I think that's a good lesson:

In a proportional voting system, if the mapmaker has the discretion to change the district magnitude (this area will have a three-seat district, this area will have a five-seat district), then gerrymandering is relatively easy.

If all districts must have the same number of representatives, then in a proportional voting system, there is much less opportunity for gerrymandering. In other words, the mildest form of proportional voting -- three-seat districts -- is must less susceptible to gerrymandering than a single-member district map *if* every district elects three members.

I wonder if Prof. Rush would agree with that?

Dan

From: Tom Round <t.round@griffith.edu.au>
To: "Dan Johnson-Weinberger" <proportionalrepresentation@msn.com>
CC: election-law@majordomo.lls.edu,rick.hasen@mail.lls.edu
Subject: RE: Message from Tom Round re: Charles and single-member districts [2]
Date: Fri, 13 Feb 2004 11:55:24 +1000

Thanks, Dan, useful info. I agree completely that uniform three-seaters
would be harder to gerrymander than uniform single-seaters. However, with
small and/or even district magnitudes (ie, no. of seats per district),
it's being able to mix and match that gives some credence to the claims
that MMEs can be easier to rig than SME boundaries.

Once over the 5-seat minimum that's generally considered essential for
"good PR", the difference in quotas -- even in jumps of two from one odd
number to another -- rapidly diminishes. The quota in a five-, a seven-
and a nine-seat district is 16.667%, 12.501% and 10.001% respectively.
These differences are within the "ballpark" of what a normal two-election
swing of votes between parties would produce anyway -- and are also only
just above the "radar" of what opinion polls can reliably detect (about
3%, I believe). So it becomes extremely difficult -- almost impossible --
to pre-determine the result by the precise placement of boundaries.

On the other hand, if one used a winner-take-all system (eg, block vote or
Approval Voting) and multi-member electorates, drawn by use of discretion
(as opposed to, eg, "Every county elects one Delegate per 100,000
residents or fraction over 50,000"), you could do enormous damage to your
opponents even with completely equipopulous districts (ie, if every
one-seater had 100,000 residents and every seven-seater had exactly
700,000). You'd consolidate your own safe strongholds into 10-seat
districts while splitting your opponent's areas into single-seaters. The
result would make the Electoral College look like the Knesset.

regards,
Tom

> "Dan Johnson-Weinberger" <proportionalrepresentation@msn.com> 13-02-2004
05:00 To: rick.hasen@mail.lls.edu, election-law@majordomo.lls.edu cc:
t.round@griffith.edu.au Subject: RE: Message from Tom Round re: Charles
and single-member districts

Thanks for that insight. If multi-seat districts that elect a political
minority (with cumulative voting rights or the single transferable vote)
are almost gerrymander-proof with 5-7 seats, and single-member districts
are impossible *not* to gerrymander in some way, doesn't it stand to
reason that three-seat districts are less susceptible to gerrymandering
than single-member districts?

To the equipment question that Tom asks (and thanks for making me a
candidate):

The equipment vendors are unreliable sources of information as to the
actual abilities of their equipment, unless a local government asks them.
And then the question becomes how much the vendors can get the local
governments to pay for the new feature.

The newer optical scan machines have enough memory in their computers to
store an image of each ballot, and thus have the hardware to process the
voting system you propose. They don't, of course, have the software to
process the votes, and developing the software is another expense the
vendors try to put onto the local government. (We're a little suspicious
of this, as the global companies all compete for Ireland's contract, so we
think they've got the software to run STV and instant runoff voting
elections). The touchscreens almost all have enough memory to do it as
well.

So, any new equipment in the United States has the potential to process
ballots that way, but none of them have the software to do it.

And yes, Peoria's use of cumulative voting rights is as you described --
with five candidates running at-large, if a voter only casts one vote for
a favourite candidate, it is counted as five votes. It is invisible to the
voter, in a sense, unlike the Amarillo Texas ballot style, where a voter
must cast three or four votes by one candidate (and each candidate has
three or four spaces where a voter may cast her votes).

Dan

>> Hello from the antipodes,

Whether multi-seat electoral districts with proportional (STV, party-list)
or semi-proportional (cumulative or limited vote) voting systems are
easier or harder to gerrymander than are single-seaters, depends heavily
on two factors: the minimum number of seats per district, and whether even
numbers are allowed.

There is a huge difference between, say, Ireland (with 3-, 4- and 5-seat
districts) and Tasmania, Malta and the Australian Capital Territory (which
use only 5- and/or 7-seaters). Once every district has at least five
seats, and an odd number (so that a party or coalition with over 50% of
the votes in a district is guaranteed more than half of its seats), the
potential gains from gerrymandering boundaries, in the sense of
deliberately manipulating where the borderlines are placed, dwindles
almost to vanishing point.

By contrast, Ireland in the 1970s saw the refinement of the art of
"Tullymandering" (named after the Minister who drew up the electoral
boundaries for the 1974 [?] Dail Eireann election). The secret here is to
put as many 3-seaters as possible in your own stronghold areas, and as
many
4-seaters as you can in the opposing party's fiefdoms. Suppose that, based
on population, each side's stronghold regions are entitled to 24 seats out
of 48. You divide your own area into eight three-seaters. PR gives you two
seats out of three in every district, making the result there 16-8. But
you divide your opponents' areas into six four-seaters. Under normal
circumstances, each electorate is deadlocked 2-2 in its representation. So
both sides win 12 of 24 seats. Overall, you've won 28 out of 48 seats with
as little as 50.1% of the votes in your half of the state and 40.1% of the
votes in your opponents' half.

[I did say "under normal circumstances" -- the Tullymander backfired
because the Minister's party polled so poorly in the four-seater districts
that it dropped below the 40% threshold needed to hold two seats, and so
was annihilated!]

Once you get over 5 seats, this becomes harder, especially since larger
multi-seat electorates will (on balance) each contain a larger percentage
of the total electorate and therefore be more evenly balanced. (If your
state has 49 single-seaters, there'll very probably be a handful that are
70% safe for one party; but it's very likely, if the state had seven
7-seaters, that in a district accounting for 14-15% of the whole state the
"red" and "blue" voters will cancel out more evenly).

Regarding cumulative voting: Here's a query for those who know more about
US voting machines than I do ... Suppose one wanted to introduce a crude
form of "semi-preferential" Cumulative Vote. The voter could approve as
many candidates as she liked (or up to the number of seats). Her vote is
then divided equally among the candidates she votes for -- the system used
in Peoria, I believe -- instead of dividing your votes 2-2-1, say, you
vote for three candidates and they each get 0.333 of your vote.

Then the lowest candidates are eliminated one by one. If one of the three
candidates you vote for is eliminated, your vote is re-adjusted so as to
give 0.500 to the two surviving candidates only. This continues until the
number of candidates equals vacancies plus one, or until the required
number of candidates have reached the Droop quota. (Note that with a
single seat this would reduce to simple Approval Voting.) So if your party
ran, say, five candidates for five seats, you could vote for all of them
and the system would reduce the number of surviving candidates to the
maximum number your party's total vote was capable of electing.

What I am curious about is this: Are existing US voting machines capable
of, say, scanning ballots and saying "43,838 ballots contain votes for
Smith, Jones and Brown... 31,622 ballots contain votes for Smith, Jones
and Weinberger ..." -- or is that too complicated?

Cheers all Tom

_________________________________________________________________
Create your own personal Web page with the info you use most, at My MSN. http://click.atdmt.com/AVE/go/onm00200364ave/direct/01/