U*****e
发帖数: 2882 | 1 Multidisciplinary Research: A Comparative Study
http://www.econ.cuhk.edu.hk/~wcsuen/howto/howto.html
but I am not sure if I understand the pictures of Environmental Science and
Politics. |
h***u
发帖数: 245 | 2 哦，这些都是孙永泉老师的幽默。他现在的港大主页上面，好玩的东西很多。
and
【在 U*****e 的大作中提到】
: Multidisciplinary Research: A Comparative Study
: http://www.econ.cuhk.edu.hk/~wcsuen/howto/howto.html
: but I am not sure if I understand the pictures of Environmental Science and
: Politics.
|
n***n
发帖数: 57 | 3 对呀，就是这样的。数学最纯美了。经济学永远在追寻增长，政治学永远在制造崩溃论
，环境科学嘛，长期耸人听闻制造末日轮。哲学就是瞎掰。 |
U*****e
发帖数: 2882 | 4 I see the point. Thanks.
【在 n***n 的大作中提到】
: 对呀，就是这样的。数学最纯美了。经济学永远在追寻增长，政治学永远在制造崩溃论
: ，环境科学嘛，长期耸人听闻制造末日轮。哲学就是瞎掰。
|
h*******l
发帖数: 359 | 5 Pol那张图应该是一条水平线，x轴是民主程度，y轴是经济水平。至于environment
science，应该画一个类似regression discontinuity design的图。
最重要的是，Jounalism那个点应该有无数条线穿过。
and
【在 U*****e 的大作中提到】
: Multidisciplinary Research: A Comparative Study
: http://www.econ.cuhk.edu.hk/~wcsuen/howto/howto.html
: but I am not sure if I understand the pictures of Environmental Science and
: Politics.
|
n***n
发帖数: 57 | 6 学术界有一段时间很推崇交叉学科研究。其实很难。经济学家，政治学家，法学家都不
一定能和谐对话，往往是自话自说，更何况自然科学和社会科学以及还算不上科学的很
多东东。 |
U*****e
发帖数: 2882 | 7 Talking about this, I met some people in Electric Engineering and they are
using Game Theory in network resource allocation, which has become a hot
area in EE.
【在 n***n 的大作中提到】
: 学术界有一段时间很推崇交叉学科研究。其实很难。经济学家，政治学家，法学家都不
: 一定能和谐对话，往往是自话自说，更何况自然科学和社会科学以及还算不上科学的很
: 多东东。
|
kx
发帖数: 16384 | 8 我觉得数学应该算作哲学
【在 n***n 的大作中提到】
: 对呀，就是这样的。数学最纯美了。经济学永远在追寻增长，政治学永远在制造崩溃论
: ，环境科学嘛，长期耸人听闻制造末日轮。哲学就是瞎掰。
|
f******k
发帖数: 297 | 9 but this often runs into the risk of lacking motivation, since the use of
game theory in economics is necessary for the purpose of modeling human
interactions, while this does not apply directly to most engineering
problems.
【在 U*****e 的大作中提到】
: Talking about this, I met some people in Electric Engineering and they are
: using Game Theory in network resource allocation, which has become a hot
: area in EE.
|
f******k
发帖数: 297 | 10 In recent years there is a strong collaboration between game theorists and
computer scientists (but more computer scientists into game theory than the
other way around). To quote a CS scientist from his blogging of GAMES 2008,
"...there was a CS-Game Theory lovefest...".
【在 n***n 的大作中提到】
: 学术界有一段时间很推崇交叉学科研究。其实很难。经济学家，政治学家，法学家都不
: 一定能和谐对话，往往是自话自说，更何况自然科学和社会科学以及还算不上科学的很
: 多东东。
|
|
|
U*****e
发帖数: 2882 | 11 sure. I did not mean you need a tooth brush to comb hair. In communication
networks, the interaction is among service providers as well as among end
users, so the application is relatively intuitive. I am also quite surprised
by knowing that EE values game theory that much.
【在 f******k 的大作中提到】
: but this often runs into the risk of lacking motivation, since the use of
: game theory in economics is necessary for the purpose of modeling human
: interactions, while this does not apply directly to most engineering
: problems.
|
v********e
发帖数: 1058 | 12 you're right. although the routers are machines, they are coded by selfish
service providers. Here's a Phd thesis on this topic by a computer scientist
, who's an AP in Stanford now:
http://www.cs.cornell.edu/timr/papers/thesis.pdf
surprised
【在 U*****e 的大作中提到】
: sure. I did not mean you need a tooth brush to comb hair. In communication
: networks, the interaction is among service providers as well as among end
: users, so the application is relatively intuitive. I am also quite surprised
: by knowing that EE values game theory that much.
|
f******k
发帖数: 297 | 13 in that case it is true that it is a good application of game theory. but i
have seen many cases that people just use the solution concept without any
serious validations.
on the other hand, i always think mechanism design/implementation theory is
close to engineering...
surprised
【在 U*****e 的大作中提到】
: sure. I did not mean you need a tooth brush to comb hair. In communication
: networks, the interaction is among service providers as well as among end
: users, so the application is relatively intuitive. I am also quite surprised
: by knowing that EE values game theory that much.
|
v********e
发帖数: 1058 | 14 what do you mean by "just use the solution concept without any serious
validations"? totally don't understand..
mechanism design/implementation theory is already one of the hottest topics
in theoretical CS community. The following is the 1999 paper that started a
decade of intensive research in this area:
http://iew3.technion.ac.il/~amirr/AMDJ.pdf
i
is
【在 f******k 的大作中提到】
: in that case it is true that it is a good application of game theory. but i
: have seen many cases that people just use the solution concept without any
: serious validations.
: on the other hand, i always think mechanism design/implementation theory is
: close to engineering...
:
: surprised
|
f******k
发帖数: 297 | 15 i always have a problem with this kind of research which has mushroomed in
recent years. basically what he does is just calculating the ratio of
aggregate utilities under NE and under Pareto efficiency. i do not see how
relevant is that as an index of NE inefficiency.
scientist
【在 v********e 的大作中提到】
: you're right. although the routers are machines, they are coded by selfish
: service providers. Here's a Phd thesis on this topic by a computer scientist
: , who's an AP in Stanford now:
: http://www.cs.cornell.edu/timr/papers/thesis.pdf
:
: surprised
|
U*****e
发帖数: 2882 | 16 Totally agree. There will be a long way to go ...
i
is
【在 f******k 的大作中提到】
: in that case it is true that it is a good application of game theory. but i
: have seen many cases that people just use the solution concept without any
: serious validations.
: on the other hand, i always think mechanism design/implementation theory is
: close to engineering...
:
: surprised
|
v********e
发帖数: 1058 | 17 the significance of the "price of anarchy" is immediate:
in a network, if we can prove the price of anarchy is pretty bad, then it
means that it will be much more efficient to have a centralized regulator
than to have a bunch of anarchic selfish routers. Building a centralized
regulator might be expensive, but you can determine whether it is worth the
resource by looking at the price of anarchy. If you only see it as Pareto
efficient, then it might not be clear, but if you realize that in a
tech
【在 f******k 的大作中提到】
: i always have a problem with this kind of research which has mushroomed in
: recent years. basically what he does is just calculating the ratio of
: aggregate utilities under NE and under Pareto efficiency. i do not see how
: relevant is that as an index of NE inefficiency.
:
: scientist
|
f******k
发帖数: 297 | 18 the thing is that the measure seems arbitrary to me. what's the axiomatic
base for this measure? why does the measure have total order? if i have two
totally different networks with the same measure, does that imply they have
the same inefficiencies? fundamentally, why should we care about cardinal
utilities in this scenario?
NE inefficiency has been extensively studied in economics in 70's and 80's.
what's new in the current line of research?
the
mandating
【在 v********e 的大作中提到】
: the significance of the "price of anarchy" is immediate:
: in a network, if we can prove the price of anarchy is pretty bad, then it
: means that it will be much more efficient to have a centralized regulator
: than to have a bunch of anarchic selfish routers. Building a centralized
: regulator might be expensive, but you can determine whether it is worth the
: resource by looking at the price of anarchy. If you only see it as Pareto
: efficient, then it might not be clear, but if you realize that in a
: tech
|
v********e
发帖数: 1058 | 19 The measure justifies itself by comparing the following two possible
scenarios: one is a laissez-faire policy, where service providers can
arbitrarily route flows to maximize their own interests, and the other is to
have a centralized mandating power that controls all the flows. The notion
of price of anarchy captures exactly the comparison between the social cost
unders these two scenarios, and gives a reference as to whether the latter
is to be preferred. If two networks have similar size and
【在 f******k 的大作中提到】
: the thing is that the measure seems arbitrary to me. what's the axiomatic
: base for this measure? why does the measure have total order? if i have two
: totally different networks with the same measure, does that imply they have
: the same inefficiencies? fundamentally, why should we care about cardinal
: utilities in this scenario?
: NE inefficiency has been extensively studied in economics in 70's and 80's.
: what's new in the current line of research?
:
: the
: mandating
|
U*****e
发帖数: 2882 | 20 “我靠 你们都好生猛啊！”
讨论越来越深入。谁来总结一下？
to
notion
cost
run
Why
【在 v********e 的大作中提到】
: The measure justifies itself by comparing the following two possible
: scenarios: one is a laissez-faire policy, where service providers can
: arbitrarily route flows to maximize their own interests, and the other is to
: have a centralized mandating power that controls all the flows. The notion
: of price of anarchy captures exactly the comparison between the social cost
: unders these two scenarios, and gives a reference as to whether the latter
: is to be preferred. If two networks have similar size and
|
|
|
U*****e
发帖数: 2882 | 21 终于读完了。
to
notion
cost
run
Why
【在 v********e 的大作中提到】
: The measure justifies itself by comparing the following two possible
: scenarios: one is a laissez-faire policy, where service providers can
: arbitrarily route flows to maximize their own interests, and the other is to
: have a centralized mandating power that controls all the flows. The notion
: of price of anarchy captures exactly the comparison between the social cost
: unders these two scenarios, and gives a reference as to whether the latter
: is to be preferred. If two networks have similar size and
|
f******k
发帖数: 297 | 22 hehe. basically my opinion is that line of research is just a mathematical
exercise but he thinks it does have lots of insights...
【在 U*****e 的大作中提到】
: 终于读完了。
:
: to
: notion
: cost
: run
: Why
|
U*****e
发帖数: 2882 | 23 没关系吧。当年portforlio theory, option pricing model也被人批评不够经济学的。
应用本身就是有价值的，如果帮助发展理论最好，不能也没关系啦。真要是有数学系的
人说经济学理论是a mathematical or statistical exercise，经济学家也没辙。
【在 f******k 的大作中提到】
: hehe. basically my opinion is that line of research is just a mathematical
: exercise but he thinks it does have lots of insights...
|
v********e
发帖数: 1058 | 24 I agree with you :-)
and a few more comments: most of those results are published in CS community
, and they bear significance in enhancing the community's understanding of
Internet specifically and networks generally, in addition to providing with
economic insights, the value of which is the main point of this thread of
discussion. It can be misleading to undervalue research in another area
simply because it doesn't fit into the understanding of one's own and
supposedly older area.
Second, I wa
【在 U*****e 的大作中提到】
: 没关系吧。当年portforlio theory, option pricing model也被人批评不够经济学的。
: 应用本身就是有价值的，如果帮助发展理论最好，不能也没关系啦。真要是有数学系的
: 人说经济学理论是a mathematical or statistical exercise，经济学家也没辙。
|
U*****e
发帖数: 2882 | 25 you are welcome.
Btw, I guess you are in CS, right?
community
with
in
be
【在 v********e 的大作中提到】
: I agree with you :-)
: and a few more comments: most of those results are published in CS community
: , and they bear significance in enhancing the community's understanding of
: Internet specifically and networks generally, in addition to providing with
: economic insights, the value of which is the main point of this thread of
: discussion. It can be misleading to undervalue research in another area
: simply because it doesn't fit into the understanding of one's own and
: supposedly older area.
: Second, I wa
|
v********e
发帖数: 1058 | 26 yes I'm in CS, and I've learn a lot here from you economists :)
【在 U*****e 的大作中提到】
: you are welcome.
: Btw, I guess you are in CS, right?
:
: community
: with
: in
: be
|
f******k
发帖数: 297 | 27 i am fine with many topics in algorithmic game theory, like combinatorial
auctions, computational complexity of NE, etc. they are essentially
economical problems solved by tools borrowed from computer science and they
have some deep impact on the economic theory itself.
but the other way around is not that exciting. do you have a new complexity
class inspired by some economic problems? does any economic result help to
address any fundamental questions in computer science?
this unidirectional inf
【在 v********e 的大作中提到】
: I agree with you :-)
: and a few more comments: most of those results are published in CS community
: , and they bear significance in enhancing the community's understanding of
: Internet specifically and networks generally, in addition to providing with
: economic insights, the value of which is the main point of this thread of
: discussion. It can be misleading to undervalue research in another area
: simply because it doesn't fit into the understanding of one's own and
: supposedly older area.
: Second, I wa
|
U*****e
发帖数: 2882 | 28 Sounds good. :)
【在 v********e 的大作中提到】
: yes I'm in CS, and I've learn a lot here from you economists :)
|
v********e
发帖数: 1058 | 29 Computer science is not all about complexity class and "fundamental problems
". You may have a distorted image about CS.
Economics (with game theory in particular) has certainly a great impact on
computer science. To name a few: in cryptography, the synergy between the
two areas was quickly recognized, and new classes of functions and protocols
are defined, motivated by economic considerations. In systems and networks
research, the area would look so differently today without considerations
insp
【在 f******k 的大作中提到】
: i am fine with many topics in algorithmic game theory, like combinatorial
: auctions, computational complexity of NE, etc. they are essentially
: economical problems solved by tools borrowed from computer science and they
: have some deep impact on the economic theory itself.
: but the other way around is not that exciting. do you have a new complexity
: class inspired by some economic problems? does any economic result help to
: address any fundamental questions in computer science?
: this unidirectional inf
|
v********e
发帖数: 1058 | 30 I just talked with another guy about this discussion. He told me I
misunderstood feedback's notion of "inefficiency". Price of anarchy is a
measure of relative inefficiency of a network with selfish users compared
with the same network controlled by a central rational power. Feedback was
arguing for an absolute notion of inefficiency, and that's why PoA doesn't
make sense. That's true, but PoA wasn't defined to address absolute
inefficiency at all.
I was also told that it's pointless to distance
【在 U*****e 的大作中提到】
: Sounds good. :)
|
|
|
f******k
发帖数: 297 | 31 it was pleasure talking to you and i learned something here too. thank you.
for NE inefficiency in the market economy, the chapter on the positive
theory of equilibrium in MGW would be a good start.
whose
but
【在 v********e 的大作中提到】
: I just talked with another guy about this discussion. He told me I
: misunderstood feedback's notion of "inefficiency". Price of anarchy is a
: measure of relative inefficiency of a network with selfish users compared
: with the same network controlled by a central rational power. Feedback was
: arguing for an absolute notion of inefficiency, and that's why PoA doesn't
: make sense. That's true, but PoA wasn't defined to address absolute
: inefficiency at all.
: I was also told that it's pointless to distance
|
v********e
发帖数: 1058 | 32 Thanks!!
【在 f******k 的大作中提到】
: it was pleasure talking to you and i learned something here too. thank you.
: for NE inefficiency in the market economy, the chapter on the positive
: theory of equilibrium in MGW would be a good start.
:
: whose
: but
|
d*******n
发帖数: 216 | 33 How come the link is not working now, does anyone know? I thought I checked
it about two weeks and it was fine. Now I even can't find him at all at the
department webpage. |
U*****e
发帖数: 2882 | 34 http://www.sef.hku.hk/~wsuen/
checked
the
【在 d*******n 的大作中提到】
: How come the link is not working now, does anyone know? I thought I checked
: it about two weeks and it was fine. Now I even can't find him at all at the
: department webpage.
|
h***u
发帖数: 245 | 35 因为Suen教授现在在港大而非中大，所以直接到他的港大主页去看：
http://www.sef.hku.hk/people/faculty/w_c_suen.html
http://www.sef.hku.hk/~wsuen/
checked
the
【在 d*******n 的大作中提到】
: How come the link is not working now, does anyone know? I thought I checked
: it about two weeks and it was fine. Now I even can't find him at all at the
: department webpage.
|
d*******n
发帖数: 216 | 36 Thanks, guys!
I like the previous title much better-- multidisciplinary research: a
comparative study--than the current one: guide to choosing college major.
【在 h***u 的大作中提到】
: 因为Suen教授现在在港大而非中大，所以直接到他的港大主页去看：
: http://www.sef.hku.hk/people/faculty/w_c_suen.html
: http://www.sef.hku.hk/~wsuen/
:
: checked
: the
|
U*****e
发帖数: 2882 | 37 me 2, hehe
【在 d*******n 的大作中提到】
: Thanks, guys!
: I like the previous title much better-- multidisciplinary research: a
: comparative study--than the current one: guide to choosing college major.
|
