[EM] Trees and single-winner methods

[Abd ul-Rahman Lomax]

At 06:16 PM 3/17/2007, Juho wrote:
One could also say that Asset voting is not free of the need for
>strategic considerations but that the strategic considerations get so
>complex that the the votes could as well forget them.

Actually, I wouldn't agree. What happens is that "strategic" 
considerations can become moot. Why bother, indeed, with them when I 
can simply delegate the complexity to someone whom I trust and who is 
presumably in a better position to use my voting power to desired effect.

Yes, this involves surrendering some authority. As if we didn't 
surrender authority through elections anyway!

(You want authority, you should want direct democracy, not 
representative democracy. Indeed, if delegable proxy is implemented 
only for representation in deliberation, and direct voting is still 
allowed, you would not have to surrender authority at all. Simply vote.)

>  I mean that as
>a voter I might be thinking that I know candidate A quite well and he
>would probably behave in a certain way when participating the further
>negotiations and elections, and therefore it would be strategically
>optimal to vote for him. But as said, this may be too complex to manage.

Indeed. And it is not at all the way I'd think about it. Rather, I'd 
think, "I know A and trust that A's decisions will be at least as 
good as my own, so I'll let A cast my vote for me."

In Asset Voting, bullet voting for a candidate is equivalent to 
saying, "I trust this person with the office, and if not for the 
office, then for exercising my vote to select who will be in the 
office." Thus it is equivalent to a Plurality vote plus the 
permission for the candidate selected to transfer the vote if the 
candidate deems it appropriate.

The point I make again and again is that it is actually only a very 
small step beyond trusting a candidate for an office and trusting the 
candidate to elect someone to the office.

>>Given, again, that there is no need that the "candidate" actually
>>be elected or electable, I can choose a candidate whom I personally
>>know. I expect the numbers of candidates to blossom if Asset is
>>adopted. And the result will be much closer to what a hiring search
>>would produce.
>
>Many points in your mail dealt with knowing the candidates
>personally. If we want this property, the basic model in my mind is
>to arrange more levels in the representational system.

Levels are inflexible and add complexity. I originally started with a 
multilevel system and realized that the complication was utterly 
unnecessary. Asset Voting creates "levels" on the fly, as needed. If 
your preferenc3e gets enough votes, there is no need for any 
intermediate level with regard to his or her election.

You are not required to know the candidate personally. However, Asset 
*allows* you to vote based on this.

>Let's start from a village of 100 inhabitants. Everyone knows most of
>the other inhabitants quite well. The village elects 5 of the
>inhabitants to represent the village in communication towards he
>external world.

By the way, one of the early people interested in the FA/DP concepts 
was a Brazilian sociologist interested in how remote villages could 
be represented.

>Then 20 villages send all their 5 representatives to a town meeting.
>All 100 meeting participants know each others reasonably well since
>this group has made decisions together many times before. The meeting
>elects 5 of the participants to represent the town in communication
>towards the external world.

This is similar to the first ideas I had. Almost twenty years ago.... 
a series of face-to-face meetings.

And I rejected it in favor of simple delegable proxy.

Unstated, entirely, was the method by which the five were elected, 
and whether or not they have equal voting power. And why five? Why 
not one or ten or ... whatever number the village chooses to send. 
What if every villager can *choose* a representative, does not have 
to compete with anyone else? TANSTAAFL, who is going to pay for the 
reps expenses? Well, those who chose him or her, that's who!

It happens to be a libertarian solution, though I was not aware of it 
at the time.

And each representative, at the next level, has as many votes as 
those he or she represents. Simple, fair, and totally flexible. In 
the system described by Juho, every village is equal, but villages do 
not come in nice round numbers like 100 inhabitants.

What would actually happen, I think: villages would generally choose 
one or two reps. Rarely, three.


>Then 20 towns form a region. Now we already cover a population of
>40'000. The next level covers population of 800'000. Then 16'000'000,
>320'000'000 and 6'400'000'000. And finally we have 5 persons that
>could represent the earth in communication with other civilizations,
>if needed.

Yes. We did the math twenty years ago. But this is a system which 
does not actually represent people. It represents blocks, and as 
stated, treats them equally. My original concept was a U.S. 
presidential election held in nine days, through meetings of ten a 
day. Ten meet and elect the best of them to the next level. Etc.

But what if you don't agree with the rest of your "ten"? Systems like 
this can, because of how the blocks are put together ("districts") 
fail to find a true majority winner, much less represent a population 
more thoroughly.

So, I thought, well, people would form the groups of ten based on 
affinity. And the complexity snowballed.

If it was going to get complex, why not go all the way? Why not make 
it a fractal, with no fixed group size; people choose their own 
representative, who chooses his or her representative, etc? Fractal 
Democracy (TM), Patent Pending. :-)

Ah, I realized.... the basic idea is actually very old and very 
basic. Proxies. Freely chosen, not elected. The only difference is 
the delegability twist. Which is a reasonable extension of the powers 
of a proxy, and, in fact, proxy assignments often provide for such 
delegation, where reasonable and necessary for fulfilling the duties 
of a proxy.

>One key positive thing in this scenario is that the representatives
>are always in direct contact with the people who elected them and
>therefore need to be able to explain to them at personal level the
>rationale behind whatever decisions or negotiations they work with.

This is true in spades with delegable proxy. There is no dilution of 
responsibility. You don't like what your proxy is doing, you lose 
trust or faith in him or her, you just change your proxy assignment. 
Effectively immediately, though in some systems, naturally, there 
might be some latency.

>One key negative thing in this scenario is that the direct
>responsibility may fade away when the distance from the village
>people to the top level decision makers increases. It is e.g.
>possible that the people at the top consider themselves to be more
>clever and more important than the people that elected them, they may
>consider their closest colleagues and direct electors more important
>than those at the lower levels.

The relationship with "direct electors" is crucial, actually. The 
relationship with "those at the lower levels" is actually impossible 
to maintain. But everyone has access to the network of relationships, 
and everyone is directly responsible to those who directly chose 
them. It is fixed terms that create much of the problem, plus 
elections. Proxy representation totally avoids both of these.

>Clearly there is a tradeoff between knowing your nearest
>representatives at personal level, and electing your top level
>representatives directly but knowing them only via TV. To me the
>additional layer of representatives and negotiations that you
>discussed represents in some sense adding one step in this hierarchy.

It adds no steps or many steps depending on how the voters and 
candidates use it. Asset only adds steps where they serve a useful purpose.

>Direct democracy has some benefits and the model above has some. Same
>with weaknesses.

Pat comment.

It would be more useful to look at precisely what the weaknesses are 
of direct democracy, and how they can be addressed and ameliorated 
without losing the strengths. I think I have an answer and there may be others.

But I haven't seen any others, really. Proportional representation is 
about the closest that others get to it.

In any case, the ultimate realization was that we did not need to 
wait for changes in law to implement delegable proxy. FA/DP 
implements it directly and immediately, whenever people recognize the 
need and simply do it.

The reason is that the core problem of democracy is how people make 
decisions collectively. And the problem of scale in democracy is the 
problem of how large numbers of people can do this.

Solve the decision-making problem *outside of government* and you 
have also solved it within government. How can large numbers of 
people communicate, cooperate, and coordinate, *voluntarily* and *efficiently*?

Solve that problem and apply it to, say, a political party. If the 
theory of the solution is correct, this party will be more successful 
than competitors, and thus it will be more able to mobilize votes and 
resources more effectively. And thus win elections or change laws. If 
necessary. It actually isn't very necessary, the changes in law, not 
nearly as necessary as I thought when I began.

And if you go down that road a distance, I think you will come up with FA/DP.