702. CalGal - March 22, 1999 - 1:42 AM PT
Slack,
In that link, why is a utility of 1 assigned if both people confess?
It seems to me that it's assigned on the grounds of "well, if I don't get it he doesn't either"?
Yet in the other two situations, there is no future assigned. Like "Well, if he confesses and I don't, he'll save me my half."
But if I were one prisoner, and the other prisoner were just like me, the possibilities would be 10 if I confessed and he didn't or 0 if we both confessed. No utility in the case of both of us confessing. To hell with any satisfaction of knowing he isn't getting it either. I'm not getting any of the money, so who cares?
In that case, I figured we'd both figure on not confessing and getting 5 each.
I'm not trying to poke holes in game theory, since I've long resigned myself to the fact that I'm just weird. But I would like to know what the rationale is for the "1" in the case of both people confessing--why in that one case, emotional satisfaction is assigned a utility, or if there is another reason for it that I've missed.
703. pellenilsson - March 22, 1999 - 2:13 AM PT
Slackjaw
Thank you very much for your series of posts. Both enjoyable and enlightening.
704. uzmakk - March 22, 1999 - 9:50 AM PT
I lurk but have no time.
705. uzmakk - March 22, 1999 - 1:58 PM PT
Message #691
stostosto's poem makes it sound as though the philosophy thread survives in a less vital and interesting form--
"But, lo, it lives beyond its death
transcending to, at loss of breath,
philosophy and math."
From vital mythology, to less vital philosophy, to the enemic Philosophy and Mathematics. Oh, God, let it not be so! (Slackjaw knows that it is not so.)
To prolong the life of Philosophy and Mathematics and to aid in the restoration of its vitality, perhaps its transformation, I have concocted a spice----
There is a Game afoot.
It is the Game of Man.
It is the Game of Nations.
It is the Game of dominion and control over "other"men.
It is a game of Ignorance.
It is a game of Fear.
It is a game of Knowledge.
It is a game of Lust filthy past belief.
706. Slackjaw - March 22, 1999 - 4:08 PM PT
CalGal,
Those are all great points, and are not off the wall in the least. Indeed, they come up frequently in contexts like this.
The story Levine told to go with the prisoners' dilemma was a little contrived, so I intend to retell it.
The important thing for right now is that, if you take that game matrix as given, the strictly dominant strategy is to confess. The reason is that, for any choice the other player might make, your best response is to confess.
Importantly, there is *no* future in that game. It is played once and then the world ends.
And don't worry about poking holes in game theory!
More later.
707. Slackjaw - March 22, 1999 - 6:42 PM PT
here is a story giving rise to the prisoners' dilemma, one of the most famous games ever. This story was originally told by Stanford psychologists (Dresher and Flood) in the '50s, while giving a seminar at the RAND corporation.
A reprobate has been apprehended by the police and informed that he stands accused, along with a consort, of a major crime. The DA doesn't have enough to get a conviction on the major crime without a confession, but can get a conviction of both criminals on a trumped up charge. So, the DA is offering leniency in exchange for a confession. If the reprobate confesses but his consort doesn't, the reprobate gets off scott-free but the consort spends 10 years in prison. Similarly, if the reprobate doesn't confess and the consort does, the consort gets off scott free but the reprobate spends 1o years in prison. If both confess, both spend 7 years in prison. If neither confesses, both are convicted of the trumped up charge, and both criminals will spend 1 year in jail. Now, assuming that utility is strictly decreasing in time spent in prison, what should a utility maximizer do?
Collectively, the best outcome is "neither confess." Aggregate jail time is 2 years. The worst outcome, collectively, is "both confess." Aggregate jail time is 14 years.
Yet it is clear that, for any given choice the consort might make, the reprobate gets higher utility by confessing. The same is true for the consort, so the outcome is the worst possible: "both confess." This consists of a strictly dominant strategy for both players.
This somewhat startingly drives home the point that unlike in price theory from economics, "equilibrium" outcomes in games need not be efficient, efficient in the sense of leading to an outcome that is strictly worse in everyone's opinion than another outcome.
708. Slackjaw - March 22, 1999 - 6:43 PM PT
It is also somewhat at odds with intuition for some people. This has a lot to do, I think, with either a "what if" that isn't true, or a belief that repetition matters.
I'll say some more about it later, but I have to go for now. Just wanted to get the story down.
709. pellenilsson - March 23, 1999 - 5:29 AM PT
Slackjaw
I posted this to the International thread:
"I heard on the car radio a little while ago about a political
scientist Eleanor Astrom (unsure about spelling) who has
been awarded a rather prestigious price by the University
of Uppsala here in Sweden. I was in a tunnel at the time
so I didn't get it all but she seems to work on cooperation
and power sharing in small-scale structures, in particular
as regards the management of collective resources such
as water and fishing grounds. I she known to anybody
here?"
And got this from PseudoErasmus:
"If you're speaking of Elinor Ostrom, author of the famous
"Governing the Commons", that's Slackjaw's area."
Any comments?
710. stostosto - March 23, 1999 - 7:56 AM PT
uzmakk
"stostosto's poem makes it sound as though the philosophy thread survives in a less vital and interesting form--
"But, lo, it lives beyond its death
transcending to, at loss of breath,
philosophy and math.""
Why, I should have thought it transmits the idea that this thread has gone to a higher level. To heaven. To paradise.
oh....
I see what you mean.
711. Slackjaw - March 23, 1999 - 8:55 AM PT
great segue pelle, for Ostrom's meal ticket is in an area intimately connected to the moral of the prisoners' dilemma.
She is a professor and honcho at Indiana University. Her research, usually of a very applied bent, is on the design and maintenance of institutions to solve 'collective action problems.' The paradigmatic example of a collective action problem is the prisoners' dilemma (pd): rational behavior on the part of individuals leads them collectively to a suboptimal outcome.
An example is an irrigation system shared by several farmers. All can draw water from it whether or not they limit their access, or help with maintenance. Moreover, 'doing one's part' to maintain the system carries a personal cost. So loosely, the strategies are 'contribute' or 'don't contribute' (e.g., in the form of self-regulated access, maintenance contributions, etc.), and if a subset of players of some predetermined size (depending on the physical limitations of the system) doesn't contribute, the project collapses and everyone is without its benefits. The trouble is, whatever others choose to do, a given farmer always does better by not contributing than by contributing, at least in a 1-shot game.
This outcome should be reminiscent of the pd, and indeed, this sort of thing is why the pd became so famous--possibly one of two or three of the most important ideas in social science. (The idea itself isn't new to game theory of course, but can be found Hume, Hobbes, and earlier works. The game representation is simply a formalization.)
Of course, this outcome--collective failure--isn't characteristic of every public project in history. Some institutions for the maintenance of these sorts of things (access to grazing lands, road maintenance, etc.) are successful, and some aren't.
712. Slackjaw - March 23, 1999 - 8:56 AM PT
This doesn't mean the pd is useless or that the game theoretic analysis of it is 'wrong,' but rather that successful institutions transform the game the participants are playing--the incentives are different from those in the 1-shot pd. Ostrom collected the insights of numerous case studies of successful and unsuccessful institutions in the book PseudoErasmus mentioned. Successful institutions, in accordance with intution as well as the insights of game theory (for repeated games, which we'll talk about later), tend to emphasize long term relationships among the interested parties and have visible channels of contributions to the public project (so defectors can be easily spotted).
Ostrom tends to be very applied, yet her work is a great example of how these theoretical motivations can yield insights into real-world problems. She may well be the best known political scientist on earth, due to her extensive research on real-world institutions for collective action in urban and rural settings in the developing world. She has also done some laboratory experimental work on collective action. Very little of her work is purely theoretical in the sense of having a lot of Greek letters in it, and indeed her book _Governing the Commons_ is accessible to any educated person, but her work is valuable to theorists because it gives us something to shoot for in formalization of incentive schemes.
713. CalGal - March 23, 1999 - 10:01 AM PT
Slack,
In the link, I misunderstood this sentence:
"If both prisoners confess, then both are given a reduced term, but both are convicted, which we represent by giving each 1 unit of utility: better than having the other prisoner confess, but not so good as going free."
For some reason, I misunderstood this and focused on the latter part--"better than...."--I thought this meant that in this one circumstance, the game was thinking of the future. Your explanation clarified things. It wasn't that I was trying to think of the future, but rather that I couldn't figure out why in this one instance the game was.
My next question is this: Why is it not instantly obvious that the best thing to do is to not confess? To both people?
I understand that in game theory, when dealing with many people, it becomes more difficult. So my question only applies to this dilemma.
714. pellenilsson - March 23, 1999 - 10:44 AM PT
Slackjaw Message #711 and next.
Thank you. I have to read Ostrom. The matter has interested me ever since I read an article about fishing grounds in the Scientific American a couple of years ago (maybe Ostrom was a co-author? But no energy to go back to the big pile for a check.) It demonstrated how a limitation of catch would benefit the fishermen as a collective, yet the individual would have no incentive to obey the limitation. If you will touch on this or any related subject I shall be an avid reader.
715. uzmakk - March 23, 1999 - 10:57 AM PT
I think we are approaching an area mentioned in an article in Le Monde Diplomatique by a Frenchman named Bourdieu(?) THE ESSENSE OF NEOLIBERALISM. Our own Pseudoerasmus has dismissed the article as the work of an idiot. Perhaps we should return briefly to those threads of yesteryear. This after the Harper's article and what can be dug up for Pelle, if anything.
716. pellenilsson - March 23, 1999 - 11:22 AM PT
uzmakk <msg nm=715><BR><BR>I wonder if this is the same Bourdieu (if so, Pierre) who created a bit of stir in Europe last year by claiming that television is trivialising and segmenting reality to the extent that it is quite impossible to understand contemporary events by watchning TV? Trite you may say, but apparently underpinned by enough chic post-modernist philosophy to make an impression om media people.<BR><BR>(If you ask me about post-modernism I will scream.)
717. Pseudoerasmus - March 23, 1999 - 12:19 PM PT
Pellenilsson:
Collective failure, free rider effects, etc. made a brief appearance in a discussion of OPEC in the international thread.
718. Pseudoerasmus - March 23, 1999 - 12:20 PM PT
Uzmak, did you see the links I left for you in international?
719. Pseudoerasmus - March 23, 1999 - 12:24 PM PT
Pelle: PD, free rider problems, collective action failures, etc. appear in every conceivable area, whether you're talking about complications related to debt forgiveness during the Latin American debt crisis, or the failure of pollution control devices to be marketable without state intervention, or even in explaining why the existence of a government programme in a democratic society does not necessarily mean that there is much popular support for it.
720. pellenilsson - March 23, 1999 - 12:27 PM PT
PE Message #719
Sure, but it doesn't mean that the subject is without interest.
721. Pseudoerasmus - March 23, 1999 - 12:40 PM PT
Message #720
?????
My point was that the subject is interesting PRECISELY because of its wide applications.
722. Pseudoerasmus - March 23, 1999 - 12:44 PM PT
My education in game theory and allied matter is limited to the contents of the core curriculum in graduate micro, so I don't have anything like Slackjaw's command of the literature. However, I can suggest numerous books on these topics which are accessible to lay readers.
Besides the books by Axelrod, Olstrom, and Dixit & Nalebuff that have already been mentioned,
Here's a book on game theory for babies, and it's only $8 to boot.
This is an excellent overview of rational choice, social choice, decision science, applications of economic analysis to social justice issues, Hobbes, Hume, etc.
A book requiring some undergraduate-level economics is The Economics of Collective Choice.
And indeed, Elinor Ostrom's Governing the Commons is non-technical and readable.
723. pellenilsson - March 23, 1999 - 12:47 PM PT
PE Message #721
Misunderstanding on my part.
724. Pseudoerasmus - March 23, 1999 - 12:49 PM PT
"The Theory of Choice" referenced above is co-authored by an economist, a philosopher, and a political scientist.
725. uzmakk - March 23, 1999 - 1:17 PM PT
Pseudo, Message #718:
I checked back 100 posts in International. Can't find my self or anything addressed to me. I don't post there much,Pseudo, are you sure you have the right thread?. Sorry. Thats why I need my magic Fray Logbook which I am beginning to use.
726. BobaFett - March 24, 1999 - 5:09 AM PT
FTC:
My newest hypercube answer is: one hundred perecent of its volume lies within 1 meter of its surface.
Here's how I came up with this latest mathematical debacle.
Volume = 100 to the 100th power
for each side, the amount of volume within 1 meter of the side will be 1m x 100 99th meters
Since there are 100 sides, the volume within 1 m of the surface is 100 x 100 to the 99th, or 100 to the 100th power
Thus, the proportion is 1:1 or 100 percent.
727. FreetoChoose - March 24, 1999 - 6:38 AM PT
Here is a simple math puzzle, originally posted as a challenge in one of the political threads. However, given that we now have a math thread, I thought I would post it here (plus, it is less likely to get lost in the shuffle.)
Actually, I'll write it as two distinct, though very similar questions (because there may have been a mistyping on my part):
A. Consider a 100-dimensional hypersphere with diameter 100 meters. What proportion of the interior volume of the sphere is within 1 meter of the surface?
B. Consider a 100-dimensional hypercube with length 100 meters. What proportion of the interior volume of the cube is within 1 meter of the surface?
728. FreetoChoose - March 24, 1999 - 6:40 AM PT
A. BTW, PE and Slackjaw, I'll give you credit for knowing the answer. Let's let someone else have a shot.
729. BobaFett - March 24, 1999 - 6:44 AM PT
Well, FTC, I know how to figure out the answer, but I don't know the formula for the volume of a 100 dimensional hyperspehre.
What's the formula? 101/100 pi r to the 100th power? (I'm just guessing based on 4/3 pi r cubed for volume of a sphere).
730. BobaFett - March 24, 1999 - 6:49 AM PT
101/100 pi r 100 - 100/99 pi r 99
or 101 pi r 99 - 100/99 pi r 99
or 100 and 1/100 pi r 99
731. BobaFett - March 24, 1999 - 6:51 AM PT
oops, scratch that. I fucked up.
732. BobaFett - March 24, 1999 - 6:55 AM PT
(instead of giving the smaller sphere a radius of 99, I gave it 99 dimensions)
733. Msivorytower - March 24, 1999 - 6:56 AM PT
Here's a simple problem for you FTC:
In a semi-logrithmic function (regression), what do the partial coefficients represent? ie., how can you interpret them?
734. FreetoChoose - March 24, 1999 - 7:01 AM PT
MsIT
I've done regression, but I don't recall the term "semi-logarithmic". If I find some time, I'll poke around, but I cannot answer it without references.
735. FreetoChoose - March 24, 1999 - 7:13 AM PT
It occurs to me that different people may have different “rules” on what resources may be accessed when answering puzzles. Let me throw out (for discussion purposes) a hierarchy:
Type 1—Answer without using reference materials, or help from other people. Use of a calculator or spreadsheet for “simple” calculations allowed. (I.e., no use of symbolic math programs, Solver, or exhaustion of possibilities.)
Type 2—Computer use allowed
Type 3—Reference materials (including Internet) allowed
Type 4—Help from others allowed
It is my intention that people solve this as a Type 1.
(Puzzle aficionados might argue that some puzzles should be solved without the use of calculators. Note that I neatly left Type 0 for those people.)
736. BunEBear - March 24, 1999 - 8:11 AM PT
Start with B, The volume of the n-dimensional "cube" is L^n. The volume within s of the surface is L^n - (L-2s)^n. The fraction is thus (1 - (L/(L-2s))^n). For L=100, s=1, n=100, this comes to 86.7%.
Now A: The volume of an n-dimensional hyper-sphere is proportional to r^n. I have no idea what the formula is, but it will be some constant K (probably involving pi) times r^n. So again, the fractional volume within s of the surface is ((D/2)^n - (D/2 -s)^n) / (D/2)^n = (1-(D/(D-2s))^n). Same formula, same result, 86.7%.
I used a calculator.
Did I screw up or is this right?
737. BunEBear - March 24, 1999 - 8:13 AM PT
BobA: What you forgot in your treatment of the cube is that the sides overlap, so you can't just sum up the volume of the sides.
738. BobaFett - March 24, 1999 - 8:15 AM PT
BunEBear:
I already acknowledged that several times. But I disregarded it because the overlap would be fairly insignificant.
Were I to do the job correctly, I would just calculate the hypervolume of a cube with a side of 99 and subtract it from the hypercube of side 100.
739. BunEBear - March 24, 1999 - 8:25 AM PT
Oop's, I seem to have typed both formulas wrong (though I think I did the calculation right). They should be:
A: (1 - (L-2s)/L)^n)
B: (1-((D-2s)/D)^n)
"Were I to do the job correctly, I would just calculate the hypervolume of a cube with a side of 99 and subtract it from the hypercube of side 100."
Right, this is just what I did in Message #736
740. BunEBear - March 24, 1999 - 8:26 AM PT
Or rather you would subtract the volume of a cube with side 98.
741. FreetoChoose - March 24, 1999 - 8:41 AM PT
BunEBear
Correct.
It's mildly interesting in some contexts that almost all of the volume is within one per cent of the surface. Not immediately intuitive, although it becomes obvious as one begins to imagine a 100-dimensional hypersphere. (Not that I can, I get to 6 dimensions and run out of steam; but I can see the trend.)
742. BobaFett - March 24, 1999 - 8:45 AM PT
You can visualize higher dimensions?
743. msivorytower - March 24, 1999 - 8:50 AM PT
FTC reMessage #734
Ah, well, here's some more specificity for you, given your lack of familiarity with log functions.
Consider how one would interpret the partial coefficients of a regression when the dependent is in log form, but the independents are in unit forms. This is known as a semi-log function (semi-logrithmic, too).
This is a simple math problem, isn't it? Just like your simple one above.
744. FreetoChoose - March 24, 1999 - 9:04 AM PT
msivorytower
Who said I was unfamiliar with log functions?
So I take it you meant to refer to a regression where one of the variables has undergone a log transform? Do economists (or any other scientists) call that a semi-logarithmic function? I've never heard the term.
The term “regression” is well-known. So is logarithm, and semi-logarithm. And I know what you mean when you state that one of the variables is in log form, but I contend you are using rather obscure jargon when you call that a “semi-logrithmic function”.
My goal was to create math puzzles that don't require knowledge of jargon known only to practitioners of a few narrow fields. I'm guessing that you have a different goal, but I could be wrong.
745. FreetoChoose - March 24, 1999 - 9:11 AM PT
msivorytower
But, roughly speaking, when doing a semi-log regression, the coefficients represent the percentage increase in the dependent variable corresponding to the unit increase in the independent variable.
746. msivorytower - March 24, 1999 - 9:14 AM PT
Huh? Excuse me? Obscure?
I don't think so. Logrithm transformation, logrithmic transformation, semi-log, semi-logrithm, semi-logrithmic, etc.
What's so obscure about the latter term?
And, of course, just everyone who's taken math knows what a hypersphere is. Everyone. No obscurity there.
And just everyone walks around with the formula for computing the volume of a cube tucked in their little ole' heads. Everyone. Why I bet millions and millions of people who took lots and lots of math in college use these terms every day. Yep, every day.
747. msivorytower - March 24, 1999 - 9:15 AM PT
Hahaha, and I bet you cheated on that one.....
748. FreetoChoose - March 24, 1999 - 10:19 AM PT
Well, it's spelled logArithm, for starters.
749. FreetoChoose - March 24, 1999 - 10:20 AM PT
and how many people do NOT know the formula for the volume of a cube?
750. FreetoChoose - March 24, 1999 - 10:27 AM PT
But I grant that your question meets two of the criteria that Jade's failed; the answer is not easily found with a simple Internet search, and there are no artificial and unstated restrictions. Plus, it was easy to dance to. I'll give it an 8.
751. msivorytower - March 24, 1999 - 10:28 AM PT
Well, take the spelling to the bank, FTC.
And I used to know it when I needed it, but I can tell you I haven't needed to know the formula for the volume of a cube (by memory) for more years than I care to remember.
Now, my little commentary here was simply in reaction to your opening line in Message #727.
Had you said,
"Here's an interesting little math puzzle for all you math affectionados out there"
I'd not have said a word. But your implication that this was a simple problem, one that "oh, everyone, ought to be able to answer" was absurd.
I'd bet not more than 10 posters here remember the formula for the volume of a cube without referencing it, or know what a hypersphere is, if they ever knew it to begin with. That is, unless they are undergraduates taking math classes anyway.
752. FreetoChoose - March 24, 1999 - 10:50 AM PT
I'll be stunned if less than 80% of posters know the formula for the volume of a cube by memory. The length cubed. What could be simpler? Volume of a sphere is a bit tougher to recall off the top of ones head.
My reference to it being a simple problem is that it required very little calculation. Dividing 98 by 100 is a calculation most people can do in the heads. Raising the result to a power requires a calculator, but that's all it took to answer the question.
My intention was to start with a simple question, then try a harder one. I didn't want some one to think I felt the problem would stump many people.
BTW, I'm heading to Philadelphia in minutes; I hope others will pose puzzlers while I'm gone, and suffer me if I post some when I return.
753. msivorytower - March 24, 1999 - 10:55 AM PT
Okay posters
Stand up and be counted. How many of you know the formula for the volume of a cube by memory?
754. BobaFett - March 24, 1999 - 10:58 AM PT
Well, I do.
755. msivorytower - March 24, 1999 - 11:00 AM PT
oh goodie.
That's one.
We have about 50 more posters to go so that FTC won't be shocked and stunned.
756. BunEBear - March 24, 1999 - 11:14 AM PT
I gather that there was some more political context in which this question was first posed by FTC. Would anyone care to bring me up to speed?
757. FreetoChoose - March 24, 1999 - 11:32 AM PT
BunEBear
Well, Boba challenged Jade (or was it the other way around) to a math contest in Politics. While the purpose there was simply one-upmanship, I thought it might be fun to post some puzzlers. Apparently some people cannot leave acrimony behind. I'm trying hard to encourage posting of such puzzles here, as it makes more sense than in the Politics thread, but I don't want to muddy this place up with the baggage of other threads. We'll see what happens.
758. shaweagle - March 24, 1999 - 11:54 AM PT
Someone posting doesn't know that:
if L = length of a side of a cube,
then volume V = L x L x L !!
759. BobaFett - March 24, 1999 - 11:56 AM PT
FTC:
Jade challenged me. I would never stoop to such a childish taunt.
And no, in this one case I am not being sarcastic.
Jade is, however, very uneducated in science and math as she's proven to me on numerous occasions. So when she said, "Do you really want to get into a math contest," of course I had to say yes, just for the giggles which would inevitably ensue.
Incidentally, she challenged me in the first place because I asked her a simple question: "Since you've brought up the spherical triangle question, could you now explain the answer?" Of course, she couldn't explain it, since it was just something she saw on line. And thus she challenged me.
760. TheDiva - March 24, 1999 - 12:45 PM PT
Slackjaw
Thanks for providing the links to info on game theory. After I do some reading, and a considerable amount of lurking, I may actually have a question or two. In the meantime, I'll try to figure out what connection this has with baseball, because it seems to me that it does.
761. chloel - March 24, 1999 - 5:14 PM PT
MsIT
I am in deep pain at the thought that you don't know how to find the volume of a cube... how do you order cement and topsoil? Heck, how do you figure out how many 9"x9" brownie recipes will fill your extra-deep bread pan?
But, putting away these childish things, let me remark to anyone who is now interested in hyperspheres that this particular question carefully ignores their most lively behavior. Cf. CRC Concise Encyclopedia of Math; this site that hops on and off line, so let me excerpt for those of you out of phase:
"Strangely enough, the hyper-Surface Area and Content reach Maxima and then decrease towards 0 as n increases.
...
, the 7-D and 5-D hyperspheres have Maximal hyper-Surface Area and Content, respectively (Le Lionnais 1983)."
After considerable truck with equations, the formula for the volume of a n-dimensional hypersphere of radius R is given as
( pi^(n/2)* R^n ) / ( G( 1 + n/2 ))
( G for the gamma function, which fortunately cancels out in the question we're actually answering. Some of you will be deducing that
G(5/2) = 3/4 sqrt(pi)
.)
I disapprove hugely of 'rules' to not look things up, since
a) entirely too many unsubstantiated hypotheses are slung about here already,
b) I hate people to miss interesting related facts, as above, and
c) IMO, the problem should be interesting, not the answer.
762. Msivorytower - March 24, 1999 - 8:32 PM PT
Chloel
"I am in deep pain at the thought that you don't know how to find the volume of a cube... how do you order cement and topsoil? Heck, how do you figure out how many 9"x9" brownie recipes will fill your extra-deep bread pan?"
I laughed long and hard at this. I've never ordered cement or topsoil in my life, and I don't ever intend to. And, much as I'm sure it will cause you more pain, I hate brownies, and so have never made them. Otherwise, I simply follow the directions for any recipe I encounter.
You see? No need for remembering those endless formulas I had to know in college. Now, ask me about applied math in statistics, and I might be able to hold my own.
763. chloel - March 24, 1999 - 8:39 PM PT
But what do you do when you have the wrong-sized pan and all your measuring-implements are in the wrong system? (not make the recipe, I suppose. But really, I know you must actually know the equations for finding volumes - there are geometric motivations for stats, which is how I learned what I know of it - you just purport not to know how they're connected to material existence. And that causes me pain because it is Beautiful. Brownies, however, I have no great brief for.)
764. Msivorytower - March 24, 1999 - 8:44 PM PT
Hahahaha
Actually, Chloel, I have the capacity to remember when I need it. Besides which, I have a shelf full of math texts that I simply reference when I need to refresh my memory.
My objection with FTC was that he implied that just everyone would remember abstract formulas and be able to pull them out of the hat regardless of how obscure and pedantic the question.
765. uzmakk - March 25, 1999 - 10:24 AM PT
Lurk.
766. AzureNW - March 25, 1999 - 3:37 PM PT
This thread is wonderful, Slackjaw. It's like a peaceful garden. I've been reading far too much history lately, not at all peaceful. And that after reading about nuclear weapons for months.
767. AzureNW - March 25, 1999 - 3:37 PM PT
Thanks for the links, PE.
768. Slackjaw - March 25, 1999 - 10:44 PM PT
collective action, in action
Imagine that you are visiting a beach resort. The guests spend lots of time at the beach, and get sand all over their belongings. They can either take care to brush off their belongings before entering their rooms, which is a gigantic pain, or they can simply wash off the stuff in their rooms. If enough people do this, however, it is very bad for the pipes, and the exorbitant plumbing repairs will end up folded into the cost of stay.
One can think of this as a stylized collective action dilemma, where the common resource is the plumbing, and the strategies are "brush off" and "rinse off." As usual, regardless of what anyone does, "rinse off" is best for any one guest. Given the choices of the other guests, the choice of an individual guest will not matter for the pipes. As brushing off is more costly, guess what everyone does? So the sand wreaks havoc on the pipes, and rooms are more expensive than they would be if everyone brushed off.
Per Ostrom, some of the reasons for this collective action failure are clear. For example, the participants generally do not interact with each other repeatedly. Even people who return year after year--and there are some--know they alone cannot as one group alter the outcome. Furthermore, if there was any agreement to brush off, defection is very difficult to detect. People mostly come and go as they choose, so monitoring is limited. (Of course, one might argue that this is because there is no payoff if there is any monitoring--there is no repeated interaction.)
769. CalGal - March 25, 1999 - 10:53 PM PT
Slack,
Is game theory ever used to identify these sorts of problems in advance, so that the "system" (used loosely) can be designed to avoid them? (or perhaps that's its whole purpose?)
770. Slackjaw - March 25, 1999 - 11:28 PM PT
indeed it is, CalGal--great point. Loosely speaking, game theory is about investigating the consequences of various institutions, rules for conducting human interaction (be they formal or informal), etc.
The issue you have raised is the province of something called either implementation theory, mechanism design theory, or the theory of incentives, depending on your continent and intellectual bent. The idea here is to identify some consequence or family of consequences (e.g., efficient outcomes, in the sense I mentioned above**), and then come up with a game that leads to, or implements, some member of that family as its outcome. This is one of my major fields. Economists consider it part of information economics for reasons we'll talk about later, and it obviously draws very heavily on game theory.
So, implementation theory makes an engineering question (of sorts) out of game theory. How can we organize interaction to make some particular thing happen? Game theory asks, if interaction is organized in some particular way, what will happen?
** Once again, when I say that an outcome is efficient, I mean it in this sense: there is no other possible outcome that makes everyone better off without simultaneously making someone else worse off. (If you play around with it for a while, you will see that the oft-cited notion of efficiency as production of something at lowest cost is a special case of this notion.)
771. Slackjaw - March 25, 1999 - 11:56 PM PT
CalGal earlier raised the point that nature of the collective dilemma in the prisoners' dilemma is fairly clear when a lot of people are involved (as in the resort), but it's less clear when just two people are involved. As she said in Message #713, when it's just two people, why can't they both just see that 'don't confess' (or 'cooperate' in the context of the collective action problem) is best for both people?
This is a very common observation about the 2 person pd, and is essentially why so much ink has been spilled over this very simple toy game in various social science literatures.
From the collective point of view, both choosing to cooperate (with each other, NOT with the DA in the context of the prisoner story!) IS best, but individuals do not, by assumption, make their decisions from the collective point of view.
From an individual's perspective, which is the best outcome? Of course, it's when you choose 'don't cooperate' and the other player chooses 'cooperate.' That way, you don't have to pay the cost of coopearating, say brushing the sand of your things, but you enjoy the benefits of the collective good, clean pipes. (Thus I am thinking of a 2 person resort game, in which both people choosing 'rinse' destroys the pipes, but one person choosing 'brush' preserves them. Also for the sake of illustration imagine that the bills are determined after the vacationers leave and pipe damage is assessed. No other vacationers ever stay at this resort.)
If it was somehow self evident that player 1 was to rinse, and player 2 to brush, player 2 would not be very happy with this situation. S/he'd want to change to 'rinse.' Of course, if it was self evident that both were to brush, player 2 again would wish to deviate to 'rinse.' This is the essence of domination.
(cont.)
772. Slackjaw - March 25, 1999 - 11:57 PM PT
The strategy choices in the pd are defect (d) and cooperate (c), where defect is the same as 'confess.' Letting (x,y) denote the aggregate strategy or 'strategy profile' in which 1 chooses x and 2 chooses y, player 1's preferences are like so: (d,c) preferred to (c,c) preferred to (d,d) preferred to (c,d). 2's prefs are (c,d) preferred to (c,c) preferred to (d,d) preferred to (d,c).
The weird thing is that (c,c) is unanimously preferred to (d,d), yet (d,d) is the solution, in dominant strategies no less. I.e., the outcome is inefficient. But once again, while (c,c) is best from the group pov, no single player has that point of view in this game.
Now it might be charged that this is a shortcoming with game theory, that people do sometimes care about 'the group' or the payoffs of another player. If you hate your brother in law, you may not only wish to win trivial pursuit yourself, but to see him lose in spectacular fashion.
Aside from the problem of defining the 'welfare' of 'the group' and thus understanding what it means to give a damn about it, it is clear that the possibility that you might care about another's payoff is in fact not at all a limitation of game theory.
The reason is that the effect on *your own* utility when another player receives a certain level of utility is *already considered* in your preferences over the outcomes, or your utility function. Otherwise, the game is mis-specified. When the effect of payoffs to another player on your own payoffs is considered, of course you have a different game. That's the one that should be analyzed. But this doesn't say a whole lot about the adequacy of game theoretic techniques for solving games. It just drives home that point that you have to bring the techniques to bear on the right game--that trivial pursuit between brothers in law may be a bit more complicated than just a zero sum game with the obvious payoffs.
773. CalGal - March 26, 1999 - 12:24 AM PT
Slack,
I understood all this, I think--but could you define "solution" as you mean it here?
"The weird thing is that (c,c) is unanimously preferred to (d,d), yet (d,d) is the solution, in dominant strategies no less. "
The solution is that they should both confess, you are saying? At which point I realized I wasn't quite clear what the stated problem or question is.
What is the question that (d, d) answers? Is it that in designing systems, this is the behavior that should be expected and planned for?
774. Slackjaw - March 26, 1999 - 12:35 AM PT
game theory jeopardy
answer: (d,d)
question: what is the solution to the prisoners' dilemma?
so yeah, in the 1-shot pd, both confess is the solution. Collective action problem.
I think I may have confused you by posting that after the post on implementation theory. This stuff on the solution to the 1-shot pd is straight ahead, good ol' fashioned game theory. However...
"Is it that in designing systems, this is the behavior that should be expected and planned for?"
is essentially right, as the starting point from an *implementation* point of view. IOW, if you want (c,c) to be the outcome, you need a different game. How can you devise a game that gives each player the incentive to consider the perspective of the group, a perspective from which, as you pointed out, (c,c) is best? In a sense, this was one of the very first questions ever asked (and answered) in the theory of incentives.
game theory: here's the game; what's the solution?
implementation theory: here's the solution; what's the game?
(more or less)
775. CalGal - March 26, 1999 - 1:30 AM PT
Slack,
No, you didn't confuse me. I was very interested in your posts on implementation theory, and was writing up an example and question. Then you answered my previous question and I did understand the difference, but I still don't understand the definition of "solution" when looking at it from a "game theory" pov.
I think--and that's what my next posts will be about--that my confusion is because I am naturally inclined to implementation theory.
The two person dilemma confuses me, in a sense, because it *isn't* a system, so I am not sure what the hell one is trying to solve.
The "solution" from the game theory standpoint, is to figure out what both will do? And the fact that it isn't the best outcome is ascribed to the "collective action problem"?
776. CalGal - March 26, 1999 - 2:13 AM PT
Slack,
An example in management system design (customer service, specifically):
The Customer Service Reps responsible for entering problems are not incented to ensure that the data provided is accurate and complete, but on how many problems they enter into the system and what percentage they resolve. They aren't penalized for inaccurate problem description.
But two other groups in the process--management and those responsible for fixing the problems one degree removed--need accurate data. Management needs it because of their incentives--reporting their efficiency , for information about the types of problems, for improving their product, etc. Second level assignment needs it because they are incented on how many problems they resolve, and since they aren't in direct contact with the customer, the better info they get, the more quickly they can resolve the problem.
Of course, management wants 27 required data elements entered for each problem, so that they can run every report known to man. And the second level assignment wants accurate data, too, and lots of it--and they are higher status than CSRs.
But power and status are irrelevant compared to the CSR incentive system--that as long as there is no penalty for inaccurate problem description and plenty penalty for low call volume, they will blitz by those 27 fields and the data will be meaningless.
Overall, everyone will benefit from clean data. Accurate reports will allow management to justify more people, better intelligence will eventually allow CSRs to resolve more problems and increase their call counts.
But incented on call counts, are they always going to spend the time determining the accurate information, which may drastically increase their time per call? Or are they going to enter whatever the hell gets the ticket entered quickly, using the fields that are meaningful to them and faking the rest?
777. CalGal - March 26, 1999 - 2:16 AM PT
Of course, some of them might enter data accurately and completely. And then the question becomes, what is the "solution" to the "game"? Clean, but limited, data, or inconsistently accurate data, but lots of it?
Even managers know the answer to that one.
So then one has to go back and redefine the game to redesign the plumbing or incent the CSRs to shake the sand off before they get to the shower.
This problem occurs in any system design where there is a significant disconnect between the incentives of different players.
I doubt the specifics will be of interest to anyone other than Diva, who may be lurking, and I apologize for the lengthy posts. But I wanted to describe it to you because to me, this is the same as the shower and sand scenario. And the reason I asked the question in Message #768 is because as I read your post, I was automatically redefining the game, rather than wondering about the solution.
Incentives are not commonly discussed in systems design in my field--at least that I can tell. Requirements are Our God, and efficiency is only considered in terms of automation. When I bring up these considerations, people are usually convinced easily, but it doesn't seem to be a factor that is generally at play.
When I say they aren't commonly discussed, it's worth remembering I have no formal training. But I've worked in the area for a long time, with people who are formally trained, and it is never discussed.
So I am wondering now if it is something that is taught in CompSci courses and I just don't know about it and neither does anyone else I've worked with in 12 years. Or if the apps with this disconnect between data entry and analysis are rare enough that it hasn't been studied yet. (These considerations wouldn't apply in most line of business apps, for example.)
778. uzmakk - March 26, 1999 - 6:21 AM PT
And ofcourse any player's decision making capacity is dependent upon the information he is capable of receiving. Information is a structured commodity needing a structured distribution system. What should the design of that system be? Shall we leave it to the free market? to human desire? to the hereditary royalty? inertia? linear progression? eggheads? philosophers? magicians? or priests? the desires of the masses? or the desires of their rulers? Things are as they are because of the constraints put on the flow of information. A rose is a rose because information is constrained in a manner to make it become a rose and nothing else. A B-2 bomber is a B-2 bomber because information is constrained in a manner to bring such a thing into existence. Change the way information flows and the way it is constrained and new things will come into existence. I am going to write to Jackson and Perkins and suggest that they name their next rose the B-2.
779. chloel - March 26, 1999 - 1:14 PM PT
Slack - on Message #770, last para, it's "everyone better off", not "anyone better off"? Is there another name for the latter definition?
CalGal -
one of the basic books on testing software (Kaner, Falk, Ngyuen) does discuss different measures of bug-creation and bug-finding. There's a Dilbert cartoon in which programmers get paid for every bug they fix - *in their own code* - which is said to be from real life.
In internal testing, my favorite counter is to keep track of how many of a tester's bugs made it past triage to be actually fixed. I don't know of a good measure of programmer productivity that takes both code complexity and bug complexity into account (and doesn't require incredible crunching to measure).
780. Slackjaw - March 26, 1999 - 11:05 PM PT
chloel: yes, there is a name for the notion you mentioned--weak efficiency. It's closely related to the one I mentioned, strict or strong efficiency (strict/strong in the sense that it's never easier to satisfy than weak, and obviously the strong notion implies the weak one). For many classes of problems considered in economics, the two notions actually coincide.
781. Slackjaw - March 26, 1999 - 11:05 PM PT
CalGal: hey, let's have no apologies for lengthy posts, alright?
'The two person dilemma confuses me, in a sense, because it *isn't* a system' Well, this depends on your definition of 'system,' but imo the rules of the 2 person pd characterize a system just as much as the n-person pd does. They describe one system for the maintanence of a collective good among the players--though not a terribly effective one.
'The "solution" from the game theory standpoint, is to figure out what both will do?' Yes, exactly.
'And the fact that it isn't the best outcome is ascribed to the "collective action problem"?' In this case it is. Let me make clear that there are many sources of inefficiency in game solutions. Collective action problems are but one particular kind that have proven especially interesting.
question--do the higher ups have the ability to contribute to the maintenance of sound data, or is the power to clean it up solely in the hands of the CSRs? If the latter, then this is certainly an incentive problem: the higher ups benefit from certain activities of the reps, but do not directly and personally bear any cost of those activities. This is called an externality.
It is closely related to a collective action problem, but is not exactly the same thing. The reason is that a collective action problem involves a sort of 2-sided externality--each party's strategy has some direct effect on the status of the collective good from which they all benefit, like clean pipes at the resort. In the data integrity issue, it's a "1 sided" externality: one party basically has all the ability to maintain a resource all parties share in common, clean data, but about which some parties (higher ups) care more than others (CSRs).
(cont.)
782. Slackjaw - March 26, 1999 - 11:06 PM PT
Of course, there *may be* a sort of meta-collective action problem in this sense: there appear to be gains from trade between the parties that are not realized. Higher ups want clean data and presumably will give something up in exchange for it, and CSRs will provide clean data in exchange for making it worth their while. Evidently this mutually beneficial trade (assuming both sides would willingly engage) is not taking place, so the trading institution--a resource the parties share in common and from which they all benefit--is not being maintained.
Whether this counts as collective action failure depends on whether the integrity of the data could ever be checked. After all, maintaining quality data is still costly for CSRs, and it appears that only they know how high the quality is. And what could be better for a CSR than to get a bonus on the pretense that quality data is being input, but still keep shoddy data? You get the bonus, don't pay the cost of quality record keeping, and nobody is the wiser. This doesn't count as a collective action problem, because the source of the inefficiency is not something that provides benefit to all but it costly for each to maintain. It is a problem, but of a different sort.
Before anyone gets his hopes up about implementation theory, I should point out that the results are not always optimistic. At the most general level imaginable, there are serious limitations on our ability to implement certain types of outcomes. We'll get to that later.
**corrigendum: collective action problems involve a MULTI-sided externality.
783. Slackjaw - March 26, 1999 - 11:07 PM PT
Uzmakk's Message #778 is worth keeping in mind, particularly the part about information processing capabilities. When discussing standard game theory, we do not necessarily have to assume selfish preferences, but we do assume that people are aware of the game they are playing and can solve it as well as any game theorist. The fact that anyone needs to take a course to learn the subject itself falsifies the premise.
However, the point is to get a handle on the basic intuition in some problem or situation. Assuming hyper-intelligence certainly has limitations in this regard insofar as real people are interacting in these situations, but it can still be a useful guide--and compared to alternatives, it is MUCH simpler!
784. Slackjaw - March 26, 1999 - 11:10 PM PT
As for distribution of information, Uzmakk asks whether the free market is a suitable mechanism. If my grandiose plans for this thread are realized, we shall discuss instances where free market provision of information leads to total failure of any transmission at all, and instances where information transmission is so basic and essential to the problem that setting up a market is superfluous.
How information and its asymmetric dispersion affects the properties (e.g., efficiency) of trading institutions, contractual relationships, etc. is the province of information economics.
785. Slackjaw - March 26, 1999 - 11:21 PM PT
I wish to regress for a moment to zero-sum games, for a little piece of trivia.
Turns out there is a rather simple theorem on the subject, which was one of the first theorems ever proved in game theory. It was proved by the mathemetician E. Zermelo in the '20s. The subject is 2 person zero sum games of perfect information. So, for every combination of strategies, the sum of the payoffs to each player is 0, and if the game is played over time, each player knows everything that has happened up to any given point.
Examples are chess, checkers, tic-tac-toe, connect 4, etc., where each player only cares about winning the game.
For games of this sort, exactly one of the following is true: either player 1 can guarantee a win, player 2 can guarantee a win, or each player can guarantee a draw.
So, from a mathematical point of view, chess is 'solved.' Indeed, it is a trivial game mathematically--not qualitatively different from tic tac toe, which has a known solution in which X can win, provided that O chooses a 'best response' (an idea to be made clear shortly) to each move of X. Of course, for rational players, the restriction that O chooses a best response is not a limitation--it is by definition O's utility maximizing move, given what X has done.
Trouble about chess is that computationally, it is rather mindbending. As I posted once before, rumor has it that there are more possible games of chess than there are molecules in the UNIVERSE (but nobody knows exactly how many possible games of chess there are). So, while chess has a solution in the above sense, nobody knows what it is. (Checkers and connect 4 have each been solved.)
786. Slackjaw - March 26, 1999 - 11:22 PM PT
fwiw, the discussion of zero sum games begins at Message #688.
787. Slackjaw - March 26, 1999 - 11:47 PM PT
plan:
so far we have talked about zero sum games and dominant strategy solutions in 1-shot games.
But, only special games have solutions in dominant strategies. Zero sum games--situations of complete, unbridled conflict--are also very special. If game theorists want to make pronouncements or predictions about how a wider class of games will be played, we must introduce some more machinery, and more restrictive assumptions.
Also, 1-shot games, in which the players simultaneously choose their strategies, are very useful, but sometimes interesting things can happen if games take place in stages. When they do, it's called a game in 'extensive form,' as opposed to 1-shot games, which are in 'strategic' or 'normal form.'
So, I will next discuss the special case of the prisoners' dilemma (in extensive form) repeated a finite number of times by the same players, and then I will move to solution concepts that have more bite than domination but can be used in many more games.
This should not stop anyone from talking more about collective action--that's just the new stuff I intend to say in the near future.
788. CalGal - March 27, 1999 - 12:35 AM PT
Slack,
Here's an archive of the discussion thus far.
"They describe one system for the maintanence of a collective good among the players--though not a terribly effective one."
Yes, but when I think of two people, I think of the ability they have to communicate and plan. Two people working in a situation like this would have a strategy--a "system", in fact. Whereas when it is a whole group of people, the result of their collective actions, with no planning, is a "system"--which is what game theory is trying to solve for.
I'm just explaining my furrowed brow resistance, btw. I'll accept that they are to be considered the same thing.
"Let me make clear that there are many sources of inefficiency in game solutions. "
Yes, I understood that. I meant that in *this* case the less than desirable results were due to the collective action problem. Cool.
The first paragraph of Message #782 is the correct description, I think. Management in *this* case isn't an externality. (The customers served are an externality, though, and a whole different problem in and of themselves).
789. CalGal - March 27, 1999 - 12:38 AM PT
I wasn't asking if this was a collective action problem per se. What I am asking is if my impression is correct--that these particular kinds of systems are presenting similar sorts of problems that you analyze with game or implementation theory.
The reason I am asking is because management systems processes have not advanced significantly in nearly 20 years, and aren't expected to any time soon. The problems are still the same, the success rate is still low, the number of companies with barely functioning processes is about the same. Yet the applications technology has advanced on a par with all other technology in this area.
It's always been clear to many of us in this field that the problems lie in the fact that these applications (call them infrastructure, management, or process systems) are a different breed from line of business apps or end-user apps. Which is why a while ago I started framing the problems in terms of incentives, redefining the goal (or changing the game), and explaining why the most efficient outcome was (d,d). (hey, I could get very into these buzzwords.)
It's not so much that I'm hoping that implementation theory will help solve these problems as I am wondering if it can help provide a construct for why the problems exist. After 20 years it's pretty clear that there *are* serious limitations on the ability to implement certain outcomes, and there are good reasons why.
But this is all very unstructured mumblings right now--it just seemed like an interesting similarity. Are there any books on implementation theory, as opposed to game theory?
790. uzmakk - March 27, 1999 - 8:35 AM PT
Slackjaw: Message #784
I am sure we will all do our best, Slackjaw, to realize your grandioze plans. After all, pure mathematics is almost the only free field of science left.
791. Slackjaw - March 27, 1999 - 3:02 PM PT
"when I think of two people, I think of the ability they have to communicate and plan." oh, wonderful. This raises a very important point in noncooperative game theory: players cannot commit to anything a priori, sign binding agreements, or anything. If there is communication and planning, it must be taken into account in the structure of the game itself, and when it comes time to carry out their plans, players must actually wish to do so.
For example, if we play the prisoners' dilemma, and I assure you and promise you that I will cooperate with you, I *still* have an incentive to defect once it comes time to play the pd.
Naturally enough, this is particularly important in extensive form games, by the way.
"What I am asking is if my impression is correct--that these particular kinds of systems are presenting similar sorts of problems that you analyze with game or implementation theory." Aha. Yes.
"I am wondering if it can help provide a construct for why the problems exist." Aha. Again, yes. An important element is verifiability--can managers verify whether the data is high or low quality? I mean, the CSRs can always claim to be entering clean data, but still doing a shoddy job. Only they know which is which. This sort of information asymmetry always spells trouble for the implementation of efficient outcomes.
792. Slackjaw - March 27, 1999 - 3:03 PM PT
As for books, sadly there is no good general treatment of implementation or the theory of incentives at an introductory level. Certainly not at the level of _Thinking Strategically_. There is a book called _Incentives_ by a fellow named Don Campbell, which uses some elementary mathematics--a review of which appears in the book. Unfortunately, the book does not treat most of the richer and more interesting results from implementation. There is a book called _Games, Strategies and Managers_ by John McMillan that covers both game theory and some implementation theory (principal-agent theory specifically) at an introductory level--no real math at all. But it is not very general either, wrt implementation. There are some other excellent books on implementation/mechanism design, but they are highly advanced.
But that's just the theory of incentives as a game theorist understands it. Elinor Ostrom's book mentioned above is substantially less formal, more eclectic, more applied, not at all concerned with implementation theory per se, but shouldn't be overlooked if you want to read about incentives in real world settings. Again, it's a most accessible book.
Nice archive, by the way. Thanks for assembling it.
793. taust1 - March 28, 1999 - 2:06 AM PT
CalGal #789
WRT management systems I think we may have not only a problem of incentives, but of a sufficient understanding of the system they are in on the part of the participants. I'm a fan of Beer (Brain of the Firm etc etc.) He has developed a desciption of the management system which at the very least gives some insight into the system. You cannot get a discussion because there is no common description of the managemment system that you can start from. Its all pop phsychology standard solutions.
794. uzmakk - March 28, 1999 - 7:56 AM PT
Taust, Message #793:
Someone who is familiar with Stafford Beer. I am grateful for your presence. I read Platform for Change 1972 version many many years ago. I understnd that there is a 1995 edition. Have you read it?
795. uzmakk - March 28, 1999 - 7:59 AM PT
Something that seems related and is apparently all the rage in management circles is The Fifth Discipline, Peter Senge (The Art and Practice of the Learning Organization) MIT. Haven't read much of it. Pop psychology? I don't know.
796. pellenilsson - March 28, 1999 - 11:01 AM PT
Slackjaw
You remember a brief discussion about Elinor Ostrom. She was awarded the Skytte prize in political science by the Uppsala University (Johan Skytte, 1577-1645, statesman, political scientist, large donations to the university). The prize is not a Nobel but 40,000 US$ will always come in handy. There was a very brief interview with here in one of the morning papers.
Q. How does your research differ from that of others?
A. Perhaps in the point of departure. During my studies in Germany I came to know about game theory which deals with the possibility of rational decisions. The prevailing opinion then was that individuals should not be allowed to make their own choices because the collective is not best served if individuals do what they think right. But when it concerns the management of water resources I think the opposite is true.
Q. Have your theories been tested in practice?
A. Not really. But some governments have expressed interest and I have started a collaboration effort with the World Bank.
I don't want to disturb your plan for the thread. But if you can fit in Ostrom I would be glad. The subject interests me very much.
797. Slackjaw - March 28, 1999 - 10:10 PM PT
"...if you can fit in Ostrom..." In the immortal words of one Noah Cross, "oh my, yes." (anyone besides CalGal know that movie?)
Though this has been said several times already, it bears repeating for you in particular. Get her book, _Governing the Commons_. It really is accessible and a great read if you are interested. There are many case study details that can't really be efficiently presented in this format. (Though she says her theories have never been tested, she derived most of her insights from case study material, rather than formal deductions as a game theorist would.)
There is no Nobel for political science, but if there was she'd be one of a dozen or so people in the hopper to get one of the first few medals (especially considering the prize cannot be awarded posthumously).
798. pellenilsson - March 28, 1999 - 10:22 PM PT
Slackjaw
I have just now ordered the book.
799. taust1 - March 29, 1999 - 1:38 AM PT
uzmakk message #794 @795 My day job got in the way of checking my edition of 'Platform for Change' From experience with reading two different editions of one of his other books I don't recall he was a big reviser.
Where did that sweeping generalisation about management books etc come from in my previous message? Am I allowed to blame a consultant?
I'm only just getting into this thread and two books to read already.
I'll stop chatting and continue reading.
800. pellenilsson - March 29, 1999 - 3:18 AM PT
For anyone who reads or has access to The Economist there is an interesting article in the science section of the current issue. It concerns friendship and cooperation between stallions and how they get more offspring out of it compared to operating alone.