[EM] Delegable proxy/cascade and killer apps

[Michael Allan]

Michael Allan wrote:
> >> > But I was speaking of overlay networks like Napster...

Raph Frank wrote:
> >> ...even with a small number of people, it was worth using.  You
> >> could use it to send files to friends.
> >
> > Yes, that's the key.  It's free of scale dependencies.  Same for open
> > voting (though I haven't been able to convince you).
> 
> I am not sure I agree.  Napster was more valuable the more people who used it.
> 
> What exactly do you mean by scale invariant.

Like you said about Napster, "even with a small number of people, it
was worth using".  But I'm mistaken to claim that Napster was
therefore "free of scale dependencies".  It's not either/or.  A
start-up threshold can be orthogonal to a network effect.  The lowest
network value (at size 1) might nevertheless be a high value.  So the
seedling network can be viable, despite a strong network effect.

Or maybe the lowest value is truly low, but there is little "friction"
in the system.  So it can nevertheless function, attract more
participants, and grow.  So again, the network would be viable at size
1.
 
> > Again, you are suggesting this because you are concerned about a
> > network effect.  I believe it's scale free.  If a sports club is
> > roughly equivalent to a city, except smaller, then I want the city.
> > Why test on a rough equivalent, when I can test on the real thing?
> 
> Well, if it really scale invariant, then a smaller group is easier to
> convince to actually start using it.

Sure, when it comes to recruiting seed members, a smaller number is
easier to recruit.  But the same argument applies to the city.  We can
start a network at size T, where T is the absolute start-up threshold
(hopefully not much larger than 2 or 3 people).  It does not matter
how big the overall population is.  It could be 50, or 50 thousand.  I
am hoping that T is the same regardless.
 
> ... Have you considered using the pollster's method of directly
> contacting people?
> 
> Ofc, they just ring up and ask if someone what is their opinion on an
> issue and it takes the person a few seconds to respond.
> 
> Maybe you could ask "who is the person you most trust?", but that
> might sound a little spooky :).  This would give you the initial trust
> links, but probably each person would just vote for 1 other person.
> 
> Alternatively, you could ask them directly to participate.

That's my current approach.  I'm starting with activists, lobbyists
and fringe party members.
 
> Actually, what about this:
> 
> You assign each resident various scores.  For example
> 
> - location
> - gender
> - age
> - income (doubt it is public and anyway, will kinda be related to the
> location/age combo)
> 
> Each resident is assigned to a location, even if they aren't participants.
> 
> Any resident can join the system and is then verified.
> 
> Each participant can assign their vote to another participant under you rules.
> 
> However, all non-participants are assigned to the participant that is
> closest to them.
> 
> This creates an incentive to join.  If you are from a group of people
> who aren't well represented, then you would be able to assign the
> votes for lots of your peers.
> 
> Another way of looking at it is that each participant's votes are
> rescaled to eliminate participation bias.  In fact, that might be a
> better way of describing it, as being allowed to vote other people's
> votes is not good public relations.
> 
> In any case, it gets you the running poll system.  If a person doesn't
> log in every month or so, they might have their participation score
> decreased.  Their vote could drop by 10% every week they are away
> after the first month.

Creating the illusion of participation?  Like decoys on a pond?  If we
need to prime the network, if T is large...  I'm still hoping the
ducks will land on an empty pond. (?)

> >  http://web.archive.org/web/20061113083043/wikocracy.com/wiki/index.php/Main_Page
> 
> Some like the right to bear arms were basically edit wars.  One of the
> compromises that I suggested was to allow the right to apply to single
> target weapons (if it can be used to just kill one person at a time),
> but not area weapons like bombs.  This was disliked by some as it
> would prevent militia from being allowed have artillery.

What's a militia without artillery?  ;)

(Still, it's a creative compromise.  I saw your proposal to assign 50%
of all corporate stock as "liability shares".  I'll come back and read
more, when I can. :)
 
> > I was thinking the most natural drafting medium to support
> > this is recombinant text...
> 
> However, care needs to prevent massive explosion of the options.
> There should be 2 forces, people making suggestions expand the
> options, and then the consensus mechanism reduces them to a manageable
> number,

That's the thing, to somehow put a lid on divergence, and encourage
convergence.  It led me to cascade voting.

Fortunately it does no such thing.  There's no force to it.  It's
purely neutral.  It only *reveals* the current state of dissensus and
consensus (without encouraging either) so people can coordinate their
actions - shifting text content, and shifting votes - with a
collective sense of purpose.  If people want consensus, they will
probably get it.  Otherwise not.

I don't know if delegable proxy permits that.  I wonder if its proxies
instead *alienate* votes from the original casters?  That's what I
find in Abd's references to "asset voting" [1], and to Lewis Carrol's
original 1884 pamphlet [2].  The same with Rodriguez and Steinbock's
"dynamically distributed democracy" [3-5].  In all of these, the
original casters are excluded from any consensus that emerges among
the delegates, because they have lost the ability to dissent from it.
They cannot shift their votes.

 [1] http://rangevoting.org/AssetCLD.html

 [2] Duncan Black. 1969.  Lewis Carroll and the theory of games.  The
     American Economic Review. 59: 2, p. 210.

 [3] Marko Antonio Rodriguez, Daniel Joshua Steinbock. 2004.  A social
     network for societal-scale decision-making systems.  NAACSOS '04.
     Proceedings of the North American Association for Computational
     Social and Organizational Science Conference.  Pittsburgh,
     Pennsylvania.

 [4] Marko A. Rodriguez, Daniel J. Steinbock. 2006.  The anatomy of a
     large scale collective decision making system.  Los Alamos
     National Laboratory Technical Report.  LA-UR-06-2139.

 [5] Marko A. Rodriguez, Daniel J. Steinbock, Jennifer H. Watkins,
     Carlos Gershenson, Johan Bollen, Victor Grey, Brad deGraf. 2007.
     Smartocracy: social networks for collective decision making.
     40th Annual Hawaii International Conference on System Sciences
     (HICSS'07).

> > ... What if the candidate-drafter provided write-access to her
> > immediate voters?  The voters could submit text, and she could be
> > editor in chief.
> 
> Another option is to have 2 versions of the page.  The wiki version
> and the "committed" version.
> 
> The committed version has been approved by the proxy in question.  The
> wiki version can be modified by anyone who has been given write access
> by the proxy.
> 
> The proxy would then transfer any info from one to the other that he
> agrees with.

It introduces another "warehouse" into the supply chain.  All else
being equal, I vote for the simpler design.  (But I guess drafters
will experiment with different arrangements to see what works.  Text
media are largely decoupled from the core voting medium, so anything
goes.)
 
> > Distortion and suppression are problems...
> 
> If your proxy does that, then you could withdraw your support and
> submit the change elsewhere.

There may be no "elsewheres" left to go.

Also, distortion and suppression in a candidate draft is inevitable.
It's an aggregate of multiple voter drafts (and equivalently of ideas
from non-drafting voters).  The candidate drafter must be selective to
some extent.  There'll be loss of information.

To prevent it being lost completely, I believe that voters must be
allowed to express their own texts separately.  They must be permitted
to have a "presence" that cannot be distorted or suppressed.  So
individual texts for individual people - if they wish.  The general
structure is therefore a population of texts, or wikis.

> > ... a cascade of Wikis...
> 
> I wonder if it is worth having a composite draft that is subject to discussion.
> 
> There would need to be a consensus agreement on how the draft should
> be formatted (as in what goes in each section).
> 
> Then the page might be
> 
> Section 3 -
> Option 1: Summary
> click to expand text
> Option 2: Summary
> click to expand text
> 
> and so on, with maybe the support for each option shown.
> 
> 2 proxies might be willing to merge their drafts like that if it meant
> that they could agree to suggest them at a higher level.
> 
> This would mean that users would be able to see the main options in each case.
> 
> One possible disadvantage is that sometimes a compromise would involve
> agreements on more than one option.

Ultimately the candidate drafter must decide and live with the
political consequences.  It will be tedious work.  Endless haggling.
Then an immediate loss of votes, as the dissenters depart.

Big moves will be more exciting.  Like where someone (maybe a
disgruntled voter) proposes a synthesis of two or more texts, and
attempts to recruit from the separate voter pools (pole-vaulting over
her candidate's head).

> > ...In the beta release, all of the electoral data will be public
> > for verification purposes.  The same data will suffice to fork the
> > electoral server, as a last resort.
> 
> But not the final release?

Oh yes, there too.  But not in the current alpha release.

  alpha > beta > final
 
> >> The point of participation is to have some change in the real world.
> >> This means that you need some way to verify that the people you are
> >> talking to (and possibly compromising with) are actually capable of
> >> fulfilling their side of the bargain.
> >
> > I guess that's the purpose of the electoral system.  If you disappoint
> > your voters and don't answer for it, you'll lose their votes.  It will
> > happen immediately.
> 
> I wasn't referring to the elected, the system should be usable in most
> general instances.  You could have 2 groups of people who agree on a
> compromise candidate that they will back.  However, when it actually
> becomes time to canvas for the compromise, only one group shows up.
> 
> It turns out the other group was actually a sock puppet army or maybe
> the proxy was just representing lazy people.
 
Sock puppets are dealt with in the voter register, separately from the
elections (one register, many elections).  We can discount the
possibility of electoral support from sock puppets.

> The effect would be that that proxy would be considered to possibly
> represent some voters, but isn't someone you can rely on to get people
> to actually do anything.

A candidate serves her voters.  If she keeps them happy, I guess she
keeps their support.

So A and B agree to back a third candidate (C).  Then A and B
consummate the deal by shifting their votes to C.  If A's voters are
happy with that, then everyone gets what they bargained for.
Otherwise, A immediately loses votes.  As a consequence, she is no
longer in a position to make deals at that high level.

-- 
Michael Allan

Toronto, 647-436-4521
http://zelea.com/