[EM] Re: UK - Lord Jenkin's Proposals
New Democracy
donald at mich.com
Sat Sep 26 04:59:10 PDT 1998
---------- Forwarded Letter ----------
Date: Wed, 23 Sep 1998 11:21:27 -0500
To: donald at mich.com (New Democracy)
From: Charles Fiterman
Subject: Re: UK - Lord Jenkin's Proposals
As a Lake County Illinois coordinating Election Judge I view these complex
proposals with horror. Even the very simple mechanical procedures now in
place are hard to administer. We average 900 voters per precinct and four
election judges per precinct. Even with modern punch card systems it is
common for judges to be still balancing their books at midnight. At least
common from my point of view as I'm one of the people who goes out to help
when they have trouble. Not common from the judges point of view or we
would have no election judges.
Any time the voter must write on anything for any reason bad things will
happen. For example the voter must sign his name, judges must look up that
name and confirm that it matches the signature. This is always a judgment
call but thankfully it tends to be an easy one. If there are ten write in
votes for an entire precinct you may expect them to eat as much of the
judges time as all of the mechanical voting.
Lord Jenkin's proposal has people writing many numbers on a ballot. I can
see the judges far far into the night, "Is that a one or a seven or a two?"
The potential for error and fraud are endless.
Getting people to put a punch card into a slot and not fold it first and
not put it in upside down even though that's impossible is very hard when
you are dealing with 900 voters and a secret ballot. We have immigrants
from countries where democracy is a sham asking for help in filling out
their ballots because they expect the judge will fill it out for them and
then pay them. And then we have crazy people and crooks and people who's
English is so limited they legitimatly can't read the ballot and we have to
explain it and not give advice.
And worst we have crooks in the legislature manipulating the rules to gain
advantage. And people ask why is this silly rule the way it is and we have
to say "That's the law, that's how it is." not "The Republicans think their
voters are more dedicated and less likely to walk away when they see a
line." Yours is a very Republican proposal, it would create a long long
line. I suggest you go to the Illinois legislature and propose it to the
Republicans.
So let me propose something simple which has similar effect.
1. Getting on the ballot should be trivial, little more than a few forms.
The long lead times are an antique in the computer age. Making it easy and
quick to get on the ballot eliminates all reason for write in votes. No
more write in votes.
2. People should be able to vote for as many candidates for an office as
they wish. This is called approval voting. If there are five candidates and
you approve of three you can vote for those three. This allows people who
like a minority party to vote for it and not feel they have wasted their
vote. Modern punch card and optical scanning systems make this no harder to
administer than traditional systems.
Charles Fiterman
- - - - - - - - - - - - - - - - - - - - - - - - -
Dear Charles Filerman,
Thank you for sending me your comments on Lord Jenkin's Proposals.
Your point on no more write in votes is new to me. I did not know that
it causes more trouble than its worth. I will have to study the policy.
I do not share your support of the Approval Voting method. I do not
consider it a valid method.
Most of time, Approval Voting is unable to give us a winner with a
majority - the problem is in the math. When we increase the number of votes
per voter we increase the majority requirement. When the votes being used
equal twice the number of voters, then it is mathematically impossible to
have a majority winner because the majority requirement becomes equal to
the total number of voters plus one. An example may help to explain.
Suppose the following results of a single seat election:
18 AB 13 BA 17 CA
17 AC 20 BC 15 CB
----- ----- -----
35 A + 33 B + 32 C = 100, with 51 being a majority
We do not have a majority on the first count of the ballots. The
method of Approval Voting will treat every choice as a vote and add them
together to yeild the following:
35 A 33 B 32 C
13 18 20
17 15 17
----- ----- -----
65 A + 66 B + 69 C = 200, with 101 being a majority
The Approval Voting people will claim that candidate C is the winner.
But I contend that 69 is not a majority. But, it is impossible for any
candidate to have a majority of 101 because this election only has 100
voters. No candidate can receive more votes than voters.
Now the Approval Voting people will also claim that the majority should
be based on the number of voters and not the number of votes. I disagree,
but let us see what we will have when we set the majority at 51 of 100
voters. We end up with all candidates with a majority - and the "Lead"
candidate is the winner. The method known as First Past the Post also picks
the lead candidate as the winner. The more things change, the more they
remain the same.
The position of the Approval Voting People is a corrpution of
mathematics - majority must be based on the number of votes in use at the
time the winner is determined.
There is a point that you should be aware of before you choose this
method. Are you aware that the choices you make are not ranked? Your second
choice will have the same affect, at the same time, on the election as your
first choice. Every voter could switch their choices and the outcome would
be the same. This condition may be acceptable for when you rate two
candidates the same, but most people want to prefer one candidate over all
others. If you prefer your first choice then you should know that your
second choice would work against your first choice.
If you find yourself in an Approval Voting election and you wish to
prefer one candidate, you do it by not making any more choices than the
one. And it follows that if everyone did this the election would turn into
First Past The Post.
One more point about Approval Voting. It is not netural. The lower
candidates on the first count of the ballots have a mathematical greater
chance of gaining votes in the lower choices.
While we are here we may as well try other methods with this same example.
Let us try Choice Run-Off aka Alternative Vote - also known as Instant
Run-off or as Bottoms Up or as Majority Preference Voting or as Majority
Preferential Vote or as Single-Seat STV or as Single-Winner STV - all eight
are the same. Single-Member STV is also the same but it refers to members
of a lawmaking body being elected from single seat districts/ridings.
Bottoms Up is the name for the same routine used to elect more than one
member per district/riding.
Choice Run-Off will drop candidate C, the lowest, and transfer the
votes to the next choice. Example and solution below:
18 AB 13 BA 17 CA
17 AC 20 BC 15 CB
----- ----- -----
35 A + 33 B + 32 C = 100, with 51 being a majority
17 A 15 B - 32 C
---- ----- -----
52 A + 48 B + 0 = 100, with 51 being a majority
Choice Run-Off claims candidate A to be the winner - with a majority.
In Choice Run-Off the choices are ranked. Your second choice will not
work against your first choice. Your vote belongs to your first choice and
will stay with your first choice as long as your first choice is a
contender.
Another thing I like about Choice Run-Off is that if we work the
method all the way down to one candidate we get an order of how well each
candidate has done in the election.
Choice Run-Off is netural if you accept the policy of dropping the
lowest candidate. But a case can be made that sometimes when we drop a
candidate we are not being netural. Consider the following election
results:
2003A, 2002B, and 2001C Are we treating candidate C with equality
when we drop him - are we being netural? Is this a flaw of Choice Run-Off?
Is it a rare flaw?
Condorcet avoids this possible flaw of Choice Run-Off because
Condorcet does not drop candidates, but Condorcet has its own flaw - and
the flaw is not rare - the circular tie occurs too many times to allow
Condorcet to be used.
A man that goes by the internet name of DEMOREP made a study of
circular ties using random examples. He came up with the result that
circular ties occur one third of the times. (DEMOREP means DEMOcratic
REPresentation)
Condorcet:
This method called Condorcet will divide the candidates into every
possible pair of candidates - called "pairing". Each pairing is going to be
a contest between only two candidates. The other candidates are dropped and
their votes transferred to the candidates in the pairing. The "head to
head" contest in the pairing will usually result in a winner with a
majority. If a candidate wins each of the pairings that he is in, he has
won contests over all the other candidates and that makes him the
"Condorcet Winner".
It is argued that because the Condorcet Winner has won each of his
pair contests by a majority he has won the election by a majority.
Also it should be noted that in Condorcet your second choice does not
work against your first choice.
Condorcet is netural because each candidate in each pairing has equal
matematical chance to all the votes that are transferred from the
eliminated canadidates.
So far, Condorcet has a lot going for it. It would be hard to argue
against it as a single seat election method - except -
- except for the circular tie - the curse of the Condorcet method. One
out of three Condorcet elections will end up in a circular tie. I will now
show you a circular tie. I'll be using the same election example as above -
which was designed to produce a circular tie. This three candidate election
will only have three pairings. A four candidate election would have six
pairings.
Election Pairing Pairing Pairing
Example A B B C A C
18AB + 17AC = 35 18 17 35
13BA + 20BC = 33 33 13 20
17CA + 15CB = 17 15 32 32
---- ---- ---- ---- ---- ----
52A 48B 51B 49C 48A 52C
Candidate A wins over candidate B, candidate B wins over candidate C,
and candidate C wins over candidate A. This is a circular tie - no
candidate has won all his pairings.
There have been a number of tiebreaker methods offered. The tiebreaker
methods do not depend on the number of pair run-off wins, and would not be
decisive if they did. Condorcet himself intended Least Absolute Margin of
Defeat as the tiebreaker. A margin is the difference between the two
candidates in a pairing. Absolute Margin is the total of all the margins of
defeat of a candidate. The winner is the candidate with the smallest total
of defeats.
In this example, Camdodate A was defeated by four votes, B by four
votes, and C was defeated by two votes. Condorcet would say that candidate
C is the winner of this election, but I say that his tie solution is
arbitrary. Some other numbers could just as easily have been used to select
a winner - and those new numbers may pick a different candidate - it's
arbitrary. All circular ties solutions are arbitrary. Besides, there is
nothing in any solution to which we can point to and say that a candidate
has won by a majority - where is the majority in this solution? There is no
majority in a circular tie. It is the same as driving an auto into a dead
end street. It is not a solution to try to drive across some field beyond
the dead end. It is best to stop and then back out of the dead end. The
circular tie should be avoided. This is why I suggest a combination of
Condorcet and Choice Run-Off.
If anyone is still not convinced that Choice Run-Off is the best
method in spite of its possible flaw, I would like to suggest a combination
of Condorcet and Choice Run-Off. It would work as follows:
Step 1. Start with the first count of the ballots.
If no majority winner, go to Step 2
Step 2. Use the Condorcet method - which should yield a majority winner
two out of three times. If we get a circular tie, stop,
go back to Step 1. and drop lowest candidate.
Step 3. Second count, less one candidate and new vote counts. If no
majority winner go to Step 4.
Step 4. Attempt Condorcet a second time - which again should yield us
a majority winner two out of three times. If again we get a
circular tie, stop, go to Step 3 and drop the new lowest candidate.
This routine is to be repeated until we have a winner - either as a
result of Condorcet or Choice Run-Off. Whichever method gives us the
winner, that candidate will have won by a majority. This combination avoids
the circular ties of Condorcet and reduces the chances of any effect from
the possible flaw of Choice Run-Off - but this combination of two methods
is not as simple as using Choice Run-Off by itself.
The supporters of Choice Run-Off, aka Alternative Vote, know of its
posssible flaw, but are willing to accept it because Choice Run-off will
produce more winners with a majority than Condorcet. Also, Choice Run-Off
is simple compared to Condorcet and its host of circular tie solutions.
Regards,
Donald Davison
---------- Forwarded Letter ----------
Date: Thu, 24 Sep 1998 12:48:37 -0500
To: donald at mich.com (New Democracy)
From: Charles Fiterman
Subject: Re: UK - Lord Jenkin's Proposals
At 12:33 PM 9/24/98 -0400, you wrote:
>Dear Charles Filerman,
>
> Thank you for sending me your comments on Lord Jenkin's Proposals.
>
> Your point on no more write in votes is new to me. I did not know that
>it causes more trouble than its worth. I will have to study the policy.
Let me give you the worst case. This is all Illinois Law, I don't know what
this means in the UK.
The voter casts a straight party ticket and one write in, say for mayor.
Since the voter's wishes are clear we must honor them. But he has voted
twice for mayor, once as the straight party punch card ballot and once as a
write in.
We must now mark the ballot as spoiled, with the explanation, and duplicate
it converting the straight party punch into the votes it represents minus
the vote for mayor. This requires a mixed pair of election judges,
Republican and Democratic. The process takes about 15 minutes. Imagine
doing this at about 9:00 PM after getting to the polling place at 5:00 AM
and working straight through. Also most of our election judges are retired
people with weak eyes getting a few extra dollars for the days work.
But the worst is people hear we are recasting spoiled ballots and don't
realize the details or that this is all absolutly required by law. They
think something corrupt is going on. I hear about this almost every year
from very shocked voters. The process must not only be honest it must
appear honest.
It is so rare for a write in to win that it makes the national news,
encouraging yet more write in candidates.
The reason for write in candidates is that printing up the ballots used to
take weeks and if you wanted to get on the ballot after the printing
process had started a write in was the only way. The underlying problem is
antique, the printer can do all the work in a day.
So I propose that you simply need to fill out some forms to get on the
ballot and need to get them in two days before the election. If we ever
have computerized voting any time before the election will be fine. But no
write in candidates, they win so rarely that eliminating them does little
to harm the system.
I wish everyone with ideas for new ways to organize democracy would work in
the system or at least learn it so they understood the impact of their
proposals. While logically sound Lord Jenkin's Proposals are about the
worst I have seen from the viewpoint of implementation. I hope you
distribute my objections with your other comments.
Charles Fiterman
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - em - - -
Dear Charles Filterman,
Computers will make it possible to implement logically sound reform
election methods like Lord Jenkin's proposals.
I would like to see others comment on your policy of No Write In
candidates.
I keep thinking about the example of a front runner being shot and
killed the day before the election - his supporters need a way to vote for
their runner-up. Although, if my policy of having the primary included in
the general election was being used, their runner-up would already be on
the ballot.
I need more input on this policy of No Write In Candidates.
Regards,
Donald Davison
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
/// N E W D E M O C R A C Y ///
\\\ Home of Citizen's Democracy http://www.mich.com/~donald \\\
/////////////////////////////////////////////////////////////////
More information about the Election-Methods
mailing list