Automatic redistricting

Anthony Simmons bbadonov at
Mon Mar 25 11:11:16 PST 2002

>> From: Joe Weinstein <jweins123 at>
>> Subject: Re: [EM] Automatic redistricting

>> Tony Simmons writes:

>> 'Perhaps the trick is to let the voters create their own
>> districts.'  [Rather than geography- or party- defined
>> partitioning of the electorate] 'it might be better to
>> think of "consituencies"'.

>> 'The obvious way to allow voters to define their own
>> constituencies is with something like STV, which allows
>> voters to group themselves through the act of voting
>> itself.'

>> I think Tony's idea is good but the last quoted sentence
>> misleadingly summarizes the proposal for implementing the
>> idea.

Yep, I wrote that pretty quick, and it's far from clear,
though it seems it was clear enough.  Actually, I wasn't
summarizing the proposal at all there, but leading up to it,
presenting STV and similar schemes as a way to define
constituencies.  My proposal really doesn't involve districts
in the usual sense at all.  Each candidate would appear on a
(potentially) different collection of ballots.  Alice would
run against Bob in some precincts and not others.

I think modifying this method to be a way of ordering
candidates on ballots in a large, fixed multi-member district
would nullify some of the advantages, though it might be an
incremental improvement on existing PR schemes.  Actually, it
wouldn't be a difference in the voting method, but a
difference in the way the ballot papers are printed.

One of my motivations for overlapping constituencies was to
eliminate the need for candidates to campaign in huge
districts, with concommitant requirement for a huge war chest
in order to be competitive.  Someone brought this up earlier;
sorry.  I don't recall who; it made an impression on me.  I'm
not pleased with the trend in the size of campaign budgets.
Choosing Congresscreatures-at-large to represent all of
California would eliminate most potential candidates who
couldn't advertise to at least ten or fifteen million people.

Think of how actual voters choose candidates, or anything
else for that matter.  Everything is dominated by

In a perfect world, where citizens devote even an hour a day
to keeping up with what's going on in the world, demand a
press that presents solid information presented in a
thoroughly professional manner, and take time to apply
serious critical thinking to elections, we might even be able
to use PR to choose candidates at large for the entire
country.  But it's an ideal.

Thus, my plan was to deliberately limit the alternatives as a
concession to reality.

And there are some other interesting properties.  For
example, each candidate would face more opponents, and each
opponent would only be an opponent in the area of overlap,
rendering negative campaigning less tempting.  And of course
gerrymandering would suffer, since the process of drawing
lines would be automatic.  In general, the idea was to
provide some of the benefits of PR without the usual price.

One aspect of Joe's proposal that is very intriguing is the
idea of using computers to arrange candidates on a variety of
continua.  There were some discussions along those lines
about a year ago, specifically about placing candidates in an
"issue space".  We might include among the "issues" such
things as party and location.

However, it really isn't necessary for any of this to be done
on the ballot, or even by the Powers That Be.  Before the
election, a voter's software could provide the information
and organize into informative displays for voters as an aid
to voting.  There's computer assisted design, computer
assisted aircraft piloting, computer assisted medical
practice ... why not computer assisted voting?

Software could use information about candidates' positions on
two or three variables of interest to a particular voter, and
display them in a 2D or 3D space, relative to an arrow saying
"You are here."  Repeated displays with different choices of
variables might provide a revealing perspective.  Such a
scheme might help a voter rationally choose thirty candidates
from among a hundred.

Of course, there are problems, as always.  How would we know
which sources of candidate data to trust?  The scheme
emphasizes measurable quantities at the expense of character,
judgement and other intangibles that constitute "leadership".
And it could encourage even more abdication of responsibility
for cultivating genuine understanding of politics, though
there is little remaining room for such abdication.

Some thoughts.

>> In fact, the potential 'constituencies' are actually
>> defined not by the individual voters but by the individual
>> CANDIDATES (i.e., by where THEY happen to live).  Also,
>> use of STV rather than some other method has nothing to do
>> with the essence of the proposal.  Any desired method for
>> choosing multiple winners - my favorite would be PAV -
>> could be used.

>> Further, one could dispense with the proposed geographic
>> criteria for shortening each voter's ballot.  In my
>> opinion these criteria are needlessly restrictive and
>> overly elaborate.  Why not instead use a single long
>> ballot? - but one that lists all candidates by place of
>> residence in a recognizable geographic order (e.g., by
>> counties or by other recognized small zones, possibly even
>> derived from a recent prior geography-based districting
>> scheme).  Then each voter could, to the extent desired,
>> factor in candidates' geographic proximity.  In
>> particular, a voter who really wants a short proximity-
>> based ballot (comprising just the nearby-residing
>> candidates) could readily find the part(s) of the full
>> ballot that really interest him.

>> For that matter, as balloting gets computerized, a
>> presented ballot could come with an indexing of the
>> candidates in terms of place of residence, political
>> party, and maybe also each of several other attributes of
>> interest.

>> Joe Weinstein
>> Long Beach CA USA

More information about the Election-Methods mailing list