Re: [tlug] Open-source Japan

[Niels Kobschätzki <n.kobschaetzki@example.com>]

On Mar 5, 2009, at 3:14 AM, Stephen J. Turnbull wrote:

> Curt Sampson writes:
>
> > So the best outcome for the group is to be absolutely honest.  
> > However,
> > in a classic game theory problem, being absolutely honest will not
> > necessarially lead to the best outcome for the individual.
>
> No, we *know* it will not.  The used car market is an obvious case.
> If you've got a lemon, you want to pass it on to somebody.

Well, I disagree here. If everybody would say the truth on the market  
for lemons and everybody knew that they are saying the truth about the  
articles for sale *and* it would be trustworthy the lemon would sell.  
The buyers just don't know which car is how much worth because they do  
not have the information about it. If the sellers would say all the  
truth and would show the problems with the used cars every purchase  
could bring up a price according to his preferences and you would  
probably need an indifference condition for the seller if all buyers  
had the same preference-set which but that would lead to another  
problem because the buyer wouldn't necessarily loose utility if he was  
not selected (because he wouldn't loose any money) by a let's say  
random process and therefore he wouldn't calculate that into the price  
he's willing to pay.

And I don't know any game theory-situation where being absolutely  
honest wouldn't lead to a equilibrium. The problem usually is either  
that you do not know, how the partner acts or if he's trustworthy (and  
therefore you will need some kind of signalling). Even so everybody  
would be totally honest it would lead either to cooperation or a nash- 
equilibrium (while I say that not telling something in the prisoner's  
dilemma is not being not honest - please take care of the double  
negation here). If everybody would be honest and we knew that, we  
wouldn't need those signalling-models anymore. Or do I see anything  
wrong here?

Ok…that got a little bit out of hand here ;)

> > This is essentially an argument for regulation.
>
> Sure.  The market that economists analyze is an abstraction; it
> doesn't exist by itself in the real world.  For example, the issue of
> "announcing a price is a commitment to trade at that price".  All the
> theorems depend on that.  But to make that stick in the real world,
> there *must* be regulation.  Essentially, by definition.  (This is
> perfectly analogous to implementation of ADTs in a real language, I
> think.)
>
> The question is, does that need to be done by an external agency (==
> government or industry cooperative), or can the market be left "free"
> to do it itself?  As you say, the NYSE does a lot of self-regulation
> and is very successful because of it.  In general, the global
> financial markets self-regulate and mostly nobody is the worse for it.
>
> It's at the level of personal credit, insurance, etc that problems
> come in.

Well, in a market where self-regulation doesn't work but *is  
necessary* it should be done imho by an external agency. Otherwise how  
would we do something against monopolies/cartels? (only as an example;  
I doubt it that markets would find a way to by themself all the time  
to prevent those)
I think the main problem is how to decide before a market failure/too  
small market occurs, if and which regulation is necessary by which  
external agency.

<snip deliberate fraud should be regulated and inherent risk + fraud  
risk -> small market -> few resources for self-regulation -> rampant  
fraud>

Niels

Attachment: smime.p7s
Description: S/MIME cryptographic signature