Quote:
Originally Posted by Iam
Edit: And if mathematical models are the players in the game, then how do the people running the models decide which values best reflect that of the general population? How do they decide which values to run?
|
Mostly by declaration and definition. It tends not to assert "this is so," but rather that "if (equation here) is the motivational schema, (outcome here) is the most stable and optimal form of action."
For instance, let's look at a simple game of suicide, with one player... let's start with a nonsuicidal player. Values will be ordinal, rather than cardinal.
Commits suicide : 0
Doesn't commit suicide : 1
Now, since 1>0, the non-suicidal have an advantage in not committing suicide. When played with a suicidal player, however...
Commits suicide : 1
Doesn't commit suicide : 0
...suicide becomes the desired outcome (1>0).
Of course, this is just using definition in a circular sort of fashion, and doesn't tell us much... so, let's create a new game, with variables... same basic thing, but...
Commits suicide : x
Attempts suicide and mangles self horribly : y
Doesn't commit suicide : z
Now, being suicidal is
still defined as x>z, and being nonsuicidal is still defined as z>x...
but, this game gives us a fascinating observation. Let's rearrange the choices, shall we? After all, if one had a say in whether one fails in sucicide, no one would end up in a carbon monoxide coma.
So... let c = "chance of suicidal success,' here, ranging from 0 to 1. 'n btw, x, y, 'n z are now cardinal, not ordinal.
tries to commit suicide : (x * c) + (y * (1-c))
doesn't try suicide : z
Now, we can plug some values into this if we wish, looking for the highest value... zero makes a good breakeven point, incidentally... but we find that suicide attempts are predicated upon...
- the desire to commit suicide (x-z).
- the chances of success, and...
- the undesireability of ending up mangled by a failed suicide attempt.
In fact, regarding the last two... as c approaches 1 (100% certainty of success), or as y approaches zero (don't care if one leads a long painful life as a deformed cripple), the chances of suicide become more and more plainly the simplified game above, the relationship between x and z. In fact, if either y or c is zero, the evaluation becomes an essentially pure x:z relationship.
Conversely, however, if one were an outside interest in suicide prevention (here pitted against the suicidal "player," who is the mathematical equation directly above), your best solution would be to
maximize the percieved costs of failure (y),
while minimizing the percieved chances of success (c). The value of any effected change to either depends on the resources avaliable to you contrasted against the value of player 1's non-suicide, and the cost and effectiveness of changing any given variable - decreasing c, for instance, is useless for all cases where y=0, x>0, and z<0, and thus would not be a valid use of resources at any cost until y was increased. Thus, the game becomes player one trying to get accurate information to make the real-world decision which actually has the optimal outcome, while player two tries to manipulate perception of costs to deter suicide with minimal resource expenditure.
If you want to see how that plays out IRL, look at the drug war in the US, where goverenment "drug education" foists bullshit while citizens try to cut through the bullshit and get accurate information for their choices.
Quote:
Originally Posted by Iam
How can they possibly say that whatever works best in the mathematical model relates to how people work? It seems like it wouldn't be at all a satisfactory mode for understanding human behavior still.
|
By itself, it is not - all the game schema for a single-player game of suicide says is that
if a person calculates their chances of success, benefits of suicide, costs of failed attempts, and undesireability of life in the fashion described by the equation,
then the equation describes whether or not, and under what conditions, a person will commit suicide.
In short - if the equation is accurate, the equation is accurate. Doesn't say much by itself, does it?
Usually, attempts to derive all the factors are produced by simple brainstorming - one is a representative human, try to think of everything that could be a factor and chuck it into the equation. It's simple, but it works a lot more often than one would expect - crap like "I prefer having a fuckload of money over not having a fuckload of money," or "I would rather not experience extreme pain than experience extreme pain," are fairly common things... and luckily, since the variables are unquantified, and since the equation describes the relation, there are no complex thought in the variables themselves - "I would like a fuckload of money, but not at the expense of others' welfare" is simply "I would like a fuckload of money" and "I would prefer not to be responsible for harming others," rather than a single complex variable... so, it's pretty easy to figure out 'n track down. Remarkably simple impulses.
The study of
how people factor their decisions is usually experimental psychology, which is a completely different field... and that involves everything from polling people's opinions, to actually running complex games and seeing if they behave as the equation predicted, to throwing people in a situation, observing what they do, trying to form an equation to describe the results, and then testing that equation by adjusting variables until one has it right. 'n often, experimental psychology involves wierd shit which doesn't
involve but still
affects game theory, including the experimental psychology of game theory... wierd shit like "tryptophan increases trust" (actual finding)... which, in game theory, would translate to a reduction in percieved risk from favoring an outside human actor, generally.
So, yeah... these mathematical models are nothing more than an if-then, with no claim of proof (in game theory, experimental psych goes looking for proofs just as much as it looks for things which affect variables). It doesn't say "if I mug you with a lethal weapon, you will give me your wallet" - it says "in mugging you with a lethal weapon, if and only if you value your life more than the contents of your wallet will you surrender your wallet." (That's a bit simplified, actually - there's chances of success or failure.. but... yeah).
It's some pretty useful shit in political work, but every now 'n then, you have to raid pubmed to answer the question of "what's the motivation," especially when the motivations appear nonsensical to one's own point of view (why the fuck would anyone outside government desire more government control?)... 'n one has to be able to revise one's equations to match the evidence, in realtime, in the field...
...but, yeah. In the "suicide" game for one, "player 1" was the equation ((x * c) + (y * (1-c))) > z. Nothing more, nothing less... 'n as the chances 'n values change, so will the player's actions in the game.
's nifty stuff. Especially for taking on governments and hostile domestic terror cells.
We Need a New "Game Theory"
Re: We Need a New "Game Theory"
Re: We Need a New "Game Theory"
Linear Mode
