Wednesday, August 27, 2008

MASS IRV koolaid drinker claims to know that IRV was a success here in NC!

Folks - there are so many people out there who claim to know that IRV worked really well in NC who were nowhere near Cary when they cast or counted votes. Here's a blog from a guy in Mass - http://www.somerville.org - who claims to know how well IRV worked from videos created by FairVote and the exit poll managed by Bob hall from DemocracyNC and done by people ranging from real volunteers to the former IRV America Director from FairVote who admitted faking a southern accent when she interviewed voters (I believe it's her voice and/or arm in the video).

So I responded to his first posting and he replied to that. I tried to respond, but was unable to do so. Check out http://www.somervilleirv.org/?p=3#comment-18 and see what he wrote in response to me and see my response he wouldn't let me post to his blog.

Greg - my statements are not misleading or false. There was no objective measure of IRV success in North Carolina that was established before the elections. That is one reason why the Raleigh City Council didn't want to pilot IRV - there was concern that we wouldn't know if it worked or not.

And actually my truthful and hard-hitting arguments against IRV have been very successful where both sides of the issue have been presented to decision makers before they took their vote. In Raleigh, where the pro-IRV presenters outnumbered me 4 to 1 (and greater odds if you counted the pounds), no one in the Raleigh City Council wanted to pilot IRV after I got done. When both sides were heard, IRV went down in flames.

So pro-IRV people started to sneak around and talk to elected officials behind closed doors for weeks and months before they took a vote. So of course they voted to use IRV - they only heard one set of facts. That is how they got Cary and Hendersonville to go along.

Greg, you seem like an intelligent fellow even though you have obviously drunk the IRV kool-aid. It two educated and intelligent fellows like us can't even agree on how to count IRV votes and if IRV was a success, how do you think that the less educated and less literate voters in my state or other states will be able understand IRV? They won't understand it and if they don't understand it they won't bother to rank candidates if they do show up, or they won't show up at all. IRV is already turning into a math literacy test for college educated voters - it will be even worse for the voters who will show up to the polling places and not know they were expected to rank their choices in the first place.

The program didn't replace the runoff - it was a one-time experiment. Cary will have a traditional elections in 2009 unless they vote to pilot IRV again in 2009. Since Julie Robison, who originally supported IRV, not longer supports IRV, I doubt they will be able to get the votes to pilot it again.

And I never mentioned that the program was extended due to its success. Please do not put words in my mouth. IRV supporters claimed it was a success. Legislators never had the chance to fully debate how IRV worked in 2007. But it should be mentioned that even before the Cary Town Council voted to participate in the pilot, the NC State Board of Election knew in March 2007 that IRV would not be used in the 2008 election because it posed risks. What those risks are, they never said. And unfortunately for NC voters, most of our legislators were too incurious to ask what those risks were. The amendment to extend the pilot was submitted at the last minute, with no time to fully study the 2007 pilot.

Now you are deliberately trying to confuse people. You can't say that IRV ensures a majority winner of the non-exhausted ballots because if there are ANY non-exhausted ballots you still have to count them. You can't declare a winner if you still have votes to count!

You are wrong with your assertion that absentee and no-shows are not counted when determining the majority. Either IRV determines a majority winner by a number that is determined by the first column votes (a number which doesn't change because it has to include all the votes cast for the top two in the first column) or you are manufacturing a majority out of a plurality win by not counting all the ballots cast.

You can't compare people who don't cast second or third round choices to people who don't show up for a traditional runoff because in that race you don't carry forward the votes from the first election. And you are wrong in stating that claiming that ballots exhausted after the first or second columns aren't counted in the denominator when determining the majority - that is how it's done in San Francisco. You can not not use 3022 as a denominator in determining the majority because you are carrying the first column votes forward. If you kept the votes for the top two and changed the denominator, you really are advocating ENRON-style vote counting methods. What would you have the denominator be at the end of the second and third column tabulations?

The majority in the Cary race was based on the 3022 ballots cast in that race. A non-exhausted ballot is one that can still be counted. By the time Don Frantz got his 1401 votes, all the votes had been counted - there were no more that could be counted for him. But he still didn't have the 50% plus one vote (or 1512 votes) out of 3022 in order for him to win.

Going by your thought process, had there been no winner at the end of the first round, and only one total second or third column vote cast for either one of the two remaining candidates, you would still have a winner with only one more vote cast. You are essentially taking a plurality and manufacturing it into a preferential majority.

But one thing that I don't get about IRV supporters is that most won't admit that IRV does not always ensure a majority winner in all elections. Some do - and they did that in the legislation to use IRV for student elections at NCSU. Their IRV bill can be found at http://students.ncsu.edu/sgims/archive-87/bill/efficient-fair-elections-act-1064.html

§103. Requirements for Enactment of Preferential Balloting.
All statutes or other acts enacting preferential balloting for an elective contest shall indicate one of the following preferential balloting methodologies to be used:
(a) “IRV Majority” is §201 Instant Run-off Voting conducted with a §301 Full Preference Ballot. It is applicable to single winner races requiring majority election.

§301. Full Preference Ballot.
On any elective contest using a Full Preference Ballot, to cast a legal ballot the voter must rank all choices presented on the ballot. If a write-in allowed in the contest and used by the voter, the voter must rank it in addition to the choices presented on the ballot.

Comment
Only a full preference ballot ensures that in IRV/STV allocation, every ballot has a place to be allocated in every round. Otherwise, ballots may become exhausted and unallocated, potentially resulting in no majority or Quota in the final round.

They do require a full preference ballot - meaning that you have to rank every candidate in every race for any of your votes to be counted. In real adult elections in the USA, we don't force you to vote a full ballot in order for your votes to be counted. But at NCSU they take the total number of ballots submitted and that becomes the denominator from which a majority must be received in order to declare a winner using this method.

§201. Allocation Under Instant Run-off Voting.
Under Instant Run-off Voting, each vote shall be initially allocated to the highest preference on the voter’s legal ballot. If no candidate or option has votes equal to or greater than a majority of legal ballots, votes will be reallocated as further provided by this section. Reallocation shall proceed by determining the candidate or option with the least amount of votes as currently allocated, and then reallocating each of that candidate’s or option’s votes to the next remaining preference on each ballot that was a source of votes. Ties for least amount of votes shall be broken as provided by §211. Reallocation shall be repeated until a candidate or option has votes equal to at least a majority of legal ballots in the contest, and is thereby declared winner. No vote shall be allocated to a candidate or option previously eliminated.

But at NCSU they take the total number of ballots submitted and that becomes the denominator from which a majority must be received in order to declare a winner using this method. They don't keep changing the denominator as your statement would suggest is done in ballots where there are no second or third column votes to be tallied.

§221. Occurrence of Traditional Run-off Election under IRV.
For §201 Instant Run-off Voting in contests requiring a majority of votes to be elected, a traditional run-off election must be ordered when, due to exhaustion of preference ballots, no candidate receives a majority of votes in the final allocation round. Such run-off shall be conducted with the last two (2) candidates or options remaining in the allocation round, and shall use a non-preference ballot.

Comment
When IRV is used with a partial preference ballot, preference exhaustion may occur, necessitating an additional election to obtain a majority. This cannot occur with a full preference ballot except in the unusual case of write-in candidates reaching the final allocation round.

But where you don't require full preference balloting, you see it right there- preference exhaustion can occur. It did occur in Cary - 1401 votes is not the 50% plus one vote (or 1512 votes) of the 3022 votes cast and counted in the first round of the Cary District B race. At least at NCSU, the students admit that IRV won't always ensure a majority win by the time all the ballots are exhausted. it's a shame the adults who pushed IRV couldn't do that in Cary.

For those voters who show up at the polling places on election day and didn't know beforehand they were supposed to rank their choices, what is left for them to do? Going by the less than professionally conducted NC exit polls, 25% of Cary voters and 33% of Hendersonville voters had this problem. In San Francisco, about a third of the voters showed up not knowing they were expected to rank their choices in 2004 - it was almost HALF the votes the next year.

What would you have those voters do? Either donkey vote for candidates they know nothing about just to keep them from having to come back for the runoff when they might possibly have taken the time to learn about the remaining candidates, or just not count ANY of their votes? Either choice hardly seems democratic.

I have been curious about that - the Secretary who signed that letter wasn't even the secretary when IRV was used in Cary. She has no experience with IRV to know about the costs and transparency. The Chair has been pushing IRV for a long time, with some of his travels sponsored by FairVote (according to a statement he made at a legislative hearing in another state). He has made many less than accurate statements about the benefits of IRV, and has contradicted himself many times. It's just too bad that most members of the Fourth Estate haven't taken the time to check the facts before they coble up a story out of a press release.

Debra Goldberg, the secretary of the Wake BOE during the IRV pilot, totally backs up my claims. She was one of only 4 county election administrators in one county out of NC's 100 counties to have any direct experience tabulating IRV ballots. Her op-ed piece titled "Instant Runoff Voting is no solution, says election official who was there" can be found here: http://www.citizen-times.com/apps/pbcs.dll/article?AID=200880826043

There is nothing misleading about what I have posted. You weren’t even here to know what happened. There is no proof that the NC Legislature has any knowledge that IRV saved money, boosted turnout or was a success. There was no debate on those topics - they were asked at the very end of the legislative session to extend the pilot. And they only did so after first amending the extension by requiring that goals, standards and criteria for the implementation and evaluation of the pilot be developed consistent with NC election law - because there was A LOT of concern that the first pilot wasn't done accurately and/or lawfully. The new law states that if IRV can't be done consistent with our tough election laws - it won't be done.

And if there are some NC voters and editorial boards see IRV that way, it's not because they really know anything about what actually happened. They just happened to be hustled by the very well-funded IRV advocacy groups that spent tens of thousands of money and staff time into pushing IRV in NC with no verifiable way to determine if it worked or not. Taking a press release and turning it into a story or an op-ed piece, or publishing a ghost-written op-ed piece doesn't prove that IRV works.

3 comments:

somervilleirv said...

Greg, here.

Chris, you are completely mistaken or deliberately lying about my denying this response from being posted to the blog. I did delete one comment recently because it contained a single word -- "test" -- as if someone was testing the site or something. I haven't denied or deleted any other comment.

The fact that the Wake BOE and the NC legislature both renewed the IRV pilot program is a testament to its success. John Hudson (chairman of the Transylvania County Election Board, the president of District One Election Board Officials, and a member of the Executive Committee of State Election Board Officials) and John Gilbert and Sharon Everett (Chair and Secretary of the Wake County Board of Elections) both recently embarrassed your colleague Joyce McCloy in the pages of the Citizen Times and Raleigh News and Observer, respectively, debunking each of her (and your) arguments one by one.

For your information, I haven't drunk the "IRV kool-aid", if there such a thing. I have studied election methods for years and independently came to the conclusion that it is the best single-winner election method available. Perhaps you have drunk the "anti-IRV kool-aid".

IRV is well-understood by election experts to be a voting system that requires a "majority threshold". In voting criteria terms, it is said to satisfy the "Condorcet Loser" criterion, meaning the winner must be preferred by a majority over some other candidate.

According to your definition to "majority winner", there is no method that guarantees it. You said that the denominator should include "all the ballots cast". Well, then it only takes 50% + 1 blank ballots cast to ensure no one can ever receive a majority of "ballots cast".

My point is that it is silly to count deliberate abstentions when determining whether someone has a majority. When someone leaves the second and third columns blank on an IRV ballot in a public election, they are deliberately abstaining from the instant runoff election, should one take place.

You wrote: "had there been no winner at the end of the first round, and only one total second or third column vote cast for either one of the two remaining candidates, you would still have a winner with only one more vote cast. You are essentially taking a plurality and manufacturing it into a preferential majority."

There's no "manufactured" majority -- everyone but one person deliberately abstained from the runoff election.

Similarly, in the first round of a traditional runoff, if every voter casts an entirely blank ballot except for one person, you would still have a "majority winner" in the first round. Abstentions never count in public elections when determining whether someone has a majority.

Now there are some non-public contexts where blanks are not counted as abstentions, rather they are counted as a binding "none of the above" vote. In these situations, blanks can count in the denominator when determining whether someone has a "majority", and it is possible for no candidate to receive a majority. In these contexts, if no one receives a majority of all ballots cast (including those with blanks), there is a revote. This is the case in some private groups and college elections where revotes are feasible, but not in any public elections to my knowledge. Blanks in public elections are always interpreted as abstentions, and voters are well aware of that. If you don't put down a choice, you don't get a say -- nothing new here.

NC Voter said...

Yes, it is true, North Carolina allowed another IRV pilot. As usual, the IRV pilot was buried within a large omnibus election law bill where few lawmakers would notice it, and ultimately passed.
It would never have passed in a bill by itself.

Some lawmakers tried to amend it to delay or kill the experiment to 2011, but intense lobbying by well funded sources prevailed. Lawmakers were desperate to get out of session and go home, and several had items in the big bill that they cared about.

The little bit of good news is that lawmakers put several restrictions on the pilot and will require that this pilot adhere to existing election laws.

WUNC has the story of a last ditch bi partisan effort to kill the IRV pilot:

Instant runoff nearly went down in flames

More here at
http://irvbad4nc.blogspot.com/2008/08/instant-runoff-messing-up-north.html

IRV doesn't even work - see Have Third Parties Been "Punked" By Instant Runoff?
http://www.opednews.com/articles/Have-Third-Parties-Been-P-by-ncvoter-080821-150.html

Chris Telesca said...

Greg - I tried several times to post a response on your blog, but your blog wouldn't let me make that one posting. That is why I posted the test. But it is no lie for me to state that I was not able to place my posting on your blog.

The Wake BOE did not renew the IRV pilot program - where did you get that from? Mr. Hudson embarrassed himself by claiming that Joyce McCloy had no business writing about IRV since she had no experience either administering or witnesses an IRV election. Those two things should have stopped Mr. Hudson dead in his tracks since they apply to him as well.

And with all due respect to Ms. Everett, she didn't observe or administer an IRV election either. Dr. Gilbert is another matter entirely - he's been biased in favor of IRV for years, so much so that it blinded him to several irregularities with the Cary IRV election - #1 being that there was no majority winner as we were told IRV would ensure. So Joyce's arguments were not debunked by Gilbert or Hudson.

I wish you could hold an intelligent chat with me. If you start out a race where no one got a majority of the ballots where there were 3022 first column votes cast - you would know that there I was not counting blank ballots. Thus, we set the denominator for the race at 3022 votes, and the threshold under NC law was 50% plus one vote. Got it? 3022 first column votes contain no abstentions.

Now for IRV advocates to claim that IRV has a runoff in one election and not two, how can you claim then that you can change the denominator to only consist of the total votes received by the too-two candidates when you have no way of knowing what that denominator will be until you get done counting the votes? It especially makes no sense when you are also taking second and third column votes from candidates who where originally included in the first column vote total of 3022. To say that The denominator drops from 3022 to some lower number is manufacturing a majority out of ENRON-style vote counting!

Let us look at how utterly ridiculous it would be to have the sort of shifting denominators that Greg suggests:

At the end of the first column, here are the vote totals:

Frantz: 1150/3022 or 36.07%
Maxwell: 1075/3022 or 35.558%
Roseland: 793/3022 or 26.25%
3 write-ins: 3/3022 or 0.10%

We are not discussing undervotes or blank first column votes here. Since no one got 50% plus one of the 3022 cast first column votes, there was no winner.

Even before the Wake BOE counted the Absentee By Mail or Provisional Ballots, then began to process the 2nd and 3rd columm votes by sorting out the ballots with 796 1st column votes for Roseland and the write-ins from the the rest of the 1st column votes for Frantz and Maxwell.

Here is the result of that talley (adjusted for calculator error and missed count on 10/11/07):

Frantz: 248/796 or 47.51%
Maxwell: 274/796 or 52.490%

Those percentages are using 796 as a denominator. Clearly we can't do that - otherwise Maxwell would be the winner.

So if we add those second column votes to the first column votes, we get:

Frantz: 1399/3022 or 46.29%
Maxwell: 1349/3022 or 44.64%
No one has a majority of the vote using 3022 as a denominator

If you change the denominator, what do you change it to and why? And if you did change the denominator at this time - why can't you declare the winner right now - why do you count the 3rd column votes?

If you add up the totals of the Frantz and Maxwell votes - 1399 abd 1349 - you get 2748. IF you use 2748 as the denominator, then you get different percentages:

Frantz: 1399/2748 or 50.91%
Maxwell: 1349/2748 or 49.090%

Clearly using 2748 as a denominator, Frantz has a majority. But what happened to the 274 votes that you would subtract from the original denominator of 3022?

But the Wake BOE did not declare Frantz the winner with 1399 out of 2748. So they ran their tabulation procedure and found a total of 7 3rd column votes. Vicki Maxwell got 4/7 or 57.143% of those votes, and Frantz got 3/7 or 42.857%.

Using the original denominator of 3022, here is what you are left with:

Frantz: 1401/3022 or 46.38%
Maxwell: 1353/3022 or 44.79%
Clearly after counting the third column votes, no one has cleared 50% plus one vote out of 3022. So there is no winner.

But wait - the Wake BOE claims that Frantz is the winner with 1401, which was 50.871% - but of what number of votes? 2754 to be exact. But 2754 is 268 votes short of the 3022 votes originally cast in the first column - what is the justification for not counting those 268 first column votes toward the winning threshold?

Now if you are only going to take the total number of votes eventually cast for the top two in IRV, here is where my example of counting one and only one 2nd or 3rd column vote comes into play.

If at the end of the first column vote totals you had the original numbers:

Frantz: 1150/3022 or 36.07%
Maxwell: 1075/3022 or 35.558%
Roseland: 793/3022 or 26.25%
3 write-ins: 3/3022 or 0.10%

And then you only one one and only one 2nd/3rd column vote was cast - let's say for Maxwell - and you used the total number of votes for the two candidates as your denominator:

Frantz: 1150/2226 or 51.66%
Maxwell: 1076/2226 or 48.33%

How would you explain to voters that 1150 wasn't a majority on Election night, but somehow it becomes a majority with IRV? And how would you explain why you are using 3022 on Election night, but 2226 the next day - where did 796 votes disappear to?

For that matter, using that rationale, if no one who cast a first column vote voted for Maxwell or Frantz in the 2nd or 3rd column (because Roseland did tell his supporters to vote for him 1st, 2nd and 3rd) - you could have been left with the exact same number of votes at the end of the second column tally as you had at the end of the first - 1150 and 1075. How could you pull a majority using IRV from numbers that weren't a majority on Election night?