关于博弈论及其它 - 未名空间MITBBS历史存档

国际科技财经博客移民网络热点娱乐民生时事公众号

Redian新闻

>未名空间

>Economics - 经济

关于博弈论及其它

关于博弈论及其它# Economics - 经济

w*g2002-09-14 07:09

1 楼

既然大家都叫我说两句，那我就罗嗦两句吧
首先，到底什么是博弈论呢？英文叫game theory，就是说是关于game的一种theory. 立
马给人一种玄虚的感觉，或者用大家常用的说法，不是“真实生活中的经济问题”。我
不知道是谁先用game theory这个词的，但我觉得它的中文翻译更能准确体现它的内涵。
那有人又要问了，到底什么叫博弈呢。在我在大学第一次见到这个词之前，我是从来没听
说这个词的；我也没查过汉语大词典什么的，但我感觉中汉语里面本来是没有这个词的。
欢迎指正。但这个词妙就妙在你一见到它，你就知道它是什么意思，尽管可能你也说不好
到底是什么意思。在以后多少窥探到一些门禁后，才知道是在搞博弈论的和不在搞博弈论
的有意的无意的把它弄的太玄虚了。其实，说白了它就是研究人作为经济体之间相互inte
ract的一种methodology. 姑且就把这做为定义吧。任何不符合这个定义的，都不是博弈
论。
有了定义就好办了。我们接下来就看什么不是，或者不一定是博弈论。第一个需要澄清的
，它本身并不是一个研究的兴趣，或者方向领域，它并不是并行于micro, macro, io,
eco

w*g2002-09-14 07:09

2 楼

usually by bargaining, people mean bargaining power, i.e. Nash-bargaiing,
where people in (re)negotiation process simply share some total pie according
to their relative bargaining power. however, there are some papers detailing
the bargaining process, i.e. the alternate-bargaining model, the employer-
employee bargaining model.
lack of common knowledge is a tricky stuff, which I dont see satisfactory
technique to actually adress the famous knowledge cycle. usually people talk
about incomplete i

w*g2002-09-14 07:09

3 楼

as long as there is some infinite cycle about knowledge, that is, not
everything is common knowledge, which in turn by itself is not common
knowledge, then it is hard to model this thought process. it is tricky becasue
of the infinitiness. as long as we assume the lack of common knowledge stops
at some point, then we can model it as uncertainty, and look back from that
point using Bayes rule. but if it is infinite, no matter what the lack of
common knowledge is about, we can not use Bayes rule

w*g2002-09-14 07:09

4 楼

the rubenstein bargaining model is qualitatively different. There is no lack
of common knowledge there: everything is perfect known; there is no incomplete
information either, though you can incorporate that issue. and what further
makes it easier is that, everytime someone takes the turn to propose,
everything is the same as the last turn when he/she proposes, when no
information releasing is involved. for the lack of common knowledge issue,
every round of the throught process is totally differ

c*s2002-09-14 07:09

5 楼

Bottom line: If we have complete information about strategies and payoff,
there are no bargaining and repeated games, as the game will be end at the
very first move and first mover has advantages in making the results in
favor of himself.
As a matter of fact, there are some interesting issues about repeated games,
in my opinion.
First of all, bayesian update is all about updating one's understandings of
his opponents, but NOT the beliefs of his opponents. For example, in a simple
case that you t

【在 w*****g 的大作中提到】

: the rubenstein bargaining model is qualitatively different. There is no lack
: of common knowledge there: everything is perfect known; there is no incomplete
: information either, though you can incorporate that issue. and what further
: makes it easier is that, everytime someone takes the turn to propose,
: everything is the same as the last turn when he/she proposes, when no
: information releasing is involved. for the lack of common knowledge issue,
: every round of the throught process is totally differ

w*g2002-09-14 07:09

6 楼

bargaining by itself only says that in the interaction between agents, no one
can propose a take-it-or-leave-it contract, either unilateral or in an
alterate way. that is all. it does not necessarily have anything to do with
repeated game or not, complete information or not.
Bayesian updating, or updating, can be about anything, either your parnter's
type, your type, or any other things.
we know time is money, so we need to discount future value. that is all we
need. by exactly how much to disco

【在 c*******s 的大作中提到】

: Bottom line: If we have complete information about strategies and payoff,
: there are no bargaining and repeated games, as the game will be end at the
: very first move and first mover has advantages in making the results in
: favor of himself.
: As a matter of fact, there are some interesting issues about repeated games,
: in my opinion.
: First of all, bayesian update is all about updating one's understandings of
: his opponents, but NOT the beliefs of his opponents. For example, in a simple
: case that you t