From: proff@suburbia.net
Date: Wed, 22 Jan 1997 22:40:54 -0800 (PST)
To: cypherpunks@toad.com
Subject: Re: Commerical applications (was: Development and validation (fwd)
Message-ID: <199701230640.WAA06727@toad.com>
MIME-Version: 1.0
Content-Type: text/plain


>From terry@lambert.org Thu Jan 23 00:20:21 1997
Return-Path: <terry@lambert.org>
Delivered-To: proff@suburbia.net
Received: (qmail 11676 invoked from network); 23 Jan 1997 00:20:09 -0000
Received: from phaeton.artisoft.com (198.17.250.211)
  by suburbia.net with SMTP; 23 Jan 1997 00:20:09 -0000
Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id RAA22488; Wed, 22 Jan 1997 17:04:17 -0700
From: Terry Lambert <terry@lambert.org>
Message-Id: <199701230004.RAA22488@phaeton.artisoft.com>
Subject: Re: Commerical applications (was: Development and validation
To: proff@suburbia.net
Date: Wed, 22 Jan 1997 17:04:17 -0700 (MST)
Cc: nate@mt.sri.com, terry@lambert.org, chuckr@glue.umd.edu,
        hackers@freebsd.org
In-Reply-To: <19970122141820.16633.qmail@suburbia.net> from "proff@suburbia.net" at Jan 23, 97 01:18:20 am
X-Mailer: ELM [version 2.4 PL24]
Content-Length:  6273

> > Terry Lambert writes:
> > > > > A weighted democracy would be one open-ended growth solution, as
> > > > > long as parametric changes could be made within the system.  I have
> > > > > suggested this before.  A trivial napkin drawing version:
> > > > 
> 
> I have doubts about such a system. How are the weights chosen?

By core team fiat.  It is they who will be giving up power, it is they
who have the power to enforce initial disbursement.  After that, it
*should* be metastable.  If it isn't, it can be reigned in.

> 	- Should all weights be votable - dynamically adjusted according
> 	  to votes * current weights?

This one: yes.

> 	- Which is more stable? w1+w2+w3+w4=1 or w1*w2*w3*w4*w5=1?

+, if we are talking probabilities applied to strange attractors.

> 	- How a new weights created?

Constitutionally, there must be a method for doing this within the
system.  Mostly, the weight values are the time rate accumulation vs.
the spend rate on voting issues.  If you are actively positive, you
affect change; the stronger you feel about something, the more likely
you are to code it (which every direction the strength of your feeling
goes.

The catch is that positive votes are inherently more valuable than
negative votes, since to obstruct progress requires a number of votes
equal to the outstanding proposals, but to move forward a proposal only
requires a number of votes equal to the proposal.  Blind obstructionism
(and blind advocacy) are uneconomical.  That's the point of having
more than one vote to potentially spend on an issue, with potentially
more issues than you can vote on all of them.  You will have to pick
you battles carefully if you want to avoid being lost in the noise.


> 	- How does one prevent factional deal making?

It won't, really, if there are places for deals to be made, and there
are accurate vote tallys published (promoting last minute bid frenzies,
auction-like behaviour).

> 	- Should weights decline over time in the same manner as
> 	  an infinitely trainable adaptive neural network?

There is a limit on the amount of weight you can throw around in a
given time because of the high water mark on the number of tokens it's
possible to build up on account.  Again, you have to pick your battles
or you will be lost in the noise.

> What about retrospectivity? On the one hand you entrench a
> pre-democratic feudal power structure and end up like Mandela's
> South Africa; a constitutionally reformed non-racially discriminatory
> capitalist society in which the blacks have all the votes, but
> the whites have all the capital. On the other (FreeBSD) hand the
> whites did all the work.

In the abstract, if you are willing to do the work, you are more
likely to throw three votes than one for a given topic.  If you are
just being obstructioninst, you will likely throw only one vote so
that you can keep being obstructionist later.

For reciprocity, it's possible to charge off percentages in the
win/lose case to bias the power concentration: if your side wins, it
costs you one less token then you voted, etc..  Again, initial
bylaws are established through constitution provision: "we have the
power that is being shared, therefore, these are the weights".


> Certainly a very interesting social engineering experiment; there
> is room here for long excursions into probability theory, game
> theory, cryptographic voting protocols (extending to protocols
> not traditionally seen as voting protocols such as Rabin's m/n
> secret sharing scheme), all excellent paper fodder.

Heh.  I was thinking more in terms of its value as a cascade trigger
to increasingly complex social organisms in the Internet implementation
space.  Representational democracies (republics, really) came about
because of rate limits on communication.  The US could not elect a
president by popular vote because there were no methods of verification,
and communication rates were limited by travel time.  Therefore, the
US has an Electoral College.  But a side effect of this structure is
a bias for bipartite seperation of interests, instead of seperation
into as many interest groups as it takes to do the job of mapping the
interest space.  This bias is not removed because the bipartite
interests have (and must continue to have) the power concentration.
This leads to continued "wasted vote syndrome", where people vote
for the lesser of two evils instead of voting their conscience... an
effect of mass psychology.  Similar pressures prevent the polling
times from being changed to opening at 8am EST and closing at 8am EST
to prevent early returns from earlier time zones influencing the
outcome of elections before people in later time zones have even voted.
For example, Ross Perot got almost 20% of the vote in the 1992
election, but 0% of the electors.  He would still have lost, given
the actual values.  There is actually a case in US history where the
winner of the electoral vote lost the popular vote... the president
was not chosen by the people, but by the electors.


> It would definitely attract a lot of welcome attention to FreeBSD.

It would be worth one or more articles in WIRED, actually, as well
as more scholarly sociology journals.  Maybe even "Wall Street Journal"
would run "Multinational Democratic State Declares Independence in
Cyberspace" or a similar silly headline.


> When viewed strictly as an experiment this idea has a lot
> of merit. If it actually pans out, then well and good, if not,
> then it could be used as some kind of Sawick poll.

Yep.  The reason I went weighted, by the way, was the volunteer
nature of the project.  In theory, number of vote tokens spent
should be proportional to willingness to actually volunteer.  As
you point out, there could be feedback here as well: for instance,
if a proposal passes, if it is completed, the tokens spent on
the vote could be refunded to those who voted for it.  If it dies
on the vine, the tokens could be refunded to those who voted
against it.  Being right would give you more license to participate,
and being wrong would not, etc..  Again, a matter for the initial
bylaws.


					Regards,
					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.