(Long) Re: AltInst: Vote on Values, But Bet on Beliefs

Robin Hanson <rhanson@gmu.edu> writes:

> This non-technical paper proposes a new form of government.
> Comments welcome, though no rush as I'll be on vacation for the next three
> weeks.
>
> -----------------------------------------------------------------------------
> http://hanson.gmu.edu/futarchy.pdf or .ps
>
> Shall We Vote on Values, But Bet on Beliefs?
> Robin Hanson, July 2000

Thank you for publishing that paper. I found it very interesting. I
had been thinking along those lines myself for some time, altho I came
to different conclusions.

I have some comments, which I mean in the most constructive way.
Please don't take them as fault-finding.

I'm skipping the first few of the 25 objections, because they speak
more to people without experience with IF, FX etc.

"Democracy mite become unstable" seems overly theoretical in both
objection and answer - I suppose that can't be helped. I notice that
the answer does not consider the possibility of direct democracy, ie
referendums instead of persons as candidates. Referendums may not
always be possible, but ISTM that since you confine democracy to
mostly defining the fitness function, referendums could and should be
more heavily used. So ISTM there is a better answer.

"Time-consistency mite be a problem" seems one of the stronger
objections, but as you point out, it's at least as bad for the status
quo. There may possibly be a financial solution in the sense of
issuing bonds to back up the commitment, perhaps a bond that pays
double if the govt goes back on its promise.

"Expressive voting could still cause problems". The answer isn't very
convincing: If it's a problem, we mite find solutions. It's hard to
see what form of voting could alter people's underlying preferences.
Also, I tend to think that the voting system would not allow the
voting system to change, much as the US national elections will not
switch to approval voting, tho it's better and there is a clear
appetite in the US for an alternative to the two ruling parties.

But OTOH, it's not clear to me that expressive voting is the only or
even major problem in democracy. I would have said the sort of
special-interest voting that Libertarians like to call "4 wolves and a
sheep voting on what to have for dinner". ISTM this would be just as
much an issue in futarchy.

"People could buy policy via betting markets". I notice your answer
centers on a specific case, dealing with a stadium where insiders try
to boy a "favorable impression" and then profit buy selling before the
bubble bursts. ISTM this doesn't really address the case of what I
call "short circuits" where the policy at issue illegitimately creates
its own funding. One could imagine other cases whose effect on policy
had the effect of funneling money to themselves, eg a defense
contractor bidding on a re-armament policy. In that case, the profit
they get from creating policy affects the equilibrium price, so they
may win even tho they take a small loss in the market. Whereas the
tax money spent is spread over hundreds of millions, so no one
taxpayer has strong incentive to move the price the other way.
So ISTM you need to address a stronger case example.

"You need a way to tell if a proposal was implemented". You appeal to
contract law, but it's not clear that policy writers would have
incentive to be as clear as possible, which contract writers do.
Couldn't they potentially profit directly or indirectly from unclear
policies, because they may later be called as expert witnesses on
their own document and can adjust their testimony according to their
investments or the investments of select others?

"A policy mite influence how welfare was measured". As you noticed,
futarchy has to appeal to external rules and judgements to deal with
this kind of "short circuit".

"Welfare measurement mite be corrupted". IMO it would surely happen
unless a much stronger basis for preventing it can be found.

"People could do harm to win bets". Good answer.

"Policies mite just be encoded in the welfare definition". I tend to
think that would be the workaround of choice. Governments have often
made policy by defining terms against the way they were clearly meant
to be understood, as sure as ketchup is a vegetable in
schoolchildren's lunches.

"Most decisions cannot detectably impact welfare". I guess the answer
works. But ISTM issues would become much less transparent, and would
sometimes be gerrymandered, even if they were privately defined (eg,
by mutual funds of some sort). I'd be more comfortable seeing some
reason to expect that it's generally or always possible to aggregate
small policies in a way that's transparent to the investor.

"You must define when a market "Clearly" estimates". As you point
out, this is a thorny problem. It's basically a free parameter in
futarchy.

"You mite not catch buggy decisions quickly enuff". The answer works,
but again ISTM it would work better with referendums. That would
eliminate the problem of legislators' insider trading. It could also
be possible to condition a referendum on a derivative issue reaching a
certain capitalization.

I am skipping a couple of objections wrt the transition to futarchy.
In fact the political resistance would be incredible, but ISTM rite
now we are still talking about whether and how it would work, not how
to get there.

"It seems hard to make one function encode all our values". Good
answer.

"So many things would have to go into a welfare definition". The
difference between this and the last is very subtle. I think it had
something to do with initialization vs steady state? In any case,
good answer again.

"You can not pay off bets if the earth is destroyed". The problem, I
guess, is that policies that increase both the probability of
destruction and of some lesser gain would be overvalued, and so the
futarchy would chug mechanically on, using the irrational policy. I'm
not sure whether the answer works.

"But infinity never arrives". Good answer.

"You have not guaranteed basic rites and freedoms". ISTM it would do
at least as well as the status quo.

"Risk premiums and excess volatility distort estimates" I wonder if
the cure, adjustng the fitness metric to correct for it, isn't worse
than the disease. ISTM it's one more opportunity for gerrymandering
the fitness metric. At least ISTM futarchy should limit the
correction parameter to a single overall risk-aversion curve.

"There is a decision selection bias". Good answer.

All in all, I'm very interested in the idea, and I'd like to see if we
can find answers for the weaker points.

-- 
Tom Breton, http://world.std.com/~tob
Not using "gh" since 1997. http://world.std.com/~tob/ugh-free.html
[To drop AltInst, tell:  majordomo@cco.caltech.edu  to: unsubscribe altinst]