City of Hudson's weighed voting system under scrutiny

[Torie]

I didn't see any discussion about about the critical point - the assumption for calculating voting power of random voting vis a vis each other by aldermen from the same ward.

Even if the court accepts the random voting assumption, there still is the issue of whether there is any public policy reason at all for weighted voting. The only one out there is that you don't have to move the lines. That's weak. For counties, there is a real rational - that the boundaries of the districts conform to the town and city subdivisions. That is a particularly compelling argument when as is the case the supervisor for towns also sits on the town council, wearing two hats.

[jimrtex]

I didn't see any discussion about about the critical point - the assumption for calculating voting power of random voting vis a vis each other by aldermen from the same ward.

Even if the court accepts the random voting assumption, there still is the issue of whether there is any public policy reason at all for weighted voting. The only one out there is that you don't have to move the lines. That's weak. For counties, there is a real rational - that the boundaries of the districts conform to the town and city subdivisions. That is a particularly compelling argument when as is the case the supervisor for towns also sits on the town council, wearing two hats.

It is not critical. And your claim of injury is that it violates your public policy sensibilities.

The voting weights can be set so that the deviation range between vote share and population share is less than 10%. In fact, the vote weights can be set so that there is no deviation at all.

At this point, you no longer have a case in federal court.

You may have a case against the county board of supervisors.

The basic concept behind weighted voting is that a representative casts a vote for each of his constituents. Kinderhook has 8501 persons and under a simple weighting system, then the town's supervisor casts 8501 votes, or at least a proportionate share of the votes. The 8501 persons represented 13.53% of a Columbia County's 62,829 residents, so the supervisor might have 1353 of 10,000 total votes, or perhaps 135 of 1000 total votes, or 478 of 3535 total votes.

This system is analogous to corporate governance where a shareholder may cast one vote for each share they own. In a county government, each resident may be thought of owning an equal share of the government. Since it is impractical for all persons to attend meetings of the board of supervisors, each supervisor in effect holds a proxy for each of his town's residents.

If supervisors who represent a majority of the county's residents vote for a motion then the motion carries. Simple weighting preserves the fundamental concept of majority rule.

Equal population districts perform somewhat similarly. If the board of supervisors had 23 members, we could divide the county into 23 districts of equal population (2732 residents), and give each supervisor 2732 votes, or each supervisor 0.043 votes (1/23 = 4.3%), or 1 of 23 total votes. Regardless of the weights used, if a majority (12) of the 23 supervisors votes for a motion, then the motion passes, but more importantly that majority of the board also represents a majority of the county's residents.

The weakness of equal-population districts is that the districts are arbitrary and capricious and subject to gerrymandering.

If the districts are malapportioned, then there is a risk that those who represent a majority of the population may not prevail on a vote, or those that represent a minority would win. This was the reasoning behind Reynolds v Sims - a faction representing a minority could prevail defeating an opposition faction representing a majority.

If each supervisor in Columbia County had one vote, then the supervisors from the 12 smallest towns and Hudson wards, who collectively represent only 28.1% of the population could carry a vote.

If the districts were not precisely equal, but roughly equal, then a faction that represents a majority will usually but not necessarily prevail. For example if 11 supervisors each represented 2868 (105% of the ideal of 2732), they would collectively represent 50.2% of the county's residents, yet would lose on a 12-11 vote.

But if the districts were drawn using some neutral criteria such as following major roads, town boundaries, and physical features, then it is quite unlikely that this situation would occur, and even more unlikely that the 11 supervisors from overpopulated districts would form a coalition.

Now back to weighted voting. If the weights are adjusted away from simple weights based on population, then some combinations of supervisors who represent a majority of the population will be unable to pass a motion.

Remember that the reason Dr. Papayanopoulos modifies the weights, is to change the distribution of critical votes. A critical vote is one in which a combination of supervisors have a majority of the total vote, but removal of a supervisor from the combination will reduce the remaining combination to a minority.

Consider all the ways critical votes can be changed:

Make a critical vote non-critical:

(1) K (for Kinderhook) is a member of a combination, and therefore the combination represents a majority of the population. Voting weights can be reduced such that the combination has a minority of the vote, yet represents a majority of the population.

{A,B,C ... K} have a majority of the population.
{A,B,C ... K} have a minority of votes.

(2) K is a critical member of a combination, and therefore the combination with K removed does not represent a majority of the population. Voting weights can be increased such that the remaining combination has a majority of the vote, even though they represent a minority of the population. Since K is no longer necessary for a majority, he does not have a critical vote.

{A,B,C ... K} have a majority of the population, but {A,B,C ... K} - {K} does not.
{A,B,C ... K} have a majority of votes, but {A,B,C ... K} - {K} also does.

Make a non-critical vote critical.

(3) T (for Taghkanic) is a member of a combination that does not represent a majority of the population, and therefore T is not a critical vote. Voting weights can be increased such that the combination has a majority of the vote, even though they represent a minority of the population. T is likely to be a critical vote, since the vote majority is likely to be slim.

{A,B,C ... T} does not have a majority of the population.
{A,B,C ... T} has a majority of votes, and if {A,B,C ... T} - {T} does not have a majority of the votes, T is critical to the combination.

(4) T is a member of a combination that represents a majority of the population, but even if T left the combination, would still have a majority. T is not a critical vote. Voting weights can be reduced such that the combination without T has a minority of the votes, making T a critical vote.

{A,B,C ... T} has a majority of the population, but so does {A,B,C ... T} - {T}, and therefore T is not critical.
{A,B,C ... T} has a majority of votes, but {A,B,C ... T} - {T} does not, and therefore T is critical to the combination.

In all four cases, a combination that represents the majority of the population does not have a majority of the votes; or a combination that represents the minority of the population, does have a majority of the votes.

It can be said that the purpose of modifying voting weights is to convert winners to losers, or losers to winners. It doesn't matter whether there was affirmative intent, or merely gross negligence. The Alabama constitution called for the legislature to be reapportioned every decade. It had never been done from 1900 through 1960 when Reynolds v Sims was rendered. Gross negligence was not a defense.

The diminishment of the right to vote under adjusted voting weight system is directed at an identifiable class of voters, those who live in more populous towns such as Kinderhook, Claverack, and Ghent; just as the apportionment in Alabama was most discriminatory against voters in larger counties such as Jefferson, Mobile, Montgomery, and Madison.

The system in Columbia is made more invidious by dividing Hudson's representation among 5 wards, four of which are smaller than Taghkanic the smallest town, by which voters in Hudson escape the injury incurred by the voters in the larger towns. And Columbia County is stuck with paying the salaries of the Hudson supervisors.

[Torie]

I understand what you wrote jimrtex, but I am not persuaded by your assertion that the non random voting assumption is irrelevant. If there were but one alderperson per ward in Hudson, Dr. Papa's mathematics would fall apart, and there would be no way for him to fix it. He could not effect the objection function of getting both population and critical voting within the 10% parameter.  

[jimrtex]

I understand what you wrote jimrtex, but I am not persuaded by your assertion that the non random voting assumption is irrelevant. If there were but one alderperson per ward in Hudson, Dr. Papa's mathematics would fall about, and there would be no way for him to fix it. He could not effect the objection function of getting both population and critical voting within the 10% parameter. 

But he could get the population deviation, which is what the SCOTUS cares about, to within 10%. I've already shown that you can get to 0%.

In Roxbury Taxpayers Association the federal court never even considered the reason for the adjusted weighting in Delaware County. The SCOTUS has rejected the use of the Banzhaf Power Index to prove an equal protection violation (Whitcomb v Chavis) or to disprove an equal protection violation (Board of Estimate v Morris).

How far is your claim that Hudson may properly used the Banzhaf Power Index to adjust voting weights, but that Dr. Papayanopoulos is calculating it wrong, going to get?

Now consider this analogy. There is a certain bus route with two buses, a half-hour apart. It turns out it is random whether each regular passenger takes the early of late bus. One passenger is superstitious and flips a coin to decide which bus to take. A waitress play's liar poker at the cafe to determine who has to do table prep for next day, and thus takes the late bus, and so on.

One of the potential passengers is named Mountain. He is 6'5 well over 300 pounds of solid muscle. No one knows if Mountain is first name, last name, or a nickname, but it fits. He is probably a longshoreman. Tim is your prototypical 98-pound weakling. His twin sister Tina is a 99-pound weakling. Though twins, they make independent determination of which bus they ride. Some days they both are on the early bus, other days on the late bus, and sometimes they ride different buses. There are a few Joe and Joanna Averages, which are not atypical.

The incident in question took place on the late bus, so there would not be another bus. The bus came upon a terrible accident, where the victim had been pinned under a car. It was critical that he be extracted from under the car, and quickly.

Mountain jumps out and begins to lift on the car, but can't quite lift it. A couple of Joe Average folks join Mountain. They lift the car an inch, but can't go further. Maybe they will relieve the pressure enough for the victim to survive. Maybe ...

But it turns out that Tiny Tim (or perhaps it was Tiny Tina) were also on the bus. With their help, the car is lifted, and while the others hold the car up, Tim (or Tina) drag the victim clear.

Who was critical to the survival of the victim? Mountain assuredly. Without his initial effort, the others wouldn't have even tried. Joe Average as well, he did what any average person would do. But Tim (or Tina) were also critical. It was dark, and they look a lot alike. We will have to check the photo from the hospital when the survivor of the accident thanks them for their critical effort. It turns out that he is quite wealthy, and wants to give each of the samaritans a reward for saving his life.

But what if both Tim and Tina had been on the late bus? They would likely be sitting next to each other. As soon as they saw the accident, they would simultaneously exclaim, "Let's Go!" and jump up. Clearly they would both be critical to survival of the victim. But that is if we use logic.

A lawyer, we will call him Banzhaf, might argue that using a power index he had devised showed that neither Tim or Tina were critical. If Tina weren't there, Tim would have been, and the car would have been lifted. If Tina were there, then Tim would be superfluous.

So if I understand it, you are going to argue that the Banzhaf Power Index, the use of which has been rejected by the SCOTUS, and the calculation of which might not actually measure what it purports to, is being calculated incorrectly?

[Torie]

That is what I am going to do, if it reaches the Constitutional stage. And Benzhaf is not dead. It still tries to address a real issue of fairness. The assumption about two persons elected from the same place at the same time voting randomly has never been litigated before, but the assumption is fundamentally flawed, and there is no real offsetting public policy advantage to weighed voting within a city without subdivisions as there is with counties. Moreover, the current lines are a form of gerrymandering because one faction in the 5th ward consistently outvotes the other, and the two election districts in the ward vote quite differently from one another. It is not homogeneous. If spit in two, the two pieces would elect different kinds of candidates.

[jimrtex]

That is what I am going to do, if it reaches the Constitutional stage. And Benzhaf is not dead. It still tries to address a real issue of fairness. The assumption about two persons elected from the same place at the same time voting randomly has never been litigated before, but the assumption is fundamentally flawed, and there is no real offsetting public policy advantage to weighed voting within a city without subdivisions as there is with counties. Moreover, the current lines are a form of gerrymandering because one faction in the 5th ward consistently outvotes the other, and the two election districts in the ward vote quite differently from one another. It is not homogeneous. If spit in two, the two pieces would elect different kinds of candidates.

The SCOTUS has agreed that the Banzhaf Power Index is based on hypothetical voting patterns that probably are not realistic at all. Re-read Chavis v Whitcomb, particularly including Justice Harlan's opinion.

Explain to me why a resident of Ward 2 has standing.

Are you claiming an equal protection violation?

BTW, on the motion to sue the BOE, weren't several wards split?

[Torie]

That is what I am going to do, if it reaches the Constitutional stage. And Benzhaf is not dead. It still tries to address a real issue of fairness. The assumption about two persons elected from the same place at the same time voting randomly has never been litigated before, but the assumption is fundamentally flawed, and there is no real offsetting public policy advantage to weighed voting within a city without subdivisions as there is with counties. Moreover, the current lines are a form of gerrymandering because one faction in the 5th ward consistently outvotes the other, and the two election districts in the ward vote quite differently from one another. It is not homogeneous. If spit in two, the two pieces would elect different kinds of candidates.

The SCOTUS has agreed that the Banzhaf Power Index is based on hypothetical voting patterns that probably are not realistic at all. Re-read Chavis v Whitcomb, particularly including Justice Harlan's opinion.

Yes, it said that in the context of being a substitute for matching the weighted voting to populations (it certainly should not), and the issue as to whether it should serve as hurdle two was never addressed.

Explain to me why a resident of Ward 2 has standing.

I would be representing a group, but first Ward 2 is undercounted, and second, all the wards but Ward 5 are screwed in voting power absent the assumption of random voting.

Are you claiming an equal protection violation?

Yes, one man, one vote, but I can win as I said based on existing state and local statutes. But I would argue the Constitutional argument, so that it would preclude the remedy being fixing the local statute, and fixing the population counts, and illegal voting.

BTW, on the motion to sue the BOE, weren't several wards split?

[jimrtex]

The SCOTUS has agreed that the Banzhaf Power Index is based on hypothetical voting patterns that probably are not realistic at all. Re-read Chavis v Whitcomb, particularly including Justice Harlan's opinion.

Yes, it said that in the context of being a substitute for matching the weighted voting to populations (it certainly should not), and the issue as to whether it should serve as hurdle two was never addressed.

Explain to me why a resident of Ward 2 has standing.

I would be representing a group, but first Ward 2 is undercounted, and second, all the wards but Ward 5 are screwed in voting power absent the assumption of random voting.

Are you claiming an equal protection violation?

Yes, one man, one vote, but I can win as I said based on existing state and local statutes. But I would argue the Constitutional argument, so that it would preclude the remedy being fixing the local statute, and fixing the population counts, and illegal voting.

So apparently you are going to sue:
(1) City of Hudson;
(2) Columbia County;
(3) Columbia County Board of Elections; and
(4) Dr. Papayanopoulos.

These are going to be your claims:

(a) Board of Elections does not maintain accurate maps.

(b) Board of Elections lets residents of Crosswinds and two houses on Harry Howard vote in Ward 4; and residents of the Columbia triangle vote in Ward 3.

(c) Board of Elections let residents of Firemen's Home vote in Ward 4. Alternatively, the city failed to convey to the Board of Elections where the ward boundary was.

(d) Hudson and Columbia County miscalculated ward populations which they used for computing voting weights, and they failed to document their populations estimates as required by state law.
(i) They misapportioned (grossly malapportioned) the Front Street block.
(ii) They included the Harry Howard notch in Ward 5's population.
(iii) They included Crosswind and the two Harry Howard addresses in Ward 5, while condoning voters voting in Ward 4.
(iv) They included the Columbia triangle in Ward 3's population.

(e) Columbia County failed to provide for equal population election areas for supervisors from Hudson, violating state law.

(f) Columbia County violates the two-cap state law.

(g) Columbia County violates equal protection by adjust voting weights away from simple population-based weights. This ensures that some combinations of supervisors who represent a majority of the population will not carry motions, and some minority combinations will succeed. The injury is directed at an identifiable class of voters, those residing in larger towns such as Kinderhook, Claverack, and Ghent.

Arguably this injury is intentional, given that the purpose of adjusting the voting weights is to change the distribution of critical votes which can only be done by permitting the majority of the population to lose, and the minority of the population to succeed.

(h) Columbia County permits Hudson to escape this injury by dividing its representation. Since they pay the salaries of the Hudson supervisors, they incur an unnecessary and irrational expense.

(i) Hudson miscalculated the voting weights, by considering the two aldermen from each ward as being independent actors.

(j) If the voting weights were calculated correctly, it would be too easy for the aldermen to dominate voting (semi-dictator).

(k) Voters in parts of Ward 5 are denied equal protection through the device of a multi-member election. This is essentially, the same claim that minorities make. Note to have standing you have to a victim who resides in Ward 5. A claim that the residents of parts of Ward 5 would elect someone more to your liking is too tenuous. This may also be a violation of state law by limiting partisan expression.

I think you have a clear win on a, b, d, g, and h.

[Torie]

The SCOTUS has agreed that the Banzhaf Power Index is based on hypothetical voting patterns that probably are not realistic at all. Re-read Chavis v Whitcomb, particularly including Justice Harlan's opinion.

Yes, it said that in the context of being a substitute for matching the weighted voting to populations (it certainly should not), and the issue as to whether it should serve as hurdle two was never addressed.

Explain to me why a resident of Ward 2 has standing.

I would be representing a group, but first Ward 2 is undercounted, and second, all the wards but Ward 5 are screwed in voting power absent the assumption of random voting.

Are you claiming an equal protection violation?

Yes, one man, one vote, but I can win as I said based on existing state and local statutes. But I would argue the Constitutional argument, so that it would preclude the remedy being fixing the local statute, and fixing the population counts, and illegal voting.

So apparently you are going to sue:
(1) City of Hudson;
(2) Columbia County;
(3) Columbia County Board of Elections; and
(4) Dr. Papayanopoulos. Why on earth would he be sued?

These are going to be your claims:

(a) Board of Elections does not maintain accurate maps.

(b) Board of Elections lets residents of Crosswinds and two houses on Harry Howard vote in Ward 4; and residents of the Columbia triangle vote in Ward 3.

(c) Board of Elections let residents of Firemen's Home vote in Ward 4. Alternatively, the city failed to convey to the Board of Elections where the ward boundary was.

(d) Hudson and Columbia County miscalculated ward populations which they used for computing voting weights, and they failed to document their populations estimates as required by state law.
(i) They misapportioned (grossly malapportioned) the Front Street block.
(ii) They included the Harry Howard notch in Ward 5's population.
(iii) They included Crosswind and the two Harry Howard addresses in Ward 5, while condoning voters voting in Ward 4.
(iv) They included the Columbia triangle in Ward 3's population.

(e) Columbia County failed to provide for equal population election areas for supervisors from Hudson, violating state law.

(f) Columbia County violates the two-cap state law. I don't understand what this means.

(g) Columbia County violates equal protection by adjust voting weights away from simple population-based weights. This ensures that some combinations of supervisors who represent a majority of the population will not carry motions, and some minority combinations will succeed. The injury is directed at an identifiable class of voters, those residing in larger towns such as Kinderhook, Claverack, and Ghent. Don't understand this either. Are you saying that in order to avoid gross disproportions in voting power, sometimes representatives representing a majority of the population need to lose a particular vote? If, so that is kind of an interesting twist of the conundrum.

Arguably this injury is intentional, given that the purpose of adjusting the voting weights is to change the distribution of critical votes which can only be done by permitting the majority of the population to lose, and the minority of the population to succeed.

(h) Columbia County permits Hudson to escape this injury by dividing its representation. Since they pay the salaries of the Hudson supervisors, they incur an unnecessary and irrational expense. How does this effect an escape?

(i) Hudson miscalculated the voting weights, by considering the two aldermen from each ward as being independent actors.

(j) If the voting weights were calculated correctly, it would be too easy for the aldermen to dominate voting (semi-dictator). Is this another way of stating the rationale for the voting power concept?

(k) Voters in parts of Ward 5 are denied equal protection through the device of a multi-member election. This is essentially, the same claim that minorities make. Note to have standing you have to a victim who resides in Ward 5. A claim that the residents of parts of Ward 5 would elect someone more to your liking is too tenuous. This may also be a violation of state law by limiting partisan expression. Is this another iteration of the voting power concept, and the assumption of random voting, in lieu of just one representative, which would then give the 40% vote of Ward 5 the power to sustain mayoral vetoes on its own? Anyway, the courts in NY have embraced the idea of voting power diverging from population weights as a finesse of the unfair domination issue. The clearest example is where one jurisdiction has a majority of the population. The state fix was by law to require two supervisors in such a town, playing the random voting game. I am not sure the federal courts have agreed that such a fix flies. I will have to reread the federal cases on Hempstead.

I think you have a clear win on a, b, d, g, and h.

My comments are in red above.

[jimrtex]

So apparently you are going to sue:
(1) City of Hudson;
(2) Columbia County;
(3) Columbia County Board of Elections; and
(4) Dr. Papayanopoulos. Why on earth would he be sued?

He is responsible for devising the method of adjusting voting weights. He knew or should have known of the problems.

These are going to be your claims:

(a) Board of Elections does not maintain accurate maps.

(b) Board of Elections lets residents of Crosswinds and two houses on Harry Howard vote in Ward 4; and residents of the Columbia triangle vote in Ward 3.

(c) Board of Elections let residents of Firemen's Home vote in Ward 4. Alternatively, the city failed to convey to the Board of Elections where the ward boundary was.

(d) Hudson and Columbia County miscalculated ward populations which they used for computing voting weights, and they failed to document their populations estimates as required by state law.
(i) They misapportioned (grossly malapportioned) the Front Street block.
(ii) They included the Harry Howard notch in Ward 5's population.
(iii) They included Crosswind and the two Harry Howard addresses in Ward 5, while condoning voters voting in Ward 4.
(iv) They included the Columbia triangle in Ward 3's population.

(e) Columbia County failed to provide for equal population election areas for supervisors from Hudson, violating state law.

(f) Columbia County violates the two-cap state law. I don't understand what this means.

MHR law says: "(b.) A plan of apportionment adopted by a county under this subparagraph may provide that mayors of cities or villages, supervisors of towns or members of the legislative bodies of cities, towns, or villages, who reside in the county shall be eligible to be elected as members of the county legislative body"

In one of the state lawsuits that proceeded Abate v Mundt, a state court interpreted this to mean that supervisors could not be members of a county legislative body by virtue of their election as town supervisors - since the State had said that a county could permit a person to run and be elected to both offices. The SCOTUS noted:

"Rockland County responded with a plan that substantially remedies the malapportionment and that, by preserving an exact correspondence between each town and one of the county legislative districts, continues to encourage town supervisors to serve on the county board."

I had misread this to mean town supervisors would be members of the county board by virtue of their being town supervisors, but that would not necessarily make sense in Rockland County, where each town, other than the smallest, had multiple legislators.

A federal court a couple of decades later overturned the Rockland County apportionment scheme. Most importantly this was because the deviation had increased significantly, but the judge also noted that Rockland County had made it illegal for someone to be a town supervisor and county legislator.

It is possible that Columbia County escapes by having a traditional board of supervisors. But if that is true, they may also escape the application of equal-size county legislator districts within a city.

(g) Columbia County violates equal protection by adjust voting weights away from simple population-based weights. This ensures that some combinations of supervisors who represent a majority of the population will not carry motions, and some minority combinations will succeed. The injury is directed at an identifiable class of voters, those residing in larger towns such as Kinderhook, Claverack, and Ghent. Don't understand this either. Are you saying that in order to avoid gross disproportions in voting power, sometimes representatives representing a majority of the population need to lose a particular vote? If, so that is kind of an interesting twist of the conundrum.

Arguably this injury is intentional, given that the purpose of adjusting the voting weights is to change the distribution of critical votes which can only be done by permitting the majority of the population to lose, and the minority of the population to succeed.

Yes. But it is not as benign as you state it. It is not sometimes. It is all the time. If you want to change the voting power, you must change the voting weights such that the result (success or failure) is changed.

Think what a critical vote is. First we will use simple, population-based weighting:

P(  ) is the population represented by a particular set of members.  P(ABC) is the population represented by A,B, and C. P(ABC...K) is the population represented by the members of a coalition that includes K.  And P(ABC...) is the population with K removed, or where switches his vote to No.

Condition I.

P(ABC...K)/P(all) > 1/2, but P(ABC...)/P(all) <= 1/2

K (for Kinderhook) is critical, because if he leaves the coalition, the motion fails.

Condition II

P(ABC...T)/P(all) > 1/2, but P(ABC...)/P(all) > 1/2

T (for Taghkanic) is not critical, because if he leaves the coalition, it still succeeds.

Condition III

P(ABC...T)/P(all) <= 1/2

The coalition is not successful, so T is not critical.

The Banzhaf Power Index for a given set of vote weights, is defined as the proportion of critical votes for each member of the representative body. The distribution of voting power can be changed, by changing the voting weights. That is the motivation for changing the voting weights. The way to change the voting power is to convert some critical votes to non-critical and some non-critical votes to critical.

We will switch to an alternative vote weighting W.

Condition I.

P(ABC...K)/P(all) > 1/2, but P(ABC...)/P(all) <= 1/2

W(ABC...K)/W(all) > 1/2, but W(ABC...)/W(all) > 1/2

We have augmented the votes of some of the members, such that now they represent a majority of the vote. K for Kinderhook is no longer critical if (ABC...K) since (ABC...) is also successful. They didn't need Kinderhook to have a majority of the vote. But P(ABC...) was not a majority, since that was necessary for Kinderhook's vote to be critical.

In addition, (ABC...) is likely to be critical to all its members since it now represents a majority of the vote, and is likely to be just a bit over 1/2.

Condition II

P(ABC...T)/P(all) > 1/2, but P(ABC...)/P(all) > 1/2

W(ABC...T)/W(all) > 1/2, but W(ABC...)/W(all) <= 1/2

Here were have reduced the voting weights of some of the other members. Before, Taghkanic was not critical, because the remainder of the coalition was successful without him. Now they need him, and he is critical.

If you were to examine all 8 million combinations of the 23 members of Board of Supervisors, and extracted those where a vote was critical using the Papayanopoulos weights, but was not critical using simple, population weights, or vice versa (not-critical with population based votes, but critical with Papayanopoulos weight), you will find that in every instance a population majority was converted to a weighted vote minority, or vice versa.

This is not just an inadvertent side effect. It is the reason the voting weights are modified - to convert majorities to minorities and minorities to majorities.

Dr. Papayanopoulos was involved in a lawsuit in Chenango County, where a similar effect was noted with regard to a 2/3 vote, where using the 1/2-decision weights, the supervisors from Norwich could not block a 2/3 vote even though Norwich was more than 1/3 of the population. He then devised a set of 2/3-weights which permitted the supervisors from Norwich to block a 2/3 vote. But what was failed to be recognized was that this necessarily resulted in another faction that represented 1/3 of the population from blocking a 2/3.

Any change in voting weights that causes a change in the results (success or failure of a motion) will necessarily have this effect. And it would be pointless to change the voting weights except to change the results.


(h) Columbia County permits Hudson to escape this injury by dividing its representation. Since they pay the salaries of the Hudson supervisors, they incur an unnecessary and irrational expense. How does this effect an escape?

Residents of Kinderhook, Claverack, and Ghent are injured because the voting weight share of their supervisors is less than their population share. The same would be true if Hudson had only one supervisor. By splitting among 5 supervisors, they don't suffer this harm.


(i) Hudson miscalculated the voting weights, by considering the two aldermen from each ward as being independent actors.

(j) If the voting weights were calculated correctly, it would be too easy for the aldermen to dominate voting (semi-dictator). Is this another way of stating the rationale for the voting power concept?

(k) Voters in parts of Ward 5 are denied equal protection through the device of a multi-member election. This is essentially, the same claim that minorities make. Note to have standing you have to a victim who resides in Ward 5. A claim that the residents of parts of Ward 5 would elect someone more to your liking is too tenuous. This may also be a violation of state law by limiting partisan expression. Is this another iteration of the voting power concept, and the assumption of random voting, in lieu of just one representative, which would then give the 40% vote of Ward 5 the power to sustain mayoral vetoes on its own? Anyway, the courts in NY have embraced the idea of voting power diverging from population weights as a finesse of the unfair domination issue. The clearest example is where one jurisdiction has a majority of the population. The state fix was by law to require two supervisors in such a town, playing the random voting game. I am not sure the federal courts have agreed that such a fix flies. I will have to reread the federal cases on Hempstead.

The MHR law says:

"iii.) The plan shall provide substantially fair and effective representation for the people of the local government as organized in political parties"

Isn't this the essence of your claim?

Delaware County no longer uses power-adjusted weighting. Saratoga County no longer uses power-adjusted weighting.

Maybe the New York courts were wrong. Maybe the Banzhaf Power Index is wrong.


I think you have a clear win on a, b, d, g, and h.

My comments are in red above.

Mine are in blue.

[Torie]

"Residents of Kinderhook, Claverack, and Ghent are injured because the voting weight share of their supervisors is less than their population share. The same would be true if Hudson had only one supervisor. By splitting among 5 supervisors, they don't suffer this harm."


Assuming Hudson is underrepresented in vote weight vis a vis its population, how does having 5 supervisors change that result?

[jimrtex]

"Residents of Kinderhook, Claverack, and Ghent are injured because the voting weight share of their supervisors is less than their population share. The same would be true if Hudson had only one supervisor. By splitting among 5 supervisors, they don't suffer this harm."

Assuming Hudson is underrepresented in vote weight vis a vis its population, how does having 5 supervisors change that result?

Hudson is not underrepresented in vote weight vis a vis its population.

Hudson wards have 1.15%, 1.23%, 1.82%, 2.04%, and 3.96% of the county population, and 1.22%, 1.27%, 1.90%, 2.07%, and 4.02% of the total voting weight.

Kinderhook has 13.53% of the population, and 12.50% of the voting weight.

Kinderhook, Claverack, and Ghent are given reduced voting weight, so that when they could have been in the majority they aren't. Greenport and Chatham are about even, and everyone else is advantage.

Have you read Banzhaf's law review article on why supposedly, weighted voting does not work?

[Torie]

"Residents of Kinderhook, Claverack, and Ghent are injured because the voting weight share of their supervisors is less than their population share. The same would be true if Hudson had only one supervisor. By splitting among 5 supervisors, they don't suffer this harm."

Assuming Hudson is underrepresented in vote weight vis a vis its population, how does having 5 supervisors change that result?

Hudson is not underrepresented in vote weight vis a vis its population.

Hudson wards have 1.15%, 1.23%, 1.82%, 2.04%, and 3.96% of the county population, and 1.22%, 1.27%, 1.90%, 2.07%, and 4.02% of the total voting weight.

Kinderhook has 13.53% of the population, and 12.50% of the voting weight.

Kinderhook, Claverack, and Ghent are given reduced voting weight, so that when they could have been in the majority they aren't. Greenport and Chatham are about even, and everyone else is advantage.

Have you read Banzhaf's law review article on why supposedly, weighted voting does not work?

I read where he said it had been misused, but no, that he had decided it was a fail, or had a serious flaw to ponder.

[jimrtex]

Have you read Banzhaf's law review article on why supposedly, weighted voting does not work?

I read where he said it had been misused, but no, that he had decided it was a fail, or had a serious flaw to ponder.
[/quote]
See "Weighted Voting Doesn't Work: A Mathematical Analysis" 19 Rutgers L. Rev. 317 1964-1965

This was written soon after the first SCOTUS OMOV decisions. At the time Banzhaf was a law student at Columbia.

In the summary he states:

"This article demonstrates that an assumption necessary to the validity of systems of weighted voting is incorrect. This is done by showing that there is no necessary correlation between the number of votes and voting power, by developing a method of measuring voting power in weighted voting situations, and by applying this measure to various examples of weighted voting systems."

But what proof is there that the Banzhaf Power Index measures "voting power"? What he demonstrated was that the Banzhaf Power Index did not correlate with the number of votes. Note that here he uses number of votes as equivalent to population.

In effect he set out to prove that there was not correlation between voting weight and voting power, and then devised a metric that did not correlate with voting weight, and therefore the metric must measure voting power.

I don't know that there is any connection between Banzhaf and Dr.Papayanopoulos. When some counties started implementing weighted voting for their boards of supervisors, state courts started striking them down, and cited Banzhaf's article. They pointed out that it was practically impossible to calculate the "voting power" without a supercomputer (there are 8 billlion voting combinations in Columbia County).

Dr. Papayanopolous at the time worked for IBM in upstate New York, with apparently the job of developing applications for an IBM 360. Somehow he hooked up with a court and calculated the voting power using the population-based weights proposed by the county. He then had the insight that not only could you calculate the "voting power" for a given set of weights, you could choose a set of weights that produce correlation between "voting power" and population.

He developed a set of voting weights for the county, which the court ordered the county to adopt.

[jimrtex]

An alternate scheme:

3 aldermen elected from each of:

(A) Wards 1 and 3
(B) Wards 2 and 4
(C) Ward 5

Alderman would be elected by simplified cumulative vote. Under a cumulative vote system, a voter may cast three votes, and distribute them among the candidates. But it is simpler to let a voter vote for one, two, or three candidates. When calculating the results, a voter is assumed to have 6 votes:

(i) Voted for 3 candidates, counts as two votes each.
(ii) Voted for 2 candidates, counts as three votes each.
(iii) Voted for 1 candidate, counts as six votes for that candidate.

This system limits voter mistakes such as using too many votes, or too few votes. The only possible mistake would be to vote for four or more candidates. For that reason the "six votes" is internal to avoid use of fractional votes.

Three candidates will be elected, and their voting weight will be proportional to the population of their area, and their share of the vote. Votes will be redistributed to ensure that all votes are transferred to three elected candidates, and that no candidate will have more than 40% of the vote.

Procedure:

(1) Count ballots, allocating each voter's six votes based on whether they voted for 1, 2, or 3 candidates.

(2) If any candidates received 40% of the total vote or more, they are elected, and have a surplus. There will be at most two such candidates.

(3) Distribute any surpluses. The largest surplus will be distributed first. The elected candidate may transfer his surplus to any remaining candidate who has less than 40% of the vote. If the transfer causes another candidate to reach 40%, then any unused surplus may be transferred to another candidate.

(4) Eliminate candidates until three remain.

(5) Candidates will be eliminated in rounds. The candidate with the fewest votes will be eliminated. He may choose another of the remaining candidates. If his transfer causes another candidate to reach 40%, then he may transfer the remainder of his vote to another candidate.

(6) If a candidate fails to transfer his vote, then it will be distributed equally among all remaining candidates.

(7) The weighted vote for each elected alderman is calculated:

votes = population * 0.9 * share_of_votes.

Example:

A candidate from Ward 5, receives 25% of the vote.

votes = 2509 * 0.9 * 0.25
votes = 564

( Council President elected at large has city_population * 0.1 = 6403 * 0.1 = 641 votes.

The 3 aldermen from each area will be more pluralistic, removing your concern about them voting the same. Their voting weight will depend on their personal support. Voters have an incentive to vote, since each vote will enhance the voting weight of their preferred candidates. Candidates will have an incentive to push GOTV. While wards 1 and 4 will lose their own aldermen, they can always plump for someone from their part of town.

Sweetener:

Adjust the ward boundaries, eliminating the ambiguities.

1st Ward:

Warren extended, Warren, South Third St to city limits.

2nd Ward:

Warren extended, North Third St, Robinson, Strawberry Alley, North 2nd Street to city limits

3rd Ward:

South Third St, Warren, Park Place, Columbia Street to city limits (Paul Avenue)

4th Ward

North 2nd Street, Strawberry Alley, Robinson Street, Warren, North 5th Street, North 5th Street extended to Harry Howard, Harry Howard to city limits.

5th Ward

Harry Howard, 5th Street Extended, North 5th Street,  Warren, Park Place, Columbia Street to city limits.

[Torie]

I rather like your scheme, and it would eliminate the random voting assumption issue, but of course it is never going to happen. Among other things, the odds are pretty good that the gentry would win a majority of the council votes. They would likely win all three alderpersons from wards 1 and 3, and one each from the other two wards. Well maybe not, since the 5th ward still has more votes.

[jimrtex]

I rather like your scheme, and it would eliminate the random voting assumption issue, but of course it is never going to happen. Among other things, the odds are pretty good that the gentry would win a majority of the council votes. They would likely win all three alderpersons from wards 1 and 3, and one each from the other two wards. Well maybe not, since the 5th ward still has more votes.

An interesting part of the plan is that it would reduce the common council to an even number, 9 aldermen and the president. With weighted voting, 5 members that represent a majority of the population could carry a motion. In an equal-population scenario, that would be a tie vote. When there are 11 members, a simple majority is actually a super-majority of 54.5%.

Another plan would be to elect one alderman from each ward, except for the fifth ward, which would elect two aldermen. Three aldermen would be elected citywide by cumulative voting. The fifth ward could be formally divided into two wards.

Currently, Hudson has a hard time getting enough candidates to contest the common council. In part this is because there is little chance to defeat an incumbent of the dominant party in a ward.

[Torie]

Having 5 equal population wards essentially divides the 5th ward into two wards. Team Scalera does not want that to happen, because at the moment one half of the ward dominates the other half, which is more evenly divided (we know that based on the election results from the two voting districts in the ward). So a division would end the gerrymander regime that needs to be protected. That is what this fuss is mostly about. If the 5th ward were more homogeneous, there would be far less of a fuss.

My guess is that the white gentry is going to lose the mayoral election next week. Team Scalera has done a good job creating a coalition with the minority community to take down the white gentry. Favors are flying all over the place. In Hudson, the more things change, the more they stay the same. And team Scalera has figured out, that if there are to be equal population wards, what is most favorable to them are 3 wards, which essentially enable the Scalera faction to control one ward, the minority community a second, leaving the gentry with the third, and absent control of the common council president, essentially toothless. Fun stuff! Oh, and I think the Scalera candidate might win the Common Council president too. The Dem candidate dropped out (my pal Victor Mandolin), and the independent candidate has basically hardly campaigned at all, and does not say much. So the Pub might actually win it - a lady I have never met or spoken too, and have never noticed at Council meetings.

More chat about weighed voting and all here, including by yours truly.

[jimrtex]

Having 5 equal population wards essentially divides the 5th ward into two wards. Team Scalera does not want that to happen, because at the moment one half of the ward dominates the other half, which is more evenly divided (we know that based on the election results from the two voting districts in the ward). So a division would end the gerrymander regime that needs to be protected. That is what this fuss is mostly about. If the 5th ward were more homogeneous, there would be far less of a fuss.

My guess is that the white gentry is going to lose the mayoral election next week. Team Scalera has done a good job creating a coalition with the minority community to take down the white gentry. Favors are flying all over the place. In Hudson, the more things change, the more they stay the same. And team Scalera has figured out, that if there are to be equal population wards, what is most favorable to them are 3 wards, which essentially enable the Scalera faction to control one ward, the minority community a second, leaving the gentry with the third, and absent control of the common council president, essentially toothless. Fun stuff! Oh, and I think the Scalera candidate might win the Common Council president too. The Dem candidate dropped out (my pal Victor Mandolin), and the independent candidate has basically hardly campaigned at all, and does not say much. So the Pub might actually win it - a lady I have never met or spoken too, and have never noticed at Council meetings.

More chat about weighed voting and all here, including by yours truly.

This is getting really fun

Had the correct ward populations been used, and the weighted votes been based on the correct figures, the resolution would have failed.  Or maybe not

As you know:

Ward 1 benefited at the expense of Ward 2 because of the misallocation of Hudson Terrace Apartments.

Ward 5 benefited at the expense of Ward 4, because of the misallocation of the Harry Howard notch.

Ward 3 benefited at the expense of Ward 5, because of the misallocation of the Columbia Triangle.

Winners: Wards 1 and 3.
Losers: Wards 2 and 4
Both: Ward 5

Now look at the results of the vote on the resolution:

Aye: Wards 1 and 3, and President.
No: Wards 2 and 4
Split: Ward 5

The current voting weights are based on the erroneous calculation that the population of
wards 1+3 is 1912; and
wards 2+4 is 2006 (4.9% larger)

This is close enough that with Ward 5 split, the president decides the issue.

But the correct populations are:
wards 1+3 is 1669
wards 2+4 is 2268 (35.9% larger).

Now whether the motion fails or not hinges on how much voting weight the president has. In an equal population scenario, each alderman would represent 1/10 of the population, or 640 persons, and that would be the value of the presidents vote (it doesn't matter whether each member has one vote or 640 as far as outcome). The difference between Wards 1+3 and 2+4 is 599. So if the aldermen had simple voting weights based on population, and the president had 640 votes, the resolution would narrowly have narrowly passed.

But let's look at some alternatives:

Hudson Voting Weights

The first was your voting weights, which were intended to demonstrate that the assumption about the alderman voted independently could produce reasonable weights, but would fail badly if that assumption was wrong.

The resolution would have passed using your weights. You helped out a bit by giving wards 1 and 3 a bit of a bonus, and penalizing Ward 2 (voting weight relative to population).

The second was my model where I used simple population weights, but varied the president's weight with the goal of optimizing the relationship between population and voting power. It happens that this would given the President 6.0% of the total vote, rather than the 9.1% he would have had under a equal-population scenario. In effect he has a light thumb which can not overcome the larger population of the northside. This could be considered an illustration of how changing voting weights can change results. With one caveat - how much population does the president represent?

The third was where I let all voting weights float so as to produce an optimum relationship  between voting power and population. The resolution would have passed using these weights, largely because the president would have a relatively strong weight.

If you could give me what you believe to be the correct ward populations, I can update these spreadsheets.

Who knows affirmatively that the population numbers are wrong? The city clerk and city attorney, I'm pretty sure do. Does the Common Council and mayor?

[Torie]

The correct ward populations, on which the City Clerk and myself agree, are posted here by Carole Osterink. She got that screenshot from me. The numbers assume that all of the Fireman's Home population goes to the 5th ward, which involves 85 people.

Everybody knows the population counts are wrong, everybody. The Mayor disingenuously expresses some confusion about the details, but is stonewalling and vetoing, because he wants the ward lines to be changed to match were folks are currently voting improperly.

[jimrtex]

The correct ward populations, on which the City Clerk and myself agree, are posted here by Carole Osterink. She got that screenshot from me. The numbers assume that all of the Fireman's Home population goes to the 5th ward, which involves 85 people.

Everybody knows the population counts are wrong, everybody. The Mayor disingenuously expresses some confusion about the details, but is stonewalling and vetoing, because he wants the ward lines to be changed to match were folks are currently voting improperly.

The numbers I used then are the same, except for the Firemen's Home.

So we almost managed a jackpot, a vote on weighted voting, where the result was changed by use of voting weights which were miscalculated due to the erroneous population figures.

[jimrtex]

SCOTUS briefs Harris, et al v. Arizona Independent Redistricting Commission, et al.

The briefs have quite a bit about the 10% deviation burden-shifting threshold, sometimes thought to be a safe harbor.

Generally, the commission and its supporters argue that getting precleared under Section 5 of the VRA was a rational state interest that justified underpopulating Democrat-leaning districts. If you read carefully, you will recognize a pitfall of citizen redistricting commissions. They rely on their expert consultants' advice. Expect to hear the words "minimal" dozens of times.

The Secretary of State, who was a named defendant in the district court case, in his/her role as administrator of the commission gerrymander, did not participate in the lower court proceedings, but has submitted a brief before the SCOTUS. Since the AG is representing the SOS, the State of Arizona is opposing its redistricting commission.

Amicus briefs of interest are:

Navajo Nation: They argue that the majority Indian district is justifiable because it is largely rural, and it permitted an increase in the share of the Indian voting population.

Sam Wang, a political scientist, who argues that the correct way to measure partisanship in redistricting is to compare the two-party share of the vote in the median district, with that of the state average. He assumes that population distribution is not skewed.

Southern Coalition for Social Justice in support of neither party. They oppose setting a bright line test for deviation, because they are involved in a North Carolina case where they are arguing that minority districts were overpopulated.

[muon2]

Sam Wang, a political scientist, who argues that the correct way to measure partisanship in redistricting is to compare the two-party share of the vote in the median district, with that of the state average. He assumes that population distribution is not skewed.

I appreciate Wang's contention that a difference between the median and mean is evidence of skew. However, I'm not sure that the statistics hold up when there are districts averaged in that only have one party on the ballot. The distribution of districts is certainly not normally distributed at the edges. There can also be significant bias introduced if one of the uncontested districts would normally be near the median, but is uncontested due to a popular incumbent. I think the better measure is to look at the results of a statewide race in each district (I use the presidential results).

His method also can fail to account for effects due to a preponderance of marginal districts for one party. Many gerrymanders will place a number of districts with enough advantage for the gerrymandering party. The median district may not reflect how many districts were gained by a gerrymander. Hence my desire to count the difference between districts that nominally should go for each party.

Finally I don't know if his method accounts for the known effect that a statewide swing has on individual districts. Each 1% swing in the vote is predicted to produce a 2% shift in the delegation. In a state like AZ there is a decided shift to the Pubs statewide, and it's not clear to me that the bias he sees isn't just due to the 2 to 1 effect. It's worth some simulation to see if that's the case.

[Torie]

The next few months will be most interesting in Hudson on this issue. The Common Council went one way in the recent election, the Mayorship the other. It does seem to be seeping through the synapses in the hood that weighted voting has to go. So now there may be maneuvering for advantage.

[jimrtex]

Sam Wang, a political scientist, who argues that the correct way to measure partisanship in redistricting is to compare the two-party share of the vote in the median district, with that of the state average. He assumes that population distribution is not skewed.

I appreciate Wang's contention that a difference between the median and mean is evidence of skew. However, I'm not sure that the statistics hold up when there are districts averaged in that only have one party on the ballot. The distribution of districts is certainly not normally distributed at the edges. There can also be significant bias introduced if one of the uncontested districts would normally be near the median, but is uncontested due to a popular incumbent. I think the better measure is to look at the results of a statewide race in each district (I use the presidential results).

With respect to election precincts the median is definitely more Republican than the mean, and the distribution is definitely skewed. But this is not an indication of gerrymandering.