Michael Allan mike at
Sun Aug 9 11:38:16 EDT 2009

Hi Thomas,

> Mike and I had an off-list conversation quite a while ago, but somehow
> it dropped of,  because I couldn´t really find clarity and also
> because it didn´t seem so important at that time.
> Now, with the delegation-tree to play with, it got much clearer for
> me. And I think it has some importance now too.
> The question is: Do we really need an ?Issue? under which we pool
> certain drafts of texts (which then form the specific delegation-tree
> for that Issue)?

This is attractive, but I foresee a problem.  It's attractive because
it holds the promise of a simpler, more elegant solution.  We remove
the boundaries that separate polls (formal decision processes) from
each other, and we allow the separate issues (outcomes) to sort
themselves out.

So I sketched a design.  There is no separation of polls.  Voters have
unlimited votes which they can cast as they please among the various
candidate positions (draft proposals, electoral candidates), but at
most one per position.  Multiples votes are automatically discounted.

The problem I foresee is that, if there are thousands of votes for a
particular position (Joe as City Councillor, or Mike's idea of Green
Roof Credits, etc.), what are the alternative choices?

Here two candidates are neck and neck.  There is dissensus.  But we
only see the dissensus because we view the results in the context of
an individual poll.  If we remove the poll boundary, then we might see
24K+ votes for one candidate and think that's an electoral consensus
(false).  The same problem crops up in normative polls.

So there's a practical problem in removing the poll boundaries.  The
alternatives are hidden from the voter.  Unable to see the rival
candidates, she cannot make an informed decision for one or the other.
She might vote for one candidate today and another tomorrow, all the
while being unaware that there are multiple choices, and she has
contradicted herself.

The theoretical side of the problem is that a decision process has
real bounds because the issue is real and precludes other rival
issues.  There can be only one city councillor for Ward 20 (one issue)
and we must decide who it is to be.  There can be only one section in
the municipal code detailing the provisions in lieu of green roof
space (whether cash-based or credit-based) and we must decide which it
is to be.  The technical requirement that follows from this is to lay
out the rival alternatives side by side.  Then the voters can see them
clearly.  We do this with separate polls, each formalizing a single
process of decision, with a single issue at stake.

> To avoid misunderstanding: I´m not talking about headlines for drafts,
> which would ideally be the same for all the drafts addressing the same
> issue. I mean the category ?issue" which technically separates one
> delegation-tree from the other.
> Is that really necessary?

(You mean that "separates one delegation-forest from another".  Each
 rival candidate is the root of a separate tree in the poll forest.  A
 voter has a familiar bird all her own, who can hop from tree to tree,
 or branch to branch.  Or she can just hover in the air, uncommitted.

 But she cannot fly to another forest.  The voter has a separate bird
 in each forest.)

> Because soon there will be many thousands of those issues. What to do
> then, if two issues were created, which actually are the same? Ask all
> the voters to merge the issues?

Yes, I think that's best.  The leading candidate in one of the polls
emails a URL to each of her immediate voters (5-20), "I've just copied
my draft over to the other poll.  Please 1) follow this URL and press
the Vote button to copy your vote over; then 2) follow this other URL
to the old poll and press Unvote."

The voters who are in fact delegates will repeat the same process for
each of their own voters.  And so on.  The old poll will then be
abandoned, and its turnout will drop to near zero.  In this way, two
polls can be merged by the voters.  A single poll can also be split
into two polls.  The process is the same, in both cases.

Polls can also be transformed in scope or theme.  For example, an
abandonded poll might be reused for an entirely different purpose.  Or
a poll with the theme of improving a particular local park (so a plan)
might expand its scope to rule on all park improvements (so become a

> Why not simply have one single list of all the drafts all the people
> have drafted and additionally have a tool to sort it by different
> criteria and to filter it...  And every user can give tags to every
> draft...  So one can easily filter out all the drafts which for
> example are tagged with ?Canada?, ?Toronto?, ?environment?, ?green
> roofs?, ?Municipal Code Bylaw 583-2009? and ?Chapter 492?. So one
> will find Mike´s, Fred´s and Thomas´s ?Greener Green Roofs?-drafts.
> And fictional Juanitas, Peter´s and Julia´s ?Green Roof ?
> Amendment?- drafts etc.

This is a classification scheme, and I agree the voters are going to
need this.  Although it's not absolutely essential, it will be useful.
I think there will even be multiple classification schemes for various
special purposes.  So they will be outside of Votorola.

There's also the poll boundaries.  These aren't a scheme for
classification, but rather for identification.  They appear to be
mandatory (reasons above), so this is built into Votorola.

> Another point which I find very import: How will we otherwise
> realize delegation across several issues? ...  For example: I want
> to delegate my vote concerning all environment- related issues in
> Dresden to Peter.  I want him to do the work for me of finding those
> issues and drafts, creating new ones, deciding on them and voting on
> them...  With the tags it seems easy, doesn´t it? ...  In all the
> drafts worldwide which are tagged "Germany", ?Dresden? and
> ?environment? Peter is my delegate.

This is possible with Votorola as is.  We spoke of it earlier and I
called it "procedural abstraction".  Basically you have an automated
voting agent working for you.  It subscribes to a particular
classification scheme for polls, and it automatically casts your
initial vote in each poll (subject to configuration and manual
overrides) according to your rough delegation preferences.  So your
vote for all environmental polls goes to Peter, initially.

This agent will not be built into Votorola, so you'll be able to
choose from different agents offered by competing suppliers.

> What do you say? Am I missing something?

I didn't see it till I sketched a technical design.  The idea is
attractive, but it looks like the poll boundaries are essential.  We
need to show alternative choices and allow for the possibility of
dissensus.  Is there any way to do this without clearly defined poll
boundaries?  If dissensus is hidden or obscured, I doubt there can be
a real consensus.

Mike Allan

Toronto, 647-436-4521

More information about the Votorola mailing list