[EM] help w. planning/working on a monte-carlo simulation?

[Abd ul-Rahman Lomax]

At 02:13 PM 5/17/2013, David L Wetzell wrote:
>Thank you Abd,
>
>I agree consensus is feasible when there are few people and they are 
>committed to working together.

Right. We learned how to accomplish this with modest-sized groups in 
the 20th century. Scaling this up has been thought impossible.

It's not. Indeed, it is *necessary*.

>I'm focusing on US elections and more so "more local" elections that 
>tend to be chronically non-competitive.

Great. It's the place to start. The goal that I see as realizable is 
that voting systems, and, more importantly, the *political process* 
can be designed to increase connectivity and trust between the people 
and government. When I started this, I felt like a voice crying in 
the wilderness. I don't feel that way any more. Lots of people are 
getting it. Structures are being built.

>I agree we all use short-cuts in voting and that that shd be 
>considered in thiinking about what sort of election rules we should use.

The interesting possibility, to me, is that the *simplest* system, 
vote-for-one, can be used in Asset as a ballot, and handling votes 
for more than one can easily be done without harm. Voting strategy is 
so simple it's hard for people to understand?

"What do you mean, vote for the single candidate you most trust? How 
would I know who this would be out of what might be thousands of candidates?"

People get it backwards. Identify the person in your community you 
most trust, talk to them and ask them to register as a candidate, and 
vote for them! And if you don't know anyone, well, that's the source 
of your sense of isolation and disempowerment right there! Meet some 
people! Or find someone else like yourself, do what it takes, and cooperate.

(People think that the "you most trust" is about comparing all the 
candidates and finding the one who most matches your opinions. A 
difficult task indeed! Trust is not about matching opinions, though 
it may correlate with them in some ways. And if we don't know what 
trust is, again, *there is what stops us!*)

>For simulations, I could generate utilities like Warren Smith did, 
>or I could sample from historical data.

For complete work, do both.

>Either way, my goal wd be to refute the notion that a Droop quota is 
>needed for 3-seat elections, when the purpose of the droop quota is 
>served by another means, like the use of an at-large seat.

That's a goal that will distort your results. Do some real science 
here. It starts with exploration. Find some historical data, I 
suggest. Become familiar with how the methods work in actual 
practice. Then design simulations, see if you can predict the 
behavior of real elections with them. The advantage of simulations is 
that you can observe how absolute utility profiles create results, 
and then you can, for the simulations, measure Bayesian regret. Real 
performance as to a measure of overall social satisfaction, known 
because the profiles were *defined.* That's what you cannot do with 
real elections, or at least not generally. There might be some ways to do it.

If you believe that the Droop quota is not needed, try to prove you 
are wrong, by designing tests. If you try to prove that your theory 
is true, you can always make up evidence to prove it.

What's likely is that the Droop quota has functions and the Hare 
quota has functions. I always suggest the Hare quota for Asset 
elections because it's thoroughly fair, all seats fully and without 
compromise represent a quota of voters. However, the Droop quota does 
the same thing, *if one may elect an extra seat*. That seat might 
have some difference in funciton; it represents what I call "dregs." 
Left over votes from elected seats, isolated electors who haven't 
found a compromise. Attempting to complete the election immediately 
is what leads to wasted votes and wasted votes means people who 
aren't represented.

Essentially, the quota can be arbitrary, it simply will produce a 
different number of seats if all votes are used. Is there harm in 
this? *How much harm*?

If you look at what seems like unfairness, I suggest it will be found 
in the rules that attampt to reqiure completion in a fixed time. What 
if the results of elected seats are announced, and not all seats have 
been elected, and over the next weeks and months, electors get it 
together to create more seats? If enough seats have been elected 
initially, the Assembly can begin to function, and it could simply 
decide to consider the threshhold for substantive motions to pass to 
be the majority of electoral votes. It could even allow electors to 
vote directly, while still restricting the right to introduce motions 
and debate on the floor to elected seats.

There are creative ways to address the problem that could transform 
the entire way we look at politics. Asset elections complete through 
*cooperation,* not competition. Every seat is *unanimously elected by 
a quota.* For N seats, that's the Hare quota.

Single-winner elections use the Droop quota! (That is, when a 
majority is required. This is the origin of the Droop quota, really.) 
Single-winner elections *require* wasted votes. The equivalent of 
using the Hare quota in a single-winner election would be a 
requirement for unanimity.

For those that don't know, the Hare quota in an N-winner election 
with V voters casting one vote -- or having one vote active -- is 
V/N. The Droop quota is V/(N+1). The methods we are considering, when 
a candidate is elected with votes in excess of the quota, the 
remaining votes after the quota is substracted are then available to 
be assigned to another caniddate. In Asset, votes held by the 
candidate may be assigned at will by the candidate, as if they were 
the candidate's "assets," hence the name. In STV, the transfers are 
assigned by a ranked ballot, the elected candidate is "eliminated" 
from the remainder of the election, and all ballots with a vote for 
that candidate drop to the next rank, and are given a fractional 
consideration so that the remaining sum is the left-over vote. At 
least that's one method.

The stunning power of asset comes from the transfers being under 
intelligent control, and available for negotiation. The voters assign 
trust to electors, who then operate publically as to any re-voting. 
The electors can use any process they choose to decide how to rassign 
their votes. I expect that seats will be elected chaotically, 
essentially one at a time, as electors find agreement to assign a 
quota of votes to a seat.

This is actually what happened with the ESF Steering Committee 
election. I was elected immediately, the only candidate with more 
than a quota of votes (Hare or Droop). I then elected the second of 
three seats, assigning him, beyond his own votes, enough to take him 
to the Droop quota, with, as I recall, still having enough votes to 
give him a Hare quota if necessary. (I don't remember exactly. The 
rules had been imperfectly and, my opinion, incorrectly specified, 
but the Asset concept is so powerful that we, as the candidates, 
could create a *unanimous election,* and with unanimity, is anyone 
gong to complain that we did not follow the announced process?). And 
then the second-highest vote-getter transferred his votes to a 
candidate who had only one vote of his own, giving him a Droop quota.

So, to make the election unanimous, I asked the remaining candidate 
-- he had two votes, as I recall -- if he would approve the election. 
He did. Now, *this could have been a contentious election*. Asset 
turned it into consensus. I happen to think that is *way cool*.

For a City Council, what would it be like? *I don't know.* But there 
is a very good chance that it would have the same effect. With a 
larger voting population and many more electors, some might be unable 
to find a place to put their votes, but as the remaining pool gets 
smaller, electors holding votes, still, will become more and more 
visible. They will find each other and find ways to cooperate, 
because, until they do, they do not gain access to the Assembly 
floor. There need be no rush.

Two electors with a half quota each who say to each other, "You 
should give your votes to me" could do that for a long time, until 
they get it together to make a common decision. It might be to elect 
someone else! (I would allow the electoral college to elect anyone 
legally eligible for the position, and willing to serve. Very much, 
it need not be an elector with votes.)

That's true, in fact, if one has 2/3 of a quota and the other 1/3. 
Neither can coerce the other. The whole competitive premise of 
politics is undermined. With Asset, "campaigning for a seat" will 
quite properly be seen as suspicious. No, the electors will *draft* 
winners through the process. The electors will have, all of them, 
real power. When I started to understand these concepts over the last 
few decades, it blew me away.

(In Asset, "candidates" are actually electors, public voters. Because 
asset wastes no votes, I expect to see *many* electors. It's really 
as many people as are willing to participate in public life, as 
people who will vote in public, as distinct from secretly, and to 
create and advise an elected assembly. In a fully open system, 
registering as an elector would be almost trivial. Ballots would not 
have candidate names on them, at all. I'd expect there to be a 
publication available at the polling place that gives a unique code 
to enter on the ballot. (Written names would be way too difficult to 
misidentify, currently many write-in votes are not recognized even 
where completely legal, which is why some jurisdictions require that 
write-ins be registered.)