you stack up the views against one another (utilizing empirically-experimental tools such as Game Theory), then the "old-school" view of "an eye-for-an-eye" is proven superior--at least if you want to get ahead in the world.

For those people who are not interested in getting ahead in the world--and would rather just argue the issue to rhetorically win a debate simply for the sake of verbally out-competing others--there is little that can be done to them ... er, I mean "for" them (they will willfully refrain from understanding).

Who did it?
Boyd, Gintis, and Bowles

What did they do?
They discovered that if 4 of 18 people (23% of a population of individuals) make a pact to punish people who deserve it rather than ignoring them and letting them commit crimes left and right--then you already get two-thirds of the benefits of inter-individual cooperation from all higher levels of willingness to personally invest in establishing justice in your society.

Why does it matter?
A lot of folks think that there is no way to clean up corruption. They think that, in order to clean up corruption, you would have to have 90% of the population on the look-out for it and fighting it tooth-and-nail--at great personal cost and risk. This leads them to want to resign from all hope on the matter. But you really only need a fourth of the population to be concerned that justice gets served--in order for justice to get served.

A related issue is the drop in violent crime noted when gun laws are loosened in order to allow more guns in the streets via increases in carry-permits. Violent crimes drop for every percentage point increase of the population carrying guns in public--up to a threshold of about 23% of the population. In this case and in the case investigated by Boyd, Gintis, and Bowles--you get 80% of the potential benefit from having just 20% of the people doing the right thing.

The upshot is that the Bad Guys know this (they know that just 20% of a population could take them down), so they work hard to keep every truth-telling, justice-monger separated from the others--so that they never coalesce into a group that is the size of a fifth of the population. Machiavelli gloated how you can dupe entire societies of people who speak the same language and have the same values--preventing them from ousting you from unearned, arbitrary, capricious positions of power.


Mr_Lineberry's picture

only games I play are tennis and golf; there is no massive wank-fest involved and it's good exercise. Wonderfully social way to make lots of new friends too.

Interesting subject

Doug Bandler The Second's picture

Tit for Tat, Prisoner's Dilemma, the Trolley Problem, etc. They all are interesting and I think they do tell us something about human nature. However, one wonders if the assumptions they are built on are legitimate. The Austrians do not like game theory also as Von Mises placed economics on what he called "Praxeology" or assumptions he held about human action. I wonder if game theory stems from the same Humean waves as does positivism.

However, Linz is wrong that I'm obsessed with game theory. I'm more obsessed with socio-biology and brain scans lol. I get worse by the day.

I also keep bouncing back and forth between Oswald as the lone gunman and the CIA/Mossad/MIC conspiracy theory. Like women, I too am daffy.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.