NationStates Jolt Archive

I just read this the other day. It's about "Mechanism Design" theory, which the three guys won the Economics Nobel Prize for this year. I touched on it before, but I think this article does a good job explaining it.

http://www.economist.com/finance/displaystory.cfm?story_id=9988840
Intelligent design
A theory of an intelligently guided invisible hand wins the Nobel prize

[...]

Mechanism-design theory aims to give the invisible hand a helping hand, in particular by focusing on how to minimise the economic cost of “asymmetric information”—the problem of dealing with someone who knows more than you do. Trading efficiently under asymmetric information is hard, for how do you decide what price to offer someone for something—a product, say, or their labour—if you do not know at what price they would sell it? On the one hand, you may not offer enough to get them to deliver the product or work, or at least do so adequately; on the other, you may overpay, wasting resources that might have been better used elsewhere.

Mr Hurwicz took up economics at a time when debate was raging about the relative merits of central planning and the market mechanism. While agreeing with the great libertarian, Friedrich von Hayek, that the dispersion of information was at the heart of the failure of planning, Mr Hurwicz saw that it went deeper than that. He observed that there was a lack of incentive for people to share their information with the government truthfully. Moreover, although the market mechanism was far less afflicted than central planning by such incentive problems, it was by no means immune from them.

His big idea was “incentive compatibility”. The way to get as close as possible to the most efficient economic outcome is to design mechanisms in which everybody does best for themselves by sharing truthfully whatever private information they have that is asked for. Even this cannot guarantee an optimal outcome, Mr Hurwicz showed, because the existence of any private information precludes the economist's holy grail, known as Pareto efficiency, even if everyone's incentives are compatible. But it will get closer to it than if incentives are incompatible (ie, when some people can do better by not sharing information or lying). Pareto efficiency means that no one can be made better off without someone becoming worse off. Mechanism design has “incentive efficiency”: given compatible incentives, no one can do better without someone doing worse.

[...]

Basically, it's about creating the rules of the game that are such that everyone doing best for themselves also produces a pareto optimal (http://en.wikipedia.org/wiki/Pareto_optimal) outcome. In modern microeconomics (http://en.wikipedia.org/wiki/Microeconomics), game theory (http://en.wikipedia.org/wiki/Game_theory) and general equilibrium theory (http://en.wikipedia.org/wiki/General_equilibrium_theory) are coming together quite rapidly (particularly because the former may hold the key to how prices come about in the latter), so mechanism design theory ultimately may hold the key to determining regulations/rules that "fix" the market's issues with information asymmetry and other market failures.

This relies on mathematics, logic and observations and surveys aimed at finding people's behaviour when making decisions. As such the process is largely free of ideology, though the overall idea has a utilitarian slant to it.

So there is a utilitarian motivation, but also a respect for people's self-interest (or rather, the realisation that you can't get them consistently acting against it). So people's desires are channeled and directed, rather than surpressed, not by day-to-day government directives, but by the overall framework of the economy, which could in theory be set up in a constitution, never to be touched again.

So my question is: do you see this sort of thinking gaining ground? If we accept that governments can only be considered legitimate if there is at least a trace of utilitarian thought in our heads, would that mean there is some "perfect/optimal government", based on a mechanism designed according to cold, hard maths?

What does this mean in practical terms? What mechanisms would be put in place to solve the asymetry of information problem?

I shall read the article proper, but my initial worry is with game theory, which mechanism design, or at least your quoted part, seems to rely heavily on.

Game theory, I believe, is obviously flawed. Humans aren’t always-rational, self-interested and calculating; we don’t make rational choices all the time, nor do we take the (supposedly rational) choices of every other actor in the ‘game’ into account before every action.

John Nash himself has expressed that the beings in his game theory model have little or no semblance to ‘real life’ humans; that he has “over-emphasised rationality”.

I don’t think it’s possible, either through game theory or any other mathematical social theory, to accurately represent the chaotic and ultimately complex human society. Nor is it the case, as game theory implies, that human behaviour is entirely motivated by self-interest.

So any theory that relies upon, or incorporates, game theory into its workings must be flawed.

What does this mean in practical terms? What mechanisms would be put in place to solve the asymetry of information problem?
It's nowhere near a practical stage right now. It's in academic papers, people running simulations and huge games, trying to see the patterns of these rules and what they do. There have been some applications, but in the area of voting rather than economic policy.

On the day the prizes were announced Jim Lehrer had an interview with one of the guys, where they briefly talk about it: http://www.pbs.org/newshour/bb/business/july-dec07/game_10-15.html

No. Even on purely utilitarian terms, there is precious little we can "prove" in practical terms with "cold, hard maths"--there are generally too many necessary simplifying assumptions.

For just one example, economic "utility" is not the same as the "utility" of utilitarian theory. Economic utility is a matter of measuring our decisions; it is an indicator of what we will choose to do. Utilitarian theory, however, wants what will actually benefit us most... and it is by no means obvious that the two will coincide always, or even usually.

Not to mention the difficulty of accounting for, say, the exact value by which the $1,000 of a rich person is less "valuable" in utilitarian terms than the $1,000 of a poor person.

Game theory, I believe, is obviously flawed. Humans aren’t always-rational, self-interested and calculating; we don’t make rational choices all the time, nor do we take the (supposedly rational) choices of every other actor in the ‘game’ into account before every action.
No, but we can agree that they should, right? And if they don't, basically there are no good rules to be made and we're forever trapped in ad hoc policies, which in fact means policies based entirely on the preference of the person in power on the day, but more importantly means that we'll never optimise anything. We'd forever be wasting resources (which is the same as wasting potential happiness).

For just one example, economic "utility" is not the same as the "utility" of utilitarian theory. Economic utility is a matter of measuring our decisions; it is an indicator of what we will choose to do. Utilitarian theory, however, wants what will actually benefit us most... and it is by no means obvious that the two will coincide always, or even usually.
So why does Mill reckon we have no scope for forcing others to do something against their will?

I had a book of collected Bentham and Mill essays around here somewhere, but it got lost. I'll try to find it and educate myself, but I'm not sure the two ideas of utility are quite as different on a philosophical level as you make it seem.

Not to mention the difficulty of accounting for, say, the exact value by which the $1,000 of a rich person is less "valuable" in utilitarian terms than the $1,000 of a poor person.
Well, economics does have a concept of diminishing marginal utility, and you could account for it by seeing how long a rich person would be willing to work the same job to earn $1,000, and how that differs from a poor person.

But the market sorts these things out already, there wouldn't have to be any rules regarding that, if everyone has the information needed available to them. In fact, the market accounts for it all the time, and in many cases it is precisely that which its critics find unfair.

No, but we can agree that they should, right?
I may think they ought to, but 'ought' =/= 'is', and we can't make up policies pretending the two are the same.

And if they don't, basically there are no good rules to be made and we're forever trapped in ad hoc policies, which in fact means policies based entirely on the preference of the person in power on the day, but more importantly means that we'll never optimise anything.
I don't see any problem with not having 'hard', mathematical rules about human society. Pretending that there is leads to farcical, and potentially dangerous situations. Just take a look at the USSR.

We'd forever be wasting resources (which is the same as wasting potential happiness).
According to a rather strange view of happiness, IMO.

So why does Mill reckon we have no scope for forcing others to do something against their will?
Because he holds an incredibly strong belief in negative liberty.

I had a book of collected Bentham and Mill essays around here somewhere, but it got lost. I’ll try to find it and educate myself, but I’m not sure the two ideas of utility are quite as different on a philosophical level as you make it seem.
As an important aside, there are many forms of utilitarianism, all with a different conception of ‘utility’; be that happiness, pleasure-net-pain, desire satisfaction, etc. Then you get onto problems such as whether one is to maximise aggregate or average utility...

Suffice to say, there are a very many different philosophical conceptions of utility, some far different to ‘economic utility’.

So why does Mill reckon we have no scope for forcing others to do something against their will?

This is a very different question.

Utilitarianism was important to the "victimless crimes" sort of argument because it required actual "harm" for immorality--so, for instance, the Catholic notion of natural law would no longer serve as a good reason to prohibit sodomy. Mill used mostly practical arguments to extend this line of thinking into the rule.

But even if we accept the utilitarian argument for anti-paternalism--which I think is highly flawed--we might still be able to recognize problems with, say, advertisement insofar as it deliberately seeks to get us to do what will not benefit us for bad "reasons." And I don't see how "cold, hard maths" can get us out of that one.

I'll try to find it and educate myself, but I'm not sure the two ideas of utility are quite as different on a philosophical level as you make it seem.

Utilitarian theories started with happiness and since have mostly moved to preference satisfaction. They are about human mental states.

Economic "utility" starts with the decision and moves backwards, but not very far (at least not credibly). It is about human behavior.

But the market sorts these things out already, there wouldn't have to be any rules regarding that, if everyone has the information needed available to them.

The market (at least under certain other simplifying assumptions that are almost certainly false in many cases) results in efficient results... but it doesn't care one way or another about maximum utilitarian benefit.

But even if we accept the utilitarian argument for anti-paternalism--which I think is highly flawed--we might still be able to recognize problems with, say, advertisement insofar as it deliberately seeks to get us to do what will not benefit us for bad "reasons." And I don't see how "cold, hard maths" can get us out of that one.
I don't think I have to quote Mill to make the point - but advertising is an attempt to convince another person to buy a certain product. If it succeeds, you obviously think that buying the product will make you better off than not buying the product. In that situation, not allowing you to buy it would actually make you unhappy - in either form of utility.

Can you be of the opinion that I shouldn't have bought the product? Certainly.
Can the product be bad for me? Physically, yes - mentally I'm not so sure.

What if the advertising is for a new cancer drug, and before I saw it I didn't know this cancer drug existed? I now go and cure my cancer, but did I do it because I saw the ad, or because I wanted my cancer cured anyway? You can't tell, in many situations I couldn't have told either.

The economic concept of utility has developed because the idea of happiness being a good thing and people wanting to maximise it is not new. It's old-school utilitarianism. But what Bentham didn't realise and Mill just brushed upon is precisely the above problem: you can't measure happiness and so any policy you make is not based on anything but your opinion. And I know that Bentham would cringe at that sort of thing, I believe he was a huge fan of the cold, hard maths.

The two have the same basic idea at the core of it, the economic conception of utility just developed from the acceptance that the best bet we have of knowing what makes people happy is by looking at their behaviour. We can modify it by reasoning with them (even though our basis for it may not always be 100% firm itself), but what we do know for certain is that if we use force to stop the behaviour, we'll make them unhappy. So that's why economists started using revealed preferences and rankings of alternatives to get an indirect measure that can be used to implement utilitarianism.

It's not perfect, but no one can come up with a better one.

And I think the advertising example is interesting anyways, simply because it deals with information dissemination and would therefore almost certainly be of interest to the mechanism designer.

I don't think I have to quote Mill to make the point - but advertising is an attempt to convince another person to buy a certain product.

No, it isn't. It's an attempt to get another person to buy a product.

If the most efficient way to get someone to do this is through rational argument, then that will be done. But the most efficient way may be instead any variety of manipulation... preying on people's insecurities, associating the product with attractive images that in fact have nothing to do with its benefits, and so on.

If it succeeds, you obviously think that buying the product will make you better off than not buying the product.

No, human beings don't always, or even usually, do what we think is best for us.

We are too vulnerable to temptation... especially in a society where the major powers have massive incentives to encourage this trait in us in certain contexts.

In that situation, not allowing you to buy it would actually make you unhappy

Maybe temporarily. Ultimately? That's not clear at all.

Can the product be bad for me? Physically, yes - mentally I'm not so sure.

You should be. What about drug addicts?

What if the advertising is for a new cancer drug, and before I saw it I didn't know this cancer drug existed?

Great! Let's have information distributors. Let's just not put profit-seeking corporations interested in manipulation and distortion rather than objectivity and benefit in charge.

(Now, we can argue about the merits of this proposal... but I highly doubt we will ever have "cold, hard maths" to prove it one way or the other, chiefly because we can never have "cold, hard maths" that will account for the different kinds of "utility.")

You can't tell

No, but you can try to create a system where only the right factor will play a part.

The economic concept of utility has developed because the idea of happiness being a good thing and people wanting to maximise it is not new. It's old-school utilitarianism.

It's also either the fallacy of equivocation (between "utility" as in "actual benefit" and "utility" as in "I will actually make x decision rather than y decision") or a transparently false assumption that the two will always be the same.

But what Bentham didn't realise and Mill just brushed upon is precisely the above problem: you can't measure happiness and so any policy you make is not based on anything but your opinion.

Really? I think we can get a pretty good idea of what makes people happy and what doesn't... never perfect, by the very nature of subjective mental states, but a pretty good approximation.

But this is an empirical question, and a very difficult and ambiguous one at that. Not a matter for resolution in the "cold, hard maths."

The two have the same basic idea at the core of it,

Of course they do, but there is still that gap.

the economic conception of utility just developed from the acceptance that the best bet we have of knowing what makes people happy is by looking at their behaviour.

I'm not sure of that at all.

We can know a lot about three things:
1. What (generally) makes people happy and what (generally) doesn't.
2. The likely consequences of certain actions.
3. The kinds of reasons people do what is not good for them.

With that kind of knowledge, we certainly can minimize (3) and in some cases even point to certain specific circumstances where people almost certainly do what is not in their interest.

but what we do know for certain is that if we use force to stop the behaviour, we'll make them unhappy.

Actually, we don't.

A person, say, being pressured by others to take a risk he or she doesn't want to take may in fact welcome such intervention... it gives him or her an easy out.

Now, you may say that this is a rare case... but can the "cold, hard maths" tell you that? And that's the important point in all of this: since the "cold, hard maths" make simplifying assumptions that clearly don't always hold in the empirical world, there is always that element of ambiguity in actually putting them into practice.

I may think they ought to, but 'ought' =/= 'is', and we can't make up policies pretending the two are the same.
We can make them such that failing to do the ought will not be rewarded.

Anyways, my reason for it is not metaphysical, but fairly simple: if people don't act rationally, they'll use scarce resource in an inefficient way, with the result that there is wastage and people not being as well off as they could have been.

Now whether or not that is a good thing can be subject to further debate, but to be honest, I don't see much point in talking about it. I'm not a fan of the whole "is vs ought" thing, because an ought that does not depend on the is is a) not relevant to the is, b) cannot be established to be correct. It's nothing but another one of those defeatist admissions of our own limitations.

I don't see any problem with not having 'hard', mathematical rules about human society. Pretending that there is leads to farcical, and potentially dangerous situations. Just take a look at the USSR.
The problem with the USSR was that it sought to rule individual behaviour. I'm not, I want to establish rules that are such that we bridge the gap between individual self-interest and economically efficient macro-outcomes.

According to a rather strange view of happiness, IMO.
Being rich is not the only way of being happy, but not being rich is a pretty good way of not being happy.

There are pretty much no people in the world for whom material wealth (at least up to some minimum level) doesn't feature in the list of things they need to be happy. Most people also prefer a bit more wealth to a bit less.

So from a utilitarian standpoint, maximising the efficiency of the economic system is paramount.

No, it isn't. It's an attempt to get another person to buy a product.
Without the use of force or fraud.

But the most efficient way may be instead any variety of manipulation... preying on people's insecurities, associating the product with attractive images that in fact have nothing to do with its benefits, and so on.
I'm not the most comfortable with my own body, and I somehow seem to manage not to get tricked into buying useless stuff. I don't see why others can't do the same.

No, human beings don't always, or even usually, do what we think is best for us.
So what do we do instead? And where do you get the knowledge of the difference?

Maybe temporarily. Ultimately? That's not clear at all.
And maybe a mother would ultimately be happy with the kid she might have aborted otherwise. The solution isn't to outlaw or eliminate abortion on the off-chance that the mother made a mistake.

You should be. What about drug addicts?
I think that drug addicts are like people with mortgages. You enter into a "contract" of sorts and you deal with the consequences. The information on how drugs work and what they do is available to everyone who will listen, if people still make the choice to get into it, they must have their reasons.

But you can try and put your theory to the test: find a drug addict, try and stop them from doing drugs and see what comes out in the end in terms of happiness. But don't get into the questionable practice of asking for any sort of consent.

Great! Let's have information distributors. Let's just not put profit-seeking corporations interested in manipulation and distortion rather than objectivity and benefit in charge.
What's with the c-word?

Prices distribute information, advertising distributes information and lots of other things do too. Mechanism design is at the moment primarily about negotations (though it can probably ultimately be expanded), but it deals with information distribution.

If someone misrepresents something in an ad, you can get them in trouble. The rest is ultimately your own responsibility, whether you're looking at an ad or a tasty-looking mushroom.

No, but you can try to create a system where only the right factor will play a part.
And you'd probably need mechanism design theory to do it. :p

It's also either the fallacy of equivocation (between "utility" as in "actual benefit" and "utility" as in "I will actually make x decision rather than y decision") or a transparently false assumption that the two will always be the same.
No one says the two are always the same. People make mistakes.

What I'm saying is that revealed preferences are the best way of approximating what makes people happy, and that you cannot come up with a better one.

But this is an empirical question, and a very difficult and ambiguous one at that. Not a matter for resolution in the "cold, hard maths."
Why? If you're saying one approximation is better than another, you've got to have some sort of basis for it.

We know that if we go with revealed preferences, the only time we'll be wrong is if people make a mistake. That mistake will be either due to a lack of information (which we're seeking to minimise) or the person's own fault in being careless.

You're proposing some other model, presumably based on stated preferences (which people can and often do distort if they sniff an incentive to do so) and an element of "I know better than they do". First of all you need to define it, then you need to develop some sort of empirical test to verify that it is in fact better. And that's a statistical procedure.

1. What (generally) makes people happy and what (generally) doesn't.
So you're aware that you're basing this on generalisations, as opposed to the market, which bases the judgement on each individual's revealed preferences.

2. The likely consequences of certain actions.
We know it, and they know it too. And if they don't, that's why we're trying to minimise asymmetric information.

3. The kinds of reasons people do what is not good for them.
Like what? What sort of thing, and what sort of reason?

A person, say, being pressured by others to take a risk he or she doesn't want to take may in fact welcome such intervention... it gives him or her an easy out.

Now, you may say that this is a rare case... but can the "cold, hard maths" tell you that? And that's the important point in all of this: since the "cold, hard maths" make simplifying assumptions that clearly don't always hold in the empirical world, there is always that element of ambiguity in actually putting them into practice.
The cold, hard maths of general equilibrium theory holds it that people act according to their own wishes. That's one assumption - and we have laws that make it reality.

If you're pressured into doing something you don't want to do, you can go to court. So the intervention exists either way, and mechanism design theory is certainly not going to take it away.

So my question is: do you see this sort of thinking gaining ground? If we accept that governments can only be considered legitimate if there is at least a trace of utilitarian thought in our heads, would that mean there is some "perfect/optimal government", based on a mechanism designed according to cold, hard maths?
Economics has always been based on the idea that people are more or less rational, so I don't think this will replace any existing ideas economists may have had about their profession. However, I think the mathematical formalism keeps economists honest, and makes it harder for them to work ideology into what they discover. That's a good thing.
As for a "mathocracy", I'm not sure that economics can create one. Like other posters have said, people do not always act rationally, and the government has to take this into account. It would still be a good thing to have in a government, though, much better than one that is completely irrational.

We can make them such that failing to do the ought will not be rewarded.

Anyways, my reason for it is not metaphysical, but fairly simple: if people don’t act rationally, they’ll use scarce resource in an inefficient way, with the result that there is wastage and people not being as well off as they could have been.
Again, people don’t act rationally at all times.

Constructing a view of how society works, and then creating policies based on that view of society, is fundamentally flawed, because society simply doesn’t work in such a manner. Applying unfitting social policies based on badly inaccurate takes on society has an inglorious history of confusion, strife, violence and oppression.

We just have to look at the example of the Soviet Union to see how an incorrect, ‘hard science’, theory of how society works can reek havoc on individuals and societies. Though I’m not saying that the atrocities committed in the name of dialectical materialism would be repeated with game or mechanism design theory, I don’t see any good argument for deluding ourselves.

Would it not be much better to allow for the fact that humans don’t act rationally at all times, and construct social policies flexible enough to handle this chaotic nature of human life? To perhaps create policies of resource management, distribution and education that minimises the wastage of resources, instead of assuming people will rationally not waste.

I’m not a fan of the whole “is vs ought” thing, because an ought that does not depend on the is is a) not relevant to the is, b) cannot be established to be correct. It’s nothing but another one of those defeatist admissions of our own limitations.
We must recognise that we have our limitations as human beings, whether that be in day-to-day life, scientific and philosophical study, or in public policy. As someone who is supportive of the idea to attempt to rationalise public policy and create hard scientific policy, surely you would be opposed to inputting incorrect data into said scientific theory?