![[Home]](http://www.bayleshanks.com/cartoonbayle.png)
Table of Contents for Governance Systems Design
broad definition: a procedure whose input is the preferences or opinions of group member, and whose output is a group decision
non-scalability of strict consensus
get-out-the-vote is very important, in macro and in micro (eg student govmt)
electoral turnout boycotts fake/corrupt high turnouts
fair voting external watchdogs
control over nominations -> circular flow of power
todo watch https://www.coursera.org/course/votingfairdiv
first-past-the-post plurality defects: spoilers no compromise candidates two-party dominance
indirect elections majoritarian unless some tweak eg voluntary constituencies watch out for top-down control of nominations
single-seat vs multi-seat; multi-seat via at-large or via constituencies gerrymandering in the sense of constituency shape; gerrymandering in the sense of non-proportionality eg the Society of Mind example voluntary constituencies
condorcet defects: does not fulfill condercet criteria in the presence of strategic voting? (look up result; was it in logic of lawmaking? or probably [1]) STV
score voting approval voting defects: non-clarity of preferences/strategy?
delegable proxy by topic? ('liquid democracy')
runoffs (must NOT be instant)
the concept of 'strategic voting' (tactical voting) https://en.wikipedia.org/wiki/Tactical_voting
voting criteria; independence of clones (indep. of irrelevant alternatives?); arrow's theorem; the condorcet criteria ; https://en.wikipedia.org/wiki/Gibbard%E2%80%93Satterthwaite_theorem ; proportionality (multiple criteria, but esp. mention the one that a faction of proportion X could control a proportion of approximately X of legislators)
todo read: https://en.wikipedia.org/wiki/Template:Decision_theory_paradoxes
simple majority: policy can turn on a dime
secret ballot
(secret ballot for representatives/proxies?)
compulsory sufferage? does it increase or decrease uninformed voting? (increase, b/c apathetic ppl vote; decrease, b/c it becomes rational to vote)
aside: prediction markets
ka-ping yee's page
cumulative voting
instant runoff
Is there such a thing as "the will of a group", where we assume that the voting expresses the will of the group? Similarly, does it make sense to say that a group "believes" something is true? Maybe not, at least not all the time. In the next two sections we should that, even if everyone in a group has a consistent preference and consistent beliefs, it's possible for the group to have inconsistent preferences and inconsistent beliefs.
Imagine a group of 3 people, Alice, Bob, and Charlie, who are voting on 3 alternatives, X, Y, and Z. Alice prefers X>Y>Z (that is, she likes X the best and Z the worst); Bob prefers Y>Z>X; and Charlies Z>X>Y. What is "the will of the group", that is, what is the GROUP's preference between X, Y, and Z? To find out, let's have a vote between two of the options, and then have another vote between the winner, and the third option.
Let's say the group first holds a vote between X and Y. Alice, who prefers X>Y>Z, likes X more than Y, so she votes for X. Bob, whose preferences are Y>Z>X, votes for Y. Charlies, Z>X>Y, votes for X. So the votes are 2 for X (Alice and Charlie voted for X), and 1 for Y (Bob), so X wins.
Next, the group holds a vote between the winner, X, and the other choice, Z. Alice (who prefers X>Y>Z) votes for X. Bob (Y>Z>X) votes for Z. And Charlie (Z>X>Y) votes for Z. So Z wins.
So the group prefers X to Y, and Z to X. So the group's preference so far appears to be Z>X>Y; Z seems to be the best and Y the worst. But not so fast; what if the group had held a vote between Z and Y?
Alice (X>Y>Z) votes for Y. Bob (Y>Z>X) votes for Y. And Charlie (Z>X>Y) votes for Z. So Y wins, with 2 votes to Z's 1 vote.
So, the group as a whole prefers X to Y, and Z to X, and Y to Z; Y>Z>X>Y. Y appears to be both best and worst at the same time. If we draw a figure with each of the options and with arrows pointing from each losing option to the option that beats it, the arrows go around in a circle, which is why we call this "cyclic".
Notice how in the previous, if we had only held a vote between X and Y and then between the winner (which was X) and Z, without ever holding a vote between Y and Z, then we would have concluded that Z is the best. If this group were a legislature, and if we started out with the law saying X, and then we held a vote on the question "should we change the law to say Y?" (which is implicitly a vote between X, the status quo, and Y), the vote would turn out X>Y, that is, "no" would win. And then if we held another vote on the question "should we change the law to say Z?" (which is implicitly a vote between X and Z), the vote would turn out Z>X, that is, "yes" would win. Then perhaps the legislature would stop, because the resolution Y had already been considered and lost; so at the end the law would say "Z". But if we had held the votes on Y and Z in the opposite order, that is, if we had voted on Z first and then Y, the results would have been "yes" and "yes", so at the end the law would say "Y". So the order of presenting the options to vote determined the outcome. This is a general principal; if the group has cyclic preferences, then he who controls the agenda controls the outcome.
Note that, in this sort of arrangement, an option has more chance of sticking if it is presented later. For this reason, some people (including me) advocate voting systems in which a choice between the winner and the status quo is always presented last; the argument is that, if the group has cyclic preferences, we may as well just stick with the status quo.
A similar argument shows that, if you imagine groups to have not only "group will" but "group beliefs" which can be measured by voting, then even if every individual in the group is rational, the group can be irrational, in the sense that the group can believe things, and believe that their beliefs logically imply other beliefs, and yet not believe in those other implied beliefs. This can occur even if everyone in the group agrees on which beliefs logically imply other beliefs.
Imagine that there is a group consisting of Alice, Bob, and Charlie. There are three beliefs; X1, X2, and Z. In this example, let's say that logic implies that if X1 and X2 are true, then Z must be true, and Z must be false otherwise. Everyone in the group agrees that IF (and only if) both X1 and X2 are true, THEN Z must be true. Alice believes X1 is true but X2 is false; so she also believes that Z is false. Bob believes that X1 is false but X2 is true; so he also believes that Z is false. And Charlie believes that both X1 and X2 are true, so he also believes that Z is true. All three of these people have consistent, logical beliefs. But look at what happens if we take a vote.
Alice and Charlie both think that X1 is true (and only Bob thinks it's false; so X1 wins with a vote of 2 to 1). And Bob and Charlie think that X2 is true (so X2 wins its vote). So in a vote, both X1 and X2 are declared true. But Alice and Bob both think that Z is false, so Z loses its vote. So the group as a whole votes for X1=true, X2=true, Z=false. Even though everyone in the group agrees that you can't have Z=false if both X1 and X2 are true.
how often do cyclic preferences occur
agenda control and always having the status quo as a runoff option
the non-uniqueness of 'the will of the group' in most situations
the irrationality of democracy (with that example where group agrees that A implies B and B implies C but not that A implies C)
representatives vs delegates (recallable) should they vote what they think their constituents want; or what they personally think is in their constitutents' interest; or what they personally think is in the organization's interest? do they represent those who voted for them, or all their constituents (what was that quote (in the Bush administration?) about ppl who 'didnt vote for us'?)?
sortition : "It is accepted as democratic when public offices are allocated by lot; and as oligarchic when they are filled by election. —Aristotle (Politics 1301a28-35)
indirect sortition (the sortition is used to select voters who vote for representatives) convened-sample sufferage my single-sample write-in indirect sortition
cryptographic vote-counting methods:
See also:
---
what's wrong with cumulative voting for multiwinner voting?
In cumulative voting, you have a certain number of points to allocate between candidates. The n candidates with the most votes win.
Some problems with this are:
---
