Meetings often make interlocking decisions 1 at a time through yes-no voting, with or without explicit rules of order, agendas, and votes. Items at the top of an agenda can make later options moot. At other times participants may talk about all options at once but never clearly tell (vote) their second and third choices -- so a small interest group united on a single proposal can appear to be the strongest group. And an individual with a good compromise but no ardent supporters might drop it from the discussion. Better voting rules avoid these problems.
Voting cannot research the issues, educate voters on policy, or dream up a compromise. It cannot implement a decision. It cannot prevent a majority from systematically treating minorities less well than it treats itself. But casting ballots can educate members about setting budgets (MVP) and priorities (LAR). This article will show how vote-winning proposals can find the major opinion groups and focus a discussion on the strongest idea from each group (STV) or on the most central options (Condorcet Series) or both (LER). Approval voting can find the most widely-acceptable compromise. LAR can find a fair-share solution that lets more than 1 proposal or group win funding.
Legislators know each others' priorities and control the options on a ballot. This makes manipulations much more effective and likely. Legislative voting rules have features to reduce manipulation, making them more complex than electoral rules.
What's wrong with the common agenda rules? Like plurality rules they are inherently erratic in their results and that is closely related to some types of manipulation. The books listed in the bibliography give examples of these and other problems. Here's how to milk common legislative rules:
Free riding II Do not spend political capital on items which have strong support. Here's an analogy: Cleaning the kitchen is a public good on which one can take a free ride on someone else's efforts if the household rules allow. This can lead to cynicism, resentment, and less of the public good than
Agenda tricks and failures are too numerous to list here. But some common ones are worth pointing out. A) Late introduction: The later an option is introduced, the fewer alternatives it must defeat. (But items at the top of an agenda can make later options moot.) Phillip Straffin gives an example in which any 1 of 4 options can win depending on the agenda's order of votes.
III 3 Ballots Rank 1 2 3 1st A C B 2nd B A D 3rd D B C 4th C D A
IV Agenda Orders and Outcomes
B versus A, A v C, C v D, D. wins.
C versus B, B v A, A v D, A. wins.
A versus C, C v B, B v D, B. wins.
B versus A, A v D, D v C, C. wins.
B) Pareto Criterion:In the first case D. wins even though the voters are unanimous in preferring B. over D. Thus sequential pairwise voting sometimes violates the "Pareto Criterion" which states that if all voters prefer B. over D, a voting rule should not produce D. as the winner.
C) Insincere ballots:The second agenda chose A, the last choice of voter 3. He can come out better by voting insincerely for C. on the first vote instead of his true preference for B. The result would be B versus C, C v A, C v D, D. wins.
"Our third voter has thus achieved his second choice instead of his last choice by this judicious bit of insincerity, and in the process has produced a rather undesirable social outcome. Sequential pairwise voting invites voters to think strategically and vote insincerely."
from Topics in the Theory of Voting.
D) One against all: Pitting 1 against all, often the status quo versus all proposed changes, requires the 1 to beat an alliance of all others. Therefore the 1 can lose even if most people like it more than any other 1.
E) Killer amendments: To reduce support for a bill, opponents may attach killer amendments, often taking a good idea too far. Some who backed the amendment then vote against the amended bill.
Democracy requires a clear public record of reps' votes on legislation, so voters may rate their reps. A rep's preference ballot ranks all competing proposals. It is much clearer than a series of votes on parliamentary procedures.
Condorcet's rule is not decisive if A beats B, B beats C, and C beats A. This is called a voting cycle. To resolve voting cycles, several people have created "Condorcet-completion" rules. One example is the Loring One-winner Rule (LOR). LOR uses the STV process of eliminations and transfers until 1 of the tied candidates tops each of the others. Under LOR, legislators might try to create a cycle if they guess they would lose by Condorcet and win by STV.
Conspiring to create a cycle is hard and risky in a large, diverse electorate with many candidates. If B is the centrist, supporters of moderate-left A may add their support to the few who sincerely rank fringe-right C above B, helping C to beat B
They risk the Condorcet-completion rule electing C But since few voters rank C first there is little risk she will win by STV. In fact, she will be the first elimination, then B will win. So this manipulation of Condorcet's rule would fail to manipulate STV or change the winner.
Less than 10% of simulated elections lead to a voting cycle when there are 200 voters and 4 options. But on a council with 3 factions, ties are fairly common. By-laws may send a voting cycle to a completion rule such as Copeland's, Dodgson's, Kemeny's, or LOR; to further discussion; to tabling the motion; or to a tie-breaking vote by the chairperson. My choice is LOR, because it is the hardest to manipulate; and the chair's choice of a tied option, because she has the least incentive to create a cycle. If the chair and LOR pick 2 different winners from the voting cycle, the reps use a simple majority vote to decide between those 2.
All decisive voting systems can be manipulated, sometimes. (Mathematicians have proven this.) Using preference ballots from a large association's presidential elections, with 5 candidates in each, Chamberlin, Cohen, and Coombs researched how often 9 voting systems were manipulable and how easy the manipulations were. The other rules were manipulable in all 10 tests, but STV was manipulable in only 1. They concluded:
"The most striking result is the difference between the manipulability of the Hare [STV] system and the other systems. Because the [STV] system considers only 'current' first preferences, it appears to be extremely difficult to manipulate. To be successful, a coalition must usually throw enough support to losing candidates to eliminate the sincere winner (the winner when no preferences are misrepresented) at an early stage, but still leave an agreed upon candidate with sufficient first-place strength to win. This turns out to be quite difficult to do."
As they imply, first preference is the rank most likely to be sincere on a ballot. It is hard to convince voters they will get a better result by lying about their first preferences. Merrill reports that findings by Tideman agree with these and states, "Indeed, since the Hare [STV] system appears very difficult to manipulate, strategic voting tends to be identical with sincere voting..."
Multi-winner STV is even tougher to manipulate. But funding candidates who are similar to a projected winner can reduce her 1st preferences and lead to her elimination. Condorcet's rule defeats that squeeze strategy and also reduces the free ride incentive.
The need to create a voting cycle may make LOR even harder to manipulate than STV. LOR often increases the number of voters who must be organized into a conspiracy.
A one-third minority should not have the power to enact one-third of the laws nor to write one-third of each law. Enacting a law is like electing a mayor, only 1 of many candidates can win. So the winner should be the 1 proposal that most people prefer over any other, the Condorcet winner.
Motions to delay a decision are common in legislative debates. But a simple majority should have the power to strike those options from the preference ballot. That makes a deadlock impossible unless a majority explicitly allows it. 89
Here are the traditional parliamentary motions that a council might require on its preference ballots: A) No change, B) The Main Motion, C) Amended versions of the main motion, D) Divide the Question to simplify a motion; and the motions to delay: E) Table the Proposal blocks debate until a motion to Take from Table passes, F) Postpone to a Definite Time delays debate to allow further study, G) Refer to Committee should require further study, and H) Postpone Indefinitely prevents voting but allows debate.
Combining these motions on one ballot speeds voting. It reduces parliamentary maneuvers that block the majority's will such as killer amendments or requiring a particular option to win a majority against all others put together. But it does not reduce the need for debate or change the rules and order of debate for parliamentary motions.
Condorcet's rule can also be used to amend constitutions. To become law, a proposed amendment must better all other proposals and it must win a super majority of 60% to 75% support against the "No change" option. If it does not, there is not sufficient agreement for a constitutional change.
By using Condorcet's rule, 1 election can decide among competing versions of an initiative. Initiatives are most appropriate for controlling elected officials by setting: election rules, public disclosure laws, and limits on campaigns, terms, salaries, taxes, debt, and so on.
Policy flip flops give new programs a chance to be tried, but only briefly. And anecdotes about haphazard changes are not as useful as deliberate policy experiments. A balanced council should let each side test its program on the issue or constituency area where it has its strongest support.
Abortion is a complex ethical issue, but policy options suggested to date fall along a one-dimensional line with various restrictions added from left to right. Candidate A says it should be legal, free, and encouraged for unwed teens. E says it should not be encouraged. J says it should require teen counseling and parental notification. P says it should require a 2 day wait for all women and private funding. U says it should not be allowed except in cases of rape, incest, or grave risk to the woman's life. Z says it should never be legal. It seems likely that 1 of the middle positions is a Condorcet winner with clear majority support over any other policy. But our current electoral and legislative voting rules fail to reveal the majority position.
Return to Democracy Evolves!
Move on to budgeting departments.
Go to the site outline. Read notes on A HREF="approval.htm" Approval voting
© 1997 Robert Loring Reprints are permitted.