[EM] Asset voting and so forth, in practice

[Abd ul-Rahman Lomax]

At 06:20 AM 11/2/2009, Michael Allan wrote:
>If I understand you Abd, we're currently developing the tools for
>voters to do essentially what you describe.  And we've made some
>progress recently:
>
>   Drafting media:  http://t.zelea.com/wiki/Toronto:Pollwiki
>   Voting engine:   http://t.zelea.com:8080/v/w/
>   Bird's eye view: http://t.zelea.com/wiki/User:Mike-ZeleaCom/p/de
>
>   More info:       http://zelea.com/project/votorola/home.xht
>
>It's all still rough work, and in flux, but the code is running live.
>So we can probably start doing this today, in small ways.  (More
>below.)

Okay, comments. First of all, great. And bells and whistles are nice. 
But it's also important to make the system bulletproof, protecting it 
from hazards and even hazards that come from an insider, a sysop, not 
to mention hackers.

You appear to have set up a delegable proxy system. Great. And for 
general polling purposes, that could be fine. However, I do have some 
strong recommendations.

1. Delegable proxy is very safe if it's an advisory system, and the 
interpretation is up to those who want to be advised. So:

2. There may be a totally anonymous layer corresponding to secret 
ballot. You would still want to have some kind of registration 
confirmation, I suspect. This would mean an engine for privately 
voting. If that's done, on this layer, everyone privately votes. 
These votes are considered in a distinct way. Can they be amalgamated 
secretly through delegable proxy? I suggest not, because then 
responsibility is lost, as is traceability. If I secretly vote for A, 
I can see the secret vote total for A go up 1 vote. However, if A 
votes for B privately, B's vote does not go up.

In Town Meeting democracy, everyone votes publicly on issues and in 
some elections (basically appointments). In Asset Voting systems, we 
assume a secret ballot layer, but from then on, in amalgamation, 
there is only public voting by those who elect to participate.

I have generally suggested that proxies be accepted to be valid; and 
that direct communication between proxy and client be set up. This, 
in fact, makes the system bulletproof, impossible to successfully 
corrupt through attacking the central mechanisms.

So, procedure. A nominates B as proxy. A's email address is 
automatically provided to B. B accepts, and B's email address is 
automatically provided to A. I'd recommend phone numbers as well. A 
should be able to expect to discuss matters directly with B and get 
personal response, or at least as much response as satisfies B. The 
theory is that this will cause traffic to be self-regulated, so that 
every connection in a DP network represents some level of mutual 
trust, at least a provisional trust. This procedure applies to the 
public system, the proxies are openly listed, and votes are openly shown.

Private voting may be done on any issue, but those votes are 
segregated and so identified. They are one-person direct votes, and 
they can't be anonymously amalgamated with the public votes because 
that could then be double-voting. In a DP system, in vote analysis, 
if a client votes directly, one is subtracted from the vote total 
attributed to the proxy.

Ideally, this is what is done: everyone names a proxy, unless they 
are willing to vote publicly, in which case they may name themselves 
if they wish. But it's actually silly, since once you are a public 
voter, you might as well have the backup of someone to *generally 
represent* you where you don't have time. But, indeed, if you don't 
trust anyone sufficiently, you can do all the work yourself!

In an Asset system, everyone votes in the secret ballot system, if 
they want. From there up it is all public voting, and the secret 
votes cannot be cast directly, until the next open election. To cast 
a vote on an issue, you must be an "elector," i.e., a public voter. 
The "election" could be every day or even continuous, but it's 
probably best if it has a specific period involved, or vote 
calculations could get hairy.

What this means is that you can actually talk to the opposition; 
that's impossible with public elections --- which can be pretty frustrating!

On the other hand, if there are security issues, people should be 
able to amalgamate their votes to a single person, because one person 
representing many can be protected. People who cast their secret vote 
for a candidate will know that the candidate didn't make a quota, if 
there is one, and hasn't accepted becoming a public voter. So they 
can change their vote, perhaps after talking with the one they chose.

Part of the whole "plot" is to get people talking with each other, 
not merely shoving electrons around on a web site. Actually meeting 
and talking about issues. And a proxy, then, becomes someone who can 
influence real votes in the real world, and more than votes, campaign 
contributions, or other exercises of individual power.

>Even if party leaders could grant a substansive democracy, that might
>not be the best way to acheive it - receiving it as a gift from on
>high.  Wouldn't it be better to build it ourselves from the ground up?

Safer, perhaps. However, it is always possible that someone with 
power will see the ultimate benefit, and will act to implement it. 
It's not normal behavior; people who have attained influence or power 
through the existing system will generally fear that a shift toward 
wide distribution of power will cause loss of intelligent direction. 
I.e., their direction. There is a truth to this; simply shifting to 
raw direct ad-hoc ochlocracy can indeed make things worse. Vide 
Wikipedia, which works very well in certain ways, but which breaks 
down badly in the presence of serious controversy, since oligarchies 
naturally formed, per the Iron Law of Oligarchy, see the Wikipedia article!

Wikipedia is an organization which *requires* consensus if it is to 
fulfill its core policy of neutrality, for the only way to be 
reliably certain that text is neutral is that it enjoys very high 
consensus, the higher the consensus, the greater the certainty, but 
because Wikipedians were generally not aware of extant 
full-consensus-seeking techniques, there was a practical settling for 
what's called "rough consensus," which has no specific defined 
meaning, plus actual decisions of any permanency are made by 
administrators with the tools to delete (actually, hide from public 
view, normal administrators can't remove changes from the database) 
or edit-protect, or block editors, and there is a schizophrenic 
double-value: decisions are supposed to reflect the cogency of the 
arguments presented in a discussion, not the "vote," but decisions 
are also supposed to represent "rough consensus." It drives people 
literally crazy, because there is little consistency and 
predictability. Wikipedia got stuck in this, and it's almost 
impossible to change. Unless Wikipedia editors organize directly to 
seek and find consensus off-wiki. If they do it on-wiki, historically 
there have been attempts, the on-wiki structures are crushed, it's 
happened more than once. Even without abuse. ("Wasting time" is 
considered a reason to delete or freeze projects. Get back to the 
content salt mines, slaves!)

And right now, because a group of editors formed a mailing list to 
support each other and discuss issues of interest, the Arbitration 
Committee seems poised to ban and block the members of the list, 
based on allegations that boil down to this being a "cabal." The list 
came to light because someone posted the list archive. The arbitrator 
who has drafted the proposed remedy has openly stated his goal as 
being to prevent people from communicating off-wiki about Wikipedia 
content.... but it's impossible to stop it, in fact, and off-wiki 
coordination is actually the norm in some circles. Such as the 
Arbitration Committee itself, or other groups that are known. But 
those groups are composed of "insiders," not outsiders. Only insiders 
are allowed to communicate privately, it would seem....

To defang the oligarchy and maintain its rightful position as servant 
of the public, it's essentialy that the oligarchy not have control 
over the communication mechanisms that allow the community to form 
consensus. As named proxies, they still have their power and 
influence, but not, ultimately, control. Real-world power is vested 
in the traditional real-world organizations, as continuously advised 
by an Asset or delegable proxy network. To goal is for the advice, 
which is visible to anyone who cares, to be as trustworthy as 
possible, so the actual office-holders get the best advice from the 
entire electorate, which includes experts on relevant topics, and 
that advice is filtered through the network so it is not just a mass 
of undigested opinion, which is way too noisy. And the voters who 
empower the office-holders in a democracy get good advice as to whom 
to vote for. Ultimately coming from people who know those 
office-holders personally, but, again, moderated through intelligent filters.

The important thing about delegable proxy is not voting, as such, it 
is communication, filtered communication.

>It's not uninteresting work, either.  But we could use some help from
>voting experts and mathematicians.  Does it look theoretically sound,
>what we're doing?

I couldn't tell enough details from the pages you pointed to.

But this is the basic principle for bulletproof delegable proxy: 
public proxy table. If it's considered necessary to have a secret 
ballot layer, let that be considered separately, it's an optional 
analysis. I.e., there are public participants and private participants.

And direct communication established when a proxy/client relationship 
is established. For some reason, Clint Eastwood became the model for 
the mass proxy. You can designate Clint Eastwood, but will Clint 
accept? It means you'd get contact information for Clint. Clint, if 
for some reason Clint wants the burden, could accept and use an 
address that is essentially opened by an agent. If the agent provides 
good service, why complain? But would the agent provide good service? 
It's expensive. If all this is used for is advice, why spend that 
money? So, what would happen is that Clint would have direct clients, 
and would advise people who want to designate Clint to name one of 
them instead. Same ultimate "voting power." But a filtering link 
inserted. Don't like the person Clint suggests? Well, that might say 
something about Clint. You can name someone else. With public proxy 
tables, you might be able to find someone you are compatible with who 
is linked to Clint. Or not, in which case you might realize that 
everyone whom you might trust and who will trust you doesn't trust 
Clint. That, too, would tell you something.

If you want Clint's advice, you can still sign up for a mailing list 
Clint controls.....