[EM] Time of trouble - Premise 2

[Michael Allan]

Juho Laatu wrote:

> Still the biggest question to me is how
> to create a system that is dominant
> (determines the de facto unquestioned
> public opinion) and at the same time
> remains in close contact and attracts
> spontaneous participation of the
> citizens and is out of the control of
> the parties or people with some party
> like agenda. Maybe the most promising
> viewpoint is to see the system as a new
> "Google" or "Wikipedia" that grows
> quickly so large that it is not easy to
> control it or question its status. This
> case is a bit different of course. One
> would need wide participation (to avoid
> being just the voice of some
> "extremists"), one could need country
> specific systems or registrations,
> decisions on what elections will be
> arranged or are important, Wikipedia
> like strong rules that keep the system
> healthy etc. This is not an impossible
> task but taking into account the role
> of politics in the societies one can be
> sure that this is not straight forward
> task either.

This raises some interesting points of technical design and strategy.
See further below.
 
> IT seems to refer to voting/polling
> specific IT. If so then it could be
> called also VT.

VT would be better, yes.
 
> >   (a.1)  FS  ->  OP(i)
> 
> I read "A -> B" as "if A is widespread
> then also B is or will be widespread".
> This could cover also cases "B is
> likely to be widespread"...

Yes, it's a causal operator.  So the relation "A -> B" means "if A
then B".  (It ought to be quantified with a probability, but we have
no numbers.)

> ... and "B can be widespread".

As in "if B then A"?  Or "if B then probably A"?  No, it's not a
dependency operator.  We allow that both of these are complete and
true statements:

  A -> B

  C -> B

So B can have multiple independent causes, or generation paths.  There
is perhaps B, on some path, without A preceding.  But there is neither
A nor C, on any path, without B following.

> >   (a.2)  OP(i) + IT  ->  OP(f)
> 
> Or "->  OR(f)". I mean that votes could
> be secret too...

If the IT allows, then an individual voter might opt to express secret
OR(f) instead of public OP(f).  I believe the IT will allow this - I
believe it ought to, as you and Kristopher convinced me earlier - as a
kind of safety valve.  It does not appear in our model, yet.

But the crucial thing for this particular class of IT is OP(f).  There
must be plenty of it.  Otherwise the technical/useability challenges
of IT(a,c,p,r,s) may be too great.  (Of course, we allow other classes
of IT, and other paths to DD.)

> ... I assume in these terms
> the "publicity" vs. "privacy" refers to
> the secrecy of the ballot, not to the
> unofficial vs. official nature of the
> vote...

Correct.  OP(f) and OR(f) are votes in general.  Where they are cast
depends on the context of the eqn's.

But in our model, so far, all terms OP(f) are in context of IT
primaries, and all OR(f) in context of RD general elections - we don't
model the privacy option of IT, nor the public votes in the RD
assembly.

> > But IT(c) ensures that any OP(f) dissensus that is not grounded in
> > a social reality will be unstable.  It will therefore fall toward
> > consensus.  So we can expect consensus.
> 
> If we have IT(c) (and the other
> conditions).

Overall, we must have all IT, which is to say IT(a,c,p,r,s).  But
IT(c) is especially crucial to (a.4) and (a.5).
 
> > Any consensus that attracts a quorum of voters (however defined),
> > and that holds stable, is PD by definition.
> > 
> >   (a.4)  OP(f) + IT(c)  ->  OP(f,consensus)  =  PD
> 
> You didn't include the quorum
> requirement in the equation. One could
> add additional requirements like
> IT(u,t), IT that is used and trusted
> widely, to derive PD.

But (u,t) are non-technical, and cannot be part of IT per se.  They
would have to be separate terms.  Something like this:

  used widely  =  OP(f,quorum)

  trust  ->  OP(f,quorum)

So:

  (a.4)'   OP(f) + IT(c) + trust  ->  OP(f,consensus,quorum)  =  PD

But:

  (a.4.1)  IT(a,r,s)  ->  trust

Therefore:

  (a.4.2)  OP(f) + IT(c,a,r,s) ->  OP(f,consensus,quorum)  =  PD

However, a quorum depends on more than trust.  There are many
variables on the input side, perhaps too many to model.  And the
dynamic is rather complicated, as in the feedback loop of a.3.

Tacking to another approach: A quorum of OR(f) is a fact in RD.  This
is our reference.  So primary turnout for OP(f) must approach the
level of general elections.  Here are some strong mechanisms:

   i) Precedent: the fact of a quorum for the sake of RD sets an
      example, and people will know they're expected to follow it, for
      the sake of DD.

  ii) Herding: feedback loop of a.3.

 iii) Corraling: argument of natural monopoly for IT, hinging on
      probability that all splits of the electorate are unstable, so
      there'll be a single system. (see further below)

  iv) Novelty: people can't ordinarilly vote on norms, so it's not
      unreasonable to expect an especially high turnout in OP(f) for
      norms - primary candidate bills for the RD assembly, etc. - in
      anticipation of eventual A(n).

   v) Synergy: investment of time in (iv) generates the competence to
      also vote OP(f) for offices - primary candidate members for the
      RD assembly, etc. - and the incentive to so.  People will know
      (from b.3) that A(n) depends on A(o).

Are these mechanisms improbable?  Are there any strong reverse
mechanisms, or blocks, that would be likely to prevent a quorum?

> > On the one hand, the IT may be configured to support primary
> > elections, and thus to to nominate cross-party candidates for
> > office:
> 
> I guess these nominations could be
> unofficial.

In the sense of non-party?  They might have to be:

(I). Organized parties cannot endorse PD(o), if they wish to survive.
Although a party might announce that the IT is henceforth its official
primary, and thus PD(o) its official endorsement, there would be
nothing to prevent another party from doing the same.  In that event,
there would be no distinguishing the two.  It follows that a party
must oppose PD(o) in principle, or cease to exist.  (Is this true?)

> >   (a.5)  OP(f) + IT(c)  ->  PD(o)
> > 
> > On the other hand, IT(p) allows issues to be raised without
> > restriction.  So PD may focus on norms (laws, plans and
> > policies) too:
> > 
> >   (a.6)  OP(f) + IT(c,p)  ->  PD(n)
> 
> Before this statement the claims seemed
> to be able to cover both official and
> unofficial votes (=> PD could be
> official too). Here IT(p) seems to
> refer to cases where regular citizens
> have more power than they have in a
> typical RD.

If I understand, yes.  The status quo affords a rough equivalent to
PD(o) in the party primaries.  Also, PD(o) is cross-party, so it's
similar to general elections.  So PD(o) is conceptually familiar.

Primary *normative* voting is less familiar.  For most citizens, a
facility to nominate legislation, and it vote up, is going to be a
novelty (iv above).
 
> > To summarize (a.1) to (a.6): IT makes PD probable:
> > 
> >   (a)    FS + IT  ->  PD
> 
> It seems you assumed IT(c,p,u,t,...),
> and PD(o,n) as the result.

Yes, (a) is the reduction of a.1 to a.6, leaving out the details.
It's asserted as true, all the same.  (It's the same eqn from previous
posts.)

>                         ... One can
> interpret "->" also so that there is a
> tendency to go in this direction but
> not that the end result would be
> guaranteed or probable. All the
> conditions and equations were maybe not
> 100% strict, and many smaller conditions
> may also be missing.

"Probable" is intended.  You may counter by saying "improbable".  We
then compare reasons.  The overall argument is probable DD.

> >   (b.1)  RD + OR(i)  ->  OR(f)  ->  A(o)
> 
> This seems to assume that OR(f) always
> refers to the official elections (and
> not to the unofficial IT stuff).

Here (b.1) it is official - state general elections.

I did not make IT's provision for secret votes explicit.  I judged it
as not on the critical causal path - we can have DD with or without
it.  (But it's better with, I agree.  This essayist also agrees:

  http://rebooting.personaldemocracy.com/node/60

  (see 4, Nyms)
 
> > But note the alignment of OR(i) and OP(f) in eqn (a.3).
> > Furthermore, from (a.3), allow a rough parity between the levels
> > of primary and general turnout.  It then follows from (a.5) that
> > OR(i) is generally aligned with PD(o).
> 
> Maybe it is safer to say that the IT
> generates such tendencies.

Clarifying: From a.3 and a.5:

  (b.1.1)  OR(i)  ~=  OP(f)  =  PD(o)

It's chicken or egg.  There's an interdependency between the private
and the public, and it's made explicit at one point in the argument
(a.3), but that's really only a gesture at a relation that's almost
ubiquitous.  It's really too complicated to model.

Suffice to say, any probability of a gross distortion/skewing in the
public votes will invalidate the argument between b.1 and b.2, if not
before.

> > So (b.1) reduces to:
> > 
> >   (b.2)  RD + PD(o)  ->  A(o)
> 
> This equation is confusing. In (b.1) you
> assumed that OR(f) -> A(o). Does this
> mean that RD + OR(i) -> PD(o) -> OR(f)
> -> A(o)? In this formulation PD(o) has
> only indirect power via OR(f) to
> determine A(o). The equation works both
> with and without PD(o).

   RD + OR(i)  ->  PD(o) ...  [not]

Not true.  RD does not cause PD(o), and OR(i) alone is insufficient.
Clarifying: This is already a fact:

  (b.1)    RD + OR(i)  ->  OR(f)  ->  A(o)

Now we must get PD(o) onto the left hand side.  But from b.1 and b.1.1
(further above) it follows:

  (b.1.2)  RD + PD(o)  ->  OR(f)  ->  A(o)

Which simplifies to:

  (b.2)    RD + PD(o)  ->  A(o)

> > ...The member will be particularly interested in the level of
> > support for the bill among her own constituents, as revealed by
> > IT(a,r,s).  Considering that her re-election bid will depend (b.2)
> > on her successful performance in the ongoing primary elections,
> 
> Official primary elections of the party
> or something less formal?

Non-party, equally formal, and potentially more effective.  Take the
simplest case.  In the last election, the member had won the primary
PD(o), and therefore she won the general election too (b.2).  That's
how she became a member.  In this case, she needs PD(o) more than she
needs a party.

(II). The other cases (party endorsed members) can be dismissed as a
temporary disequilibrium.  Once a given seat is filled by public
"endorsement" in primary PD(o), then, provided that constituency
retains a quorum from election to election, that seat will henceforth
filled by PD(o).  But, from (II), all parties must oppose PD(o).
Therefore a ratchet effect will eventually clear the assembly of all
party endorsed members.  (Is this true?)

> >   (b)    PD + RD  ->  DD
> > 
> > QED. 
> 
> I think you proved the tendency (or
> possible chain of consequences or
> existence of the path) but not the
> necessity of this to happen or that
> it will happen in full strength.

I should not have used the term "proof".  The argument is only the
outline of a theory - a hypothesis - stripped down to its causal
skeleton.  It can only be proven by subsequent facts.  But if we allow
the devil's advocate to attack the argument in advance, at any
juncture, and thus to discredit it, and if he gives up, then we may
take it that he accepts the argument as plausible.

More importantly, the formal argument makes it easier to separate the
practical questions (how does it work?) from the moral (is it good?).
Those two tend to get confused.  I raised this point in a related
discussion, the other day:

  http://metagovernment.org/pipermail/start_metagovernment.org/2009-February/001186.html

> DD seems to be still defined as RD
> that acts as if it was a true DD.

In this latest version of the argument, I take the action as the
proof.  If society is *actually* steered by the people, then it is
actually DD - but perhaps not formally so.  By the same token, if it
is formally steered by the people, then it is formally DD - but
perhaps not actually so.

  (c)    RD + FS + IT  ->  DD

There is some formal DD.  The structural part is present in IT.  If
(c) is true, then IT is an institution of governance.  The functional
part - the relation RD + IT is not formalized.  (Nor was the relation
RD + party-system formalized, at first.  Only later did it form,
sporadically, in proportional assemblies.)

> Note that a true supporter and loyal
> member of some current democratic system
> could claim that the system already
> implements PD + RD -> DD (with the help
> of polls, media, party memberships, open
> discussion, free organisations,
> elections, politicians that listen to
> the opinions). The question is then how
> strong the arrows ("->") are in each case.

  RD + FS  ->  DD  [not]

Not true.  Product DD is not a fact.  Any attempt to show it by
argument would die around eqn a.6.  The people of RD do not propose
the laws under which they live, nor do they have a vote in them.
Likewise for the plans and policies of RD.  But no need for argument,
when DD is asserted as a fact of the status quo.  Here are two rough
empirical measures of DD:

(1) Initiative. Given these:

  N' potential law, conceived in the mind of a single person - but not
     yet expressed

  C' potential (max) consensus on the topic of N' - but not yet formed

We must arrive at:

  C  actual consensus formed on the topic of N'

  N  actual law, promulgated as a statute

For a typical person capable of N', how likely is N?  Will the person
even bother to express N'?  Or, at what point will she give up?  How
large, at that point, will be the actual consensus (C) of her
discouraged co-supporters?

One measure is mean C/C'.  For RD, it is probably close to zero.  For
DD, it should be near to 1.  (In this measure, there is no need for
actual N.  Even a less likely N' with a small C' will suffice.)

(2) Review.  For each law that exists in the statute books of the
state, what is the level of consensus among the people?  Likewise for
each plan that is executed by the state?  Likewise for each policy
that is followed?

One measure is n(c)/n, where n(c) is the number of consensus norms,
and n is the total number of norms.  For RD, the quantity is unknown.
It is certainly close to zero.  For a mature DD, it should be close to
1.

> > We can further reduce (a) and (b): to
> > 
> >   (c)    RD + FS + IT  ->  DD
> > 
> > Given that RD and FS are constitutional facts, the
> > immediate cause of
> > DD is the introduction of IT.
> 
> IT could be said to have potential to
> strengthen the DD properties of the RD.

Or to strengthen its democratic properties.  RD and DD are different,
except they share D.  This DD includes a formal RD (as before), and an
actual DD.  Does it include an actual RD?

One designer of RD was James Madison.  One of his design constraints
was to ensure that the new state's debts, incurred during the War of
Independence, would be repayed to the lenders (mostly landowners, like
himself).  In this case, RD had to ensure that anullment of these
debts (the popular action) was unlikely^[1].

> > ... But the people do have the power to replace the government at
> > the next general election, and their ability to make deliberate
> > use of that power is enhanced (b.2).
> 
> I read this as an interest and
> possibility to use IT to strengthen
> the "DD chain".

Yes, or more graphically, to steer RD - like with bit, reins, and
stirrup.  Without them, the horse goes where it pleases, and it's all
the rider can do to hold on.

> > It might not be worthwhile for a mailing list to start voting on
> > issues.  It is not a city or sovereign state.  Its power is
> > limited to issues of technical administration.  The important
> > issues of election methods are decided by cities and sovereign
> > states.
> 
> Are you saying that they should not
> even try? Who should then try?

Who should raise the issue, and vote on it?  I guess it depends on the
action (A), and the actor.

  A1. Enforce a voting method for Helsinki's general elections.

    actor = City of Helsinki (say), then:

    voters = People of Helsinki

  A2. Enforce a voting method for the chess club.

    actor = Chess club, then:

    voters = Members

The ideal method for A1 and A2 may differ.  Not to say that abstract
questions - ideal voting method for *all* chess clubs the world over -
cannot be raised, and voted by a body of experts.  That could be
useful.

On the other hand, Helsinki has its own voting experts.  We can expect
them to be prominent delegates in the consensus - assuming
IT(c=delegate cascade) - and thus to inject their expertise with
considerable weight.  Simultaneously, they may come to the list,
asking advice of their colleagues, and initiating discussion here.  So
the list would participate in a natural way, doing what it does best.
 
> Why would one system among the others
> emerge as the stable system? (In the
> beginning of this mail I referred to
> the Google and Wikipedia phenomena as
> potential patterns to follow.)

Why a particular one, and not another?  Google is maybe not an
example.  A search engine (Web scraper) isn't a natural monopoly.  Two
big engines are possible, I would think.

Wikipedia is an example.  Two wikipedias are not possible, except in
brief disequilibria.  I don't know if the success of Wikipedia offers
any practical lessons... I don't know who the competitors were.  But
it's a vastly different domain - encyclopedia vs. voting system.

Most of the current crop of competitors for the voting IT are actually
business firms (at least 4), and not volunteer outfits (only 2).  So:

(III).  Assume a business firm, like Google, wins the IT toss.  It
then monetizes the user interface with advertisements, in an attempt
to recoup its capital investment, and turn a profit.  The users
respond to this by voting up a consensus resolution: "No more ads,
please".  How could the firm respond?

> I referred to Wikipedia like strong
> rules above. Maybe current information
> technology allows this type of free wide
> scale voluntary activities to escape the
> traditional control points of the old
> rule.

I think so.  I think (III) is only the tip of the iceberg.  I think
the users will drive the system (IT) toward an ideal design, and
nothing will get in their way.  Not even Google.  Imagine:

  1. Fact of monopoly is, IT(GoogleVotes)

  2. PD(n) demands A(n) = IT(JoeVotes)

Now what is Google to do?

> > If PD *is* deliberately hampered, then the captor must have
> > accepted
> > the premise of probable DD, and be acting in response to
> > the perceived
> > danger of it - either in protection of its own narrow
> > interests, or of
> > the broader interests of society. 
> 
> I must note here that the original fine
> principles of democracy and its further
> enhancements are still valid but somehow
> politicians and others have found ways
> to play the game in a somewhat less ideal
> way than originally (idealistically)
> intended. I'm sure that there will be
> similar tendencies also in the future.
> The new system must be planned well to
> avoid this type of pitfalls as
> efficiently as possible.

Hopefully, the system can adapt to unforseen bumps.  There's a
three-way tension, and any designer's time is going to be divided by
these priorities:

  1. Design well.

  2. Deploy fast.

  3. Look ahead.

I don't worry so much about 1 and 2 - it's the usual professional
race, and we all just run it - it's what we do.  And I don't worry
about 3 in the sense of running into a blind alley - a bad design
pocket.  We'll just turn around, and race off in another direction.  I
think the only things that can go wrong are non-systemic externalities
- the "troubles" of the parent thread.
 
> I mentioned the possibility of the chain
> towards DD being weaker than certain and
> automatic. The inevitable corruption
> and/or inclusion of various interests,
> less than perfect understanding and
> monitoring etc. is another potential
> factor that may weaken the system.
> 
> At this point I might also mention one
> more topic that I consider important.
> That is the role of the chairman and
> opinion formation in the system. One
> can not trust IT(p,c) to be a sufficient
> mechanism to keep the discussion focused
> and lead to selection of the best
> formulations of the questions and
> modelling of the society and topics in
> question. Better new mechanisms are
> needed also on that sector.

Here you diverge from the ideal?  IT(p,c) is rather strong.  It says,
to paraphrase it:

  IT(p) Anyone can raise a new issue (class of candidates), or an
        option of that issue (instance of candidate)

  IT(c) If society has the potential of consensus on the issue, then
        actual consensus will be expressed - without suppression of
        dissent

On the surface, these should suffice.  We desire a power-free system -
in sociological theory, an ideal speech situation - with no intrusions
of power.  To the extent the chairman had sytematic power (force or
threat of force), the resulting consensus might be an artifact of that
power.

On the other hand, if the chairman is external to the voting IT, and
does not intrude any power, then it is OK.  So the chairman may
convene a discussion roundtable on the issue.  He may enforce all
kinds of rules during that discussion.  That discussion need not, in
itself, be an ideal speech situation.  It can be anything, provided:
a) people are free to leave the table; and b) their access to the IT
(the ideal) is not restricted.  (This would be an example, I suppose,
of what the essayist Mark Murphy has called an "ecosystem", of which
the IT and its components would be a part.  Another term for that
ecosystem is just the "public sphere".)


[1] I think this is from John Dunn's "Democracy", which traces the
    history of the word, and its changing meaning.

-- 
Michael Allan

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