System for determining preferences (theoretical)

[Scott of Hyperborea]

Raikoth has a language, a religion, a culture, a history, a geography, and so on, but as for the government, I've mostly been mumbling that it works on a combination of utilitarian principles and futarchy and then running away before anyone can ask any further questions.

So now I'm working on fleshing that out. I'm trying to combine voting with a sort of utility function type of thing. Which turns out to be harder than it sounds. I don't have any single model that satisfies me, but one model ended up much more interesting than the others.

Give every citizen a survey/ballot with the following scenario (transplanted from Raikoth to America to make it more comprehensible):

Imagine you had $40,000. Further, imagine that every other American had the same amount, and that their spending would exactly mirror your own. So if you spend $5 to buy a burger , every other American will spend $5 and buy a burger. If you give $50 to a scholarship fund, every other American will donate $50 to that scholarship fund. You are permitted to allocate money to a "general fund for doing stuff I think of later", but then all Americans will do the same. How would you spend your money?

$40,000 is approximately the average US per capita income. So the question is asking: if you made as much as the average American, how would you spend your money?

The survey would then include the prices of various goods at various levels as predicted by studies. So, for example, it might say "if no money was invested in police, we predict crime rate would be 50 murders / 100,000. For $30, you can lower it to 10 murders / 100,000. And for $100, you can lower it to 1 murder / 100,000." Of course, it would be much more complicated than this. Or "For $10, we could halve malaria in sub-Saharan Africa; for $20, we could eliminate it completely within 20 years."

If everybody put down that they would spend all their money on things like food and housing and plasma TVs, then the government sets the tax rate at zero and self-disbands (the history of Raikoth actually includes one period of government self-disbandment, so this isn't as weird as it sounds).

But to the degree that people spend their money on things like a police force, a strong military, or saving the rainforest, the system assumes these projects are best handled cooperatively. The percent of money devoted to cooperative projects is averaged across all citizens, and that average is the tax rate (well, plus a little extra for administrative costs). Then the budget reflects the average of all citizen preferences. For example, consider a state with the following two citizens:

<b>Conor the Conservative</b> would spend $8,000 on mortgage, electricitity, heating, etc; $5000 on groceries; $3000 on luxury goods; $2,000 on gas and auto maintenance; give $1000 to his church; $3000 on insurance of various sorts; $5000 on savings; $3000 on vacations; $2000 on educating his children; $4000 on national defense; $1000 on helping the poor; $2000 on police; $1000 on saving the environment.

<b>Linda the Liberal</b> would spend $10,000 on mortgage, electricity, heating, etc; $6000 on groceries; $1,000 on luxury goods; $1000 on transportation; $3000 on insurance of various sorts; $4000 on savings; $1000 on vacations; $2000 on educating her children; $2000 on national defense; $5000 on helping the poor; $1000 on police; and $4000 on saving the environment.

So, of Conor's preferences, $10000 are best done by a government; of Linda's, $14000. Average is 12000/40000 = .3 so tax rate is set at 30%. We average out the numbers to get $2000 on education, $3000 on defense, $3000 on the poor, $1500 on police, $2500 on the environment.

A couple things Raikoth would have to do to make the system work. First of all, it'd need good benchmark figures, so it could say with some credibility "If you want to end world hunger, that'll take $200", and not get embarrassed when people spend $200 and then there's still lots of world hunger. All I can say here is that the country's already got a strong system of prediction markets and things like takeonit but with automated rankings.

It'd need to be efficient enough to make people trust it. The same kind of institutions - prediction markets, careful auditing, and occasionally just having the government donate lump sums to private charities or back to citizens as vouchers after it's done the hard work of getting the money would solve this problem.

It'd also have to prevent people from lying. For example, if someone wanted more money to be given to defense, then assuming everyone else was honest they would be better off allocating all $40,000 to defense than answering honestly. Of course, I trust the Raikothlin not to lie, but it's still an inelegance in the system that needs to be solved. But there are ways to solve it. For example, each year, a certain proportion of people, instead of paying taxes, simply have to actually spend their real money exactly the way they allocated their hypothetical money (within a margin of, say, 20%, and with reasonable adjustment for difference between their income and $40,000). This would give someone who allocated all $40,000 to the military a very bad year. This same intervention would hopefully lessen signaling issues.

I don't know how to prevent lying by setting the correct tax rate, but then giving all of your tax money to one cause, but I should work on it.

This system has some interesting properties. It would let us fix the cost of solving world poverty at about $200 (note that this figure is per American, not per person - if the rest of the world helped, it would be much easier). If God came to me and offered to end world poverty if I gave Him $200, I'd do it, and I assume most other Americans would also. The fact that we're not all spending $200 to end world poverty suggests the problem is mainly one of coordination instead of a lack of will. Making sites like Pledgebank ten thousand times bigger and giving everyone the understanding of why they're useful would solve this, but while I'm designing a utopian society I might as well just build in the solution.

Also, it might just be me, but I imagine this system ending out much more liberal than the current one. Would you pay $400 per year for the war in Iraq? How about $60, one lump sum, to permanently protect the entire Amazon rainforest? Well, right now you are paying that for the Iraq war, and you're <i>not</i> paying that for the Amazon. I think if the problem were phrased in these terms, even people who think they're conservative would find they were less conservative than they thought. Likewise, I think people who think they're libertarian would find that if the question were presented as "how would you allocate your resources in a case where spending $200 would end world poverty?" they'd be just as willing to accept the tax as anyone else.

Next question is how I would find preferences for laws - for example, whether the utility to nonsmokers of banning smoking is greater than the disutility to smokers. Got to start considering that one some time.

[Kaiser Mors VI]

umm... I think I'm a bit confused as to the point of this.

[Malliki Tosha]

I'm a bit confused too since this is completely unrealistic. But in this setting, who cares I suppose.

[Scott of Hyperborea]

Unrealistic how?

[Malliki Tosha]

I hope you're joking.

[CJ Miller]

Hold on.

What about centrists, libertarians, anarchists etc?

Surely you can't believe that the political spectrum is only black and white.

[Scott of Hyperborea]

Hold on. What about centrists, libertarians, anarchists etc? Surely you can't believe that the political spectrum is only black and white.

The only two examples given were a liberal and a conservative, but obviously the preferences of anarchists and libertarians and whatever people choose to call themselves would be collected and averaged as well. What label you give has no relationship to how the system collects preferences.

[Andreas the Wise]

As I've come to expect from you, a very interesting and economically/philosophically enlightened system, albeit requiring important predictive power. I wonder if you're aware how similar it is to some Para-Econ proposals (Participatory Economics, an attempt to be between Capitalism and Socialism) - although yours only allocates government spending by this method and not the rest of the economy, which makes it much easier and much more effective.

Considering the issue of lying/signalling - I take it you are assuming a rational population? Maybe with a utilitarian bent? If so, it would work in Hyperborea; in the real world, where people aren't rational and certainly not highly ethical, lying would be much more of a problem.

The other thing is of course this system encourages a lot more political participation/involvement of citizens; the upside being democracy and hopefully political education; the downside being tyranny of the majority. Maybe it's just from reading a lot of Mill and Tocqueville lately (I'm doing an essay on liberal fears of the tyranny of the majority), but I can imagine people with views that a liberal would consider ... innapropriate ... getting a lot of money allocated to purposes that the more educated members of society would realise are quite innapropriate or even dangerous. As a simple example, if a lot of the population fear immigration, that might not get reflected in Parliament normally; but if they all have the option of spending $1000 on "Stopping boatpeople and getting lazy immigrants out of the country" ... well, you can see where I'm going.

[Scott of Hyperborea]

Ah, good old Andreas, I knew there'd be at least one person who totally understood what I was doing and had a meaningful critique.

The immigration question is an odd one. It assumes that in a normal republic, such fears would get filtered out before making it to the legislature. Recent events in the US have dulled my faith in this a bit - I think part of the point of the Tea Party movement is saying "There used to be this thing where educated people who didn't reflect the prejudices of the populace at large got into Congress, but now we're going to apply a strict ideological purity test to make sure that we only nominate candidates who are as uneducated and bigoted as we are." Watch this November's elections to see how that plays out.

More to the point, I would want this system to be for determining outcomes rather than methods. So for example, a person would allocate money not to "getting rid of immigrants" but to "preserving American jobs", and then it would be the job of economists and sociologists to figure out whether getting rid of immigrants would actually preserve American jobs. But I admit I don't have a strong idea of how I would enforce this, and there might be some people who just genuinely hate immigrants whether they take jobs or not.

The other thing would be that this would not be the system for creating laws, just budgets. So you couldn't allocate money to getting rid of immigrants unless the law said you could get rid of immigrants. This gets us to a very sticky situation in US politics (I dunno if it's the same in Australia) in that we have a lot of laws on the books that we don't enforce well (immigration being a good example of this). I always found this to be weird - either enforce the law or don't have it. I suppose if Raikoth had the same sort of issue, they could use this system to increase enforcement of the laws, but maybe that would be an incentive not to get unjust laws on the books in the first place.

[Andreas the Wise]

Your explanation that this only determines the budget, not imposes policy, is a good one for avoiding the tyranny of the majority issue. Probably you can never produce a system that is democratic and yet guarantees against you - you have to rely on educating citizens.