ISSUE # 29b ... DON'T KNOW MUCH ABOUT ALGEBRA .... Part II
February 21
Paul Menter is also concerned about IRV. Paul's letter gives us the 30,000 foot view of the mess.
See his letter below. (Also see updatedinformation following the Orginal Posting.)
To the Editor,
I’ve been hearing the distant, gravelly voice of Edward P. Benton recently. My 8th grade social studies teacher’s mantra was …”You study the past so you can make better decisions in the present and improve the future.”
Or something like that…
There are certainly plenty of lessons from history going unobserved today. The one on my mind at the moment is elections. Back in 1787 the fifty-five arguably most intelligent men in the United States met in Philadelphia and birthed our Constitution and along with it the first version of our current national system for electing the President – the Electoral College.
It was brand new. Nothing like it had existed before. It had never been tested and there was no historical reference for use as a waypoint in crafting its structure or controlling for its unanticipated consequences. A compromise system, it gave power to the states to create whatever method they desired to vote for their electors who would then vote for the President.
It was nothing short of a disaster.
A series of unintended consequences spurred by Alexander Hamilton’s nefarious manipulations led to unexpected results in 1796 as opponents John Adams and Thomas Jefferson ended up serving together as President and Vice President, and in 1800, when the election was thrown to the House of Representatives for lack of a majority winner. It took 36 ballots over five days before Thomas Jefferson was narrowly elected days before Congress was to adjourn.
Constitutional crisis averted, just barely.
Monday night Aspen’s City Council will consider a new method of electing themselves called instant runoff voting, or “Irv” for short. Without any specifics, away back in November of 2007, Aspen’s voters granted the City carte blanche to create whatever “Irv” system they desired. Hmmm…Sound familiar?
Over a year later, a committee consisting of a few citizens and two (presumably) incumbent candidates unveiled two muddled options. With antiseptic terms to describe them, they sound more like euthanization procedures, “sequential elimination” and “batch elimination”, than election systems.
From these options I understand a third “compromise” solution has emerged; a sort of hybrid “two vote batch elimination” system.
It’s brand new; a system that has never been used anyplace else before for multi seat elections (like the Aspen City Council elections) and for which no historic reference exists for use as a “waypoint” to guide its development and avoid its unintended consequences. As you can imagine, rumors abound about the potential for Hamiltonian style manipulation. Sound familiar?
Technical in nature, the system requires that a custom, one of a kind software program be written in order to tabulate the votes. It is my understanding that Council are requiring no provisions for a secondary hand count or other method to confirm the results of this brand new election system that has never been used anywhere else in the Country and requires a custom software program. Sound familiar?
Our most sacred public process reduced to a black box? Hope I’m wrong!
What would Mr. Benton say? If we don’t heed the mistakes of the past they will haunt us. Aspen, prepare to be haunted.
Paul Menter
Comment from Marilyn 2.21.08:
In an unbelievable turn of events, the city staff is now recommending that council adopt a counting procedure Monday night that has not even been written. The public, including potential candidates, have not even had proper notice and opportunity to read the dozens of pages of the 2 other methods being discussed.
Furthering our suspicions is the fact that there is to be a special software program written for Aspen’s unique method of counting votes. There are no independent testing procedures in place for that new software program. And unbelievably, staff is recommending that the Council deny the citizens’ request for hand counting the ballots to verify the results.
See http://theredant.squarespace.com/storage/IRVMemo_2nd%20Reading_2.pdf
This train needs to be stopped.
==============================================================================
Neil Siegel writes about IRV---INSTANT CHAOS
The City Council is poised to enter the statistical morass of instant run-off voting ( IRV). There is one clear path, consistent with prior run-off practice, the recommendation of the taskforce and consent of the voting public, but given the track record of this incantation of city government to do as it sees fit, nothing is certain. The Council is messing with our votes that could dramatically effect the result of the election. It needs to get it right, there is no compromise, especially when different outcomes are possible.
On the table are two, possibility three schemes to eliminate candidates during IRV. The results vary from scheme to scheme, and critically can differ significantly from the method Aspen has historically used.
When the City Council approved Ordinance 38 calling for IRV and Resolution # 86 two reasons were given:
“ …..eliminating the need to conduct expensive run-off elections at a later date”
“…..the convenience of only having to vote once and financially responsible with taxpayer dollars”.
Never once was there any mention of any perceived defect in the method of vote tabulation used in past run-off elections.
The yeoman efforts of Kathryn Koch and Jim True deserve recognition. But for reasons yet explained implementation is about to run of the rails.
Nowhere in the United States has IRV ever been used in a multi-seat election requiring a majority. Aspen is the first, we are the white rats for IRV. There is no software in existence, to say nothing of vote certified software, to tabulate votes for an Aspen type IRV election. There is no plan to comprehensively test the IRV software and certify its accuracy. There is no contemplation than any hand recount will be possible to verify the results of the election. We are at the mercy of a “black box”.
Importantly, never was there any approval to substantively change the manner of candidate ranking and elimination, only to conduct the election at one time and avoid subsequent run-off elections. Now we are in a worse condition, the City Council incorrectly divining that it has a mandate to determine how our votes will be handled.
The City Council is asked to follow a runoff scheme advanced by the one recorded dissent of that task force, a sitting member of this council expected to seek re-election. Worse yet, the proposed methodology has a built-in bias for an incumbent.
The prudent and correct course is two-fold: [1] adopt the task force recommendation since it replicates most closely our own prior runoff process and most importantly comports with the legislative history and the public will and [2] use IRV this election as a test to gain confidence in the system.
View Printer Friendly Version
Email Article to Friend
Reader Comments (15)
The instant runoff voting proposal by Jack Johnson has been used in literally thousands of elections for major offices in countries like Australia and Ireland. It is the method of IRV that has been approved and is being used in many American cities like San Francisco, Oakland, Minneapolis, Santa Fe and Sarasota (FL). It is spelled out in Robert's Rules of Order as a good system for any elections of officers held by mail. This IRV system also easily can be counted by hand, as it indeed it is for all the elections for parliament and president in Ireland and Australia.
The task force came up with something new. It's interesting and innovative, but indeed, something entirely new. It also would be harder to count by hand.
You can see why FairVote is recommending the council go with Johnson's proposal.
Rob,
can you give us some examples of elections which use IRV in multi-seat, majority required races, like Aspen's Council race?
The people I've talked with have not been able to point to thoe examples yet.
It would be helpful to look at their systems.
Marilyn
Mr. Richie,
I understand the work FairVote does to promote IRV in single seat elections. In many ways I applaud your goals. Runoff and instant runoff elections both can make significant contributions to the democratic process. However I would urge you to consider the consequences of using an iterative SE process to fill two seats, each with a majority required. This process has the effect of giving the voter who votes for the winner of the first seat a greater degree of influence in the outcome of the election while stripping the second preference of voters who vote for the candidate who finishes second in the first election. This flaw is described fully in the document Understanding IRV which is posted above. This voting discrepancy violates the one person, one vote democratic philosophy. Since one person, one vote is actually the voting law as described in the Supreme Court ruling Reynolds v. Sims supporting such a method opens this IRV process to a significant legal challenge. A ruling against this form IRV could set your cause back decades. Even though Aspen represents a unique voting situation other municipalities and states will be reluctant to support a system that is being litigated or overturned in Aspen.
If you feel that equality among voters is not violated in an iterated SE system please explain your point of view. If after analysis you agree that voter equality is not maintained I would urge you to reconsider FairVote’s support of this system.
Douglas
DOESN'T 2+2=4?
I must have missed Su Lum's column on her dachshund's experience of getting his toe nails clipped which prompted a response from Elizabeth Milias whose response prompted a response from Hilary Burgess of Woody Creek.
My concern is not so much the terrifying experience the dog must have had as evidenced by his involuntary bodily response because the dog will get over the experience of the essential nail clipping visit.
What does concern me is Aspen Times reporter Carolyn Sackasiason writes:
"This spring, Aspen residents will vote for their favorite City Council members in an election system that has never been tried before in the United States....The Aspen City Council on Monday is expected to decide between two instant runoff voting (IVR) methods, both of which would produce different outcomes based on mathematical probabilities."
In America, democracy rules and the majority vote wins. To me that means, 2 + 2 = 4? Is it really constitutional to have an election where different outcomes would be produced depending on the mathematical probablities?
Thank you for clarifyng further. Also, can I vote for the same person in seat #1, seat #2 and sest #3 or do I have to vote for three different people for the three seats?
Thanks.
Toni
Toni, the reason different issues arise is in the attempt to have a 50+% majority. If 10 candidates run for 2 seats it is unlikely any one of them will receive 50+% of the vote. So IRV attempts to simulate the runoff procedure by eliminating candidates so that at some point in the voting there are only 4 candidates running for two seats or two candidates running for 1 seat. By reducing the number of candidates it can be assured that someone will receive 50+% of the votes. Depending on if you choose to count the top one or two votes, eliminate candidates one at a time or in a group, or fill fill one seat at a time or both seats at the same time all can lead to different outcomes.
To answer your other question each candidate should only receive one ranking from you, if someone is your favorite you should just rank them first. Ranking them again as your second and third choice will not be counted for anything (and potentially cause your ballot to not be counted as a spoiled ballot).
Douglas
Jack Johnson's proposal is clearly the least problematic for Aspen's parlaimentary style Council voting scheme (which was problematic in my opinion BEFORE Irv, but thats another conversation). While it is comforting that other western nations use this particular IRV system, those facts do not ameliorate the process gaps used by the City in moving towards adoption.
(PAUL SAYS HE MADE AN ERROR IN THE ABOVE STATEMENT RE: JOHNSON'S PROPOSAL) Marilyn
Asking voters to approve a concept as opposed to a tested and final process was the first mistake, and Aspen's voters, most of whom, like the Irish and the Aussies, would rather party than pay attention to silly things like elections, fell for the efficiency argument without even considering the impact on their opportunity to deliberate their options for leadership in a, well, deliberative manner. Instead they chose the track meet scoring method.
Most importantly, Ireland and Austrialia's use of Irv does not address the critical concept (and legal principal) in US election law summarized above by Mr Siegel - that of one person one vote. As I am sure you know US political history is filled with examples of attempts to manipulate our election processes. Incorporating Irv into multi seat races adds the dimension of mathematical probability to the equation in ways heretofor not anticipated.
Irv unquestionably creates the opportunity for votes to count differently (in comparison to each other) depending upon who you vote for and also the number of candidates for whom you cast a ballot (assuming not all voters will cast a ballot for every candidate).
For me, this alone is reason to scrap the concept altogether, but since I am clearly in the minorty on that matter - at least in Aspen - I think preserving the principal of one person one vote is the most important factor at the moment for the Aspen City Council to consider in their choice of options.
Paul
Copy of Letter to Mayor Ireland and City Council Members
Hi guys and you too Jackie
I attended the IRV meeting with Jim True last week and was thoroughly intrigued, a bit disturbed, and somewhat confused.
I would like to make a couple of comments concerning IRV and it's potential ramifications on the next election.
The one concerning single seat(mayor) is relatively easy to understand.
But the Multiple seat methods under consideration are complex and confusing. And that's with me attending the meeting and studying at home.
There are multiple areas of contention that need to be addressed.
Failure to consider the following is courting disaster that will only add to some of the problems Council has had to face in the last few months. No matter what system that may be chosen, the issues outlined below need to be answered.
1) the voters of Aspen will need serious education to make the proper decisions that they want to make. By this, I mean if they are not totally aware of all the possibilities, they could easily elect a third, fourth, or even fifth choice. And this might happen even with an educated electorate.
2) this system(whichever one you choose) has not even had the software written for it yet. Is it even going to be tested before the election and if so how?? What if it fails on election day?
3) my understanding is if any of the candidates are unhappy with the results that they can demand a physical recount with possible litigation if unhappy with any part of the election.
These are only some of the potential problems and issues. Are you really comfortable with what you are contemplating here?
Why are we so concerned with implementing a new, unproven system with so much at stake? Let's take a step back; take a deep breath; and pick a system that we can test with complete transparency and see if it is actually what we want.
Let's do a traditional election then experiment with whatever method of IRV that you want for the next election.
Thanks for your consideration.
Michael Wampler
Letter sent to Council 2.23.09
Mayor Ireland, and Councilmembers Johnson, Kasabach, Romero and Skadron,
By now, you know my objections to the IRV methodology process that Council put in place, so I won’t repeat those points.
It seems apparent that public did not understand the risks in what they voted for, and Council made no disclosures about such matters at that time of the November, 2007 ballot measure.
Council supported the vote only referencing efficiency and financial savings. Inadequate disclosures, it seems, for a ballot measure concerning election methods, in which Council would be given so much discretion.
I am posing the following questions, and requesting that the answers be made available to the public prior to a decision on IRV methodology by Council:
1) Were accurate and full disclosures about IRV complexities made to public at time of the November, 2007 ballot question? (Enough to withstand potential legal challenges?)
2) Do you believe that the public understood the risks inherent in the IRV system when they voted on it? (including that methods under consideration could produce different outcomes? Apparently the Task Force did not understand that until well into their deliberations.) If not, what actions will be taken to remedy this?
3) What advice has the City received from election law and methodology experts on percentage of spoiled ballots to expect in this multi-seat ranking election? (Informal conversations with experienced attorneys lead me to believe that the spoiled ballot percentage will likely be high.) Has this been factored into the decision?
4) How will you educate the public to reduce the number of spoiled ballots?
5) Will voters be given materials which fully explain the risks and benefits of voting strategies with IRV? There is a problem of having the “shrewd, educated” voters and then all others casting their votes at election time.
6) How will you develop public confidence and the system given its complexities potential failures and the fact that several people on the task force had inherent conflicts of interest?
7) Is there a clear and easy to understand chart or memo which gives adequate presentations of all methods currently being considered so the pros and cons of each method are fairly presented for benefit of the public and Council?
8) What is the reason for denying a hand count of the ballots? (if the previous answer of cost is the reason, what is the cost estimate?)
9) What is the cost of the TrueBallot software, travel and consulting? Is it more expensive than several elections of hand-counting ballots?
10) Will the TrueBallot software be certified by Colorado Secretary of State in time for the May election? (It appears that IRV software is not easy to certify, based on some states’ records.)
11) Does Aspen’s tabulation method require certification by the Sec of State?
12) What independent testing will be conducted on the TrueBallot customized software?
13) What work has been done to assess the potential constitutional challenges (unequal votes, etc.)? Do we have opinions from specialist attorneys on this matter, given that we are breaking new ground with a multi-seat, 50%+1 system ?
14) Please make public the legal work done to conclude that the various methods under consideration are in compliance with Sec 3.2 , Article III of our Charter, requiring the “highest number of votes” and “…50% plus one vote or more of the votes cast”, as the terms “votes” and “votes cast” may be defined by our charter or appropriate Colorado Statute.
15) Was the public notice given for the 2nd reading today legally adequate when the City has not made the proposed procedures available to the public as promised? As of this morning, the proposed procedures were still not available on the web. The staff recommended method does not seem have written procedures for the public to review. How can the public engage it’s elected officials on the decision with so little information?
Please see some of the comments posted by members of the public on this subject at :
http://theredant.squarespace.com/red-ant-blog/2009/2/21/dont-know-much-about-algebra-part-ii.html#comments
and
http://theredant.squarespace.com/red-ant-blog/2009/2/16/dont-know-much-about-algebra-issue-29.html#comments
Thank you for your consideration in ensuring that the public has an adequate level of knowledge to engage in the public process as you make your decisions. The public has a direct interest in the election process running smoothly and fairly and being protected from legitimate legal challenges.
Marilyn Marks
Aspen, CO
Paul,
I understand your desire to maintain one person, one vote. In fact I think that consideration should come before all others when choosing an IRV system. I also understand that you desire a simple system that is easily understood. However of all the proposed systems Jack Johnson's while being simple leads to the greatest inequality among voters. The flaws in iterated SE lead to some voters’ second preference not count while other voters have their second preference fully count in the second seat. This is an issue that would occur in almost every election to a significant number of voters. This happens to voters even if they fill out their ballot fully and correctly. If you are interested in this issue you can read about it in Understanding IRV under the first fairvote proposal.
I fully agree with your reasoning that one person, one vote should be preserved as best as possible to implement IRV. Also you are correct that in depending on if people fully fill out their ballots or have errors on their ballot that every IRV system can lead to voter discrepancies. However in some of the proposals this risk is minimal or nonexistent when people fill out their ballots fully and without errors. At least in those proposals an educated and careful voter will have their vote count fully. Johnson’s proposal would lead to voter discrepancies between two voters who both fully filled out their ballots with no errors.
I also agree with your conclusion that IRV should be shelved (or scrapped) in Aspen. There is still no clear voting method and there is not a great deal of time to approve a method, educate the public, have counting software written and verified, educate the candidates, and have an election.
Douglas
Just circled back to this blog.
Marilyn - consecutive IRV elections to fill multiple seats was used for years to elect the Australian senate before they decided to change to a proportional representation system. I believe there have been other uses too, but true, the great majority of IRV elections have been for one seat.
Douglas -- I see consecutive IRV elections as an IRV election to fill one seat, followed by an IRV election to fill a second seat. Everyone has one vote in each of those contests, although backup preferences if their first choice loses. Everyone has the same voting power.
In a one-seat IRV election, some people have their ballot count the whole time for their first choice. Others have their backup choice count if their first choice loses. But all voters have equal voting power, and judges have held that principle up repeatedly, most recently in Minneapolis
I see it just the same In a two-seat IRV election. For anyone whose first choice doesn't win in the first IRV election, their ballots going to a second choice are backup choices - not giving them extra votes. For anyone whose first choice choice does win the first IRV election, their ballot is going to a second choice for the second seat because this is a majoritarian, winner-take-all system, as was the previous system and as was the intent of the council when it placed IRV on the ballot.
In a majoritarian system for two seats, some people elect two people and some none. Some people elect two people and others one. But with consecutive IRV elections, to help elect two you don't have to cast equally weighted votes - IRV is liberating in that lower choices don't count against higher choices.
I believe the voters supported improving their old runoff system, not just figuring at way that would come close to duplicating it Consecutive IRV elections meet that goal.
No, voters did not select IRV to improve the voting system!! Unfortunately it was only because they believed their govt who offered them a voting means to save time and money by just doing it all at once. Furthermore, why should we trust a system only used before in Australia (last I checked this is outside of the US) before they too changed their system to something else!?
Elizabeth. Instant runoff voting has a history in many places besides Australia. And I think voters DID want to improve the runoff - that's why the moved to a majority threshold of 50% plus one in council races, rather than the old non-majority threshold.
Rob,
Let me address your main first assertion that the elections for the two seats should be viewed completely independently. This is not the case, there are not two elections for two seats there is ONE election for two seats. This is a key difference to understanding the problems with iterated SE elections. This leads to a fundamental difference in influence that the Minnesota judge never had to consider; after the first candidate is elected a person’s lower ranked choice will count because their candidate was ELECTED not ELIMINATED. There is a distinct philosophical difference in allowing a voter to cast votes for a lower ranked candidate when their candidate is no longer in contention for office and counting a lower ranked preference when their candidate has been elected to office.
Now to address your second point that in all majority elections some people don’t have their votes count for a winner. You’re arguing against a straw man there, I never claimed that everyone should have their vote count for a winner. I stated that everyone should have their top two preferences count. Voters who cast their top preference for the candidate that finishes second for the first seat will rarely have their second preference counted. In a single seat IRV election that’s fine, however when filling the second seat now their first preference is equal to the second preference of voters who voted for the winner of the first seat. This can change the outcome of the election because the order of elimination in IRV can be very important. If there are two seats for city council all voters should have the opportunity to cast votes for two candidates. If the only way IRV can “liberate” me from having a lower preference count against a higher preference is to take my second preference away I’d rather not be liberated.
For people interested in how situations can change with iterated versus simultaneous implementations consider the following game. Take the four aces out of a standard deck of cards. There are two men each will be dealt one of the four cards. Each is asked to place a bet if the ace they receive will be red or black. If both men are dealt cards simultaneously each will have a 50% chance of winning their bet. However if instead of dealing simultaneously the cards are dealt one at a time first man gets a card, he either wins or loses with a 50% probability still. However the second man now has additional information he can look and see what color ace the first man received and place his bet for the opposite color giving him a 67% probability of winning. This isn’t a perfect parallel to what is happening with IRV but it does show how taking a simultaneous process and iterating it can influence a game or system.
Douglas
Let me add one voting scenario as showing how the different counting for second ranked votes can change the outcome of the election. Consider a town with 100 voters: 40 Republicans(R), 40 Democrats (D), 10 Moderate Republicans (MR) and 10 Moderate Democrats (MD). Assume there is a race of 4 candidates each representing 1 of the 4 groups. Consider the following rankings of voters.
39 R voters rank: R, MR, MD, D
1 R voter ranks: R, MD, MR, D
40 D voters rank: D, MD, MR, R
10 MR voters rank: MR, MD, R, D
9 MD voters rank: MD, MR, D, R
1 MD voter ranks: MD, MR
So if one vote is counted with SE the tallies after round 1 are 40-R, 40-D, 10-MR, 10-MD. The MR and MD candidates are eliminated and the race is between R and D. The R candidate wins 50-49 over the D candidate because of the one voter who only listed their top two choices.
Now in the race for the second seat because all the R voters have their second choice count the first round tally will be 40-D, 49-MR, and 11-MD. The MD candidate is eliminated and the MR candidate wins the second seat 60-40. So in a town almost evenly divided they end up with two candidates from the same party.
Note that head to head the outcomes would be:
MD beats MR, R, and D
MR beats R and D
R beats D.
So the candidate that could win a runoff against every other candidate is not elected.
If the only thing that changed in the election scenario was the incomplete ranking happening by a MR voter instead of the MD voter the D and MD would be elected because now the D’s second choice would count.
Now consider what happens when everyone’s top two votes count. The tally after the votes are cast is 40-R, 40-D, 59-MR, and 61-MD. So the MR and MD candidates are elected. This is the promise of IRV to encourage moderation and coalition building. Why do you propose a system that benefits the party/coalition of the winner of the first seat? If voter inequality doesn't bother you then at least this system not adhering to your stated principals should cause you to reconsider.
Marilyn,
WORstCESTER, THE CITY ATTORNEY, IS INDEED WORRIED about IRV.
Focus on him and corner him because this a federal and state issue where he'll meet his legal peers, who may not buy his sly manner of legal advice. He's not deep intellectually in the "Colorado Supreme Court case method" .....look at the advice he's given over the years to McCabe at APCHA.
You are bringing national sunshine to a parocial (sic) world that didn't realize the inpact Colorado 82 would have on their world. In their heart of hearts, they don't dislike you or the Red Ant,.... it's CDOT's Tripani who beat them at their own game. Tripani now works for Brinkerhoff (?).....