-Aspen's historic May 5, 2009 IRV election audited as single ballots- 5/5/09 Aspen CO held an instant runoff election (IRV) for mayor and 2 council members. Interpreted contents of each ballot, scanned by True Ballot, were publicly released. Open records requests for a CD of image scans were denied. Aspen has been sued to protect records from destruction and to allow inspection of the scanned ballot files. A Court of Appeals ruling holds that unidentifiable ballots are public records.

Search this and related blogs

Sunday, April 18, 2010

Criticism of Caleb Kleppner Guest Editorial in Aspen Times Sept. 17, 2009: Aspen Election Transparency

This is an email written by Harvie Branscomb to Caleb Kleppner of True Ballot Inc. Sept. 17th, 2009 in response to a guest editorial written by Kleppner for the Aspen Times.  This email was not responded to in written form but there was a telephone conversation between Caleb Kleppner and Harvie Branscomb. This email, critical of the Caleb Kleppner's op-ed statement,  corresponds to the recent email of Marilyn Marks to members of the Aspen Election Commission linked here: http://aspenelectionreview.blogspot.com/2010/04/city-of-aspen-trueballot-inc.html

An open records request was filed to learn about city cooperation in the writing of this op-ed.  The result is found in the form of a pdf by clicking here .  On the pdf you will find Marilyn Marks Sept. 17, 2009 criticism of the op-ed as well as an email from Caleb Kleppner to Aspen City Clerk Kathryn Koch offering to cooperate on writing the op-ed and also a submission of a draft to Kathryn Koch for edits or modifications.
[above is by Harvie Branscomb]


You have taken the salesman’s route of promoting transparency in Aspen, which is my agenda too, but you have rested your case on some misconceptions about the election which force me to argue with you.  I am sorry you did that, because I want to support the transparency of True Ballot’s system regardless of the election method. Unfortunately by glossing over facts as you (and the City of Aspen) have done, the public does not get a clear picture of what took place and where it needs to be improved.  I have been circumspect in my criticisms of the Aspen election, but this new publication forces me to expose additional details as I know them to the public.

Here are the line by line exceptions I have to your editorial. Please give me your rebuttal ASAP. As a first hand observer, I will rely on your corrections of my statements where your corrections are definitive from  your first hand experience.

(original guest editorial from the Aspen Times, interspersed with commentary by Harvie Branscomb in yellow.)

Caleb Kleppner
Special to The Aspen Times
Some recent commentary on the instant runoff (IRV) election has distorted the facts and missed the big picture about Aspen's May election.

HB>; Caleb, you are correct, and this editorial of yours is an example of that, unfortunately.  This letter will seek to elaborate on your distortions. I think perhaps you are correct about the big picture, but this picture must not be founded upon mistaken facts.

The big picture is that the election was a model of transparency, verifiability and honesty, and the post-election audit was among the most thorough ever conducted. It also had the highest voter turnout in a municipal election in Aspen history and elected a clear majority winner for mayor without a June runoff election.

HB>; It ought not be said that the election was a model, but it does represent a template for better transparency and verifiability- which was not fully correctly implemented.  There is no connection between the election type or the procedures, and honesty. This pronouncement of honesty can only be based on the simple, and limited observations of Mr. Caleb Kleppner and perhaps his colleagues, and it is his opinion.  I have no interest in or basis for arguing with this portion of his opinion The declaration that the election achieved a “majority” fails to expose the means by which this majority for the city council contest was achieved, by redefining the size of the electorate among whom the majority is to be achieved.

Let's start with transparency. Ballots were first counted in polling places using the county's Premier Accu-vote optical scanners. Then, in plain view of the public and cable viewers, TrueBallot Inc. scanned every ballot using commercial imaging scanners, processed the data on the images, and publicly reviewed every ballot not once but twice to make sure the computer interpretation of ballots matched human interpretation of voter intent. These images were projected publicly in City Hall and broadcast on cable television. The Election Commission reviewed all potentially ambiguous ballots to assure that they were counted as the voter intended according to state law. The IRV tallies were conducted and announced on Election Night.

HB>; Ballot pages were counted by the Premier Accuvote system and found to be discrepant with the True Ballot counts of pages. This issue has not been publicly resolved. Vote counts were only partially counted by Accuvote and the election cannot be tabulated from that alone. The True Ballot “public review” consisted of exposing the ballot to the people in the City Council chamber for at most a few seconds each. Two systems were operating simultaneously, so it would take a team of observers to practically participate in checking ballot interpretations at this time.  Subsequently it is reported that about 60 ballots were further reviewed based on errors located by software.  This additional step did not apply to all ballots, and it may be that poorly marked ballots were missed by this step. Only a portion of the ballot was exposed to the public each time, so it is misleading to claim that the ballot was functionally exposed twice.  The review of each ballot was not “to make sure the computer interpretations of ballots matched human interpretation of voter intent” unless some true ballot employee was presumably performing this function. Only some of the images were broadcast on TV. The Election Commission did not review “all potentially ambiguous ballots”. IRV tallies were conducted and announced  well after the end of the public TV show and during a process which few were able to observe or understand.

In this election, unlike virtually any other public election in the country, members of the public could observe the removal of ballots from voting machines, the transportation of the ballots to City Hall, the removal of ballots from sealed bags in City Hall, the scanning of ballots, and the review of ballots for voter intent.

HB>; Members of the public could not observe removal of ballots from the voting machines unless they were among limited assigned poll watchers-  Likewise the transportation to City Hall- this requires substantial coordination for good public oversight, and does not apply to mail –in ballots which arrive through the mail and are handled at the time and place of choice by the designated election official. You have little credibility on this topic as you did not to my knowledge see any ballots removed from voting machines.

But there's more. The city adopted an ordinance to make all election data public, and the city clerk has provided CD-ROMs with all of the election data to anyone who asked. This means that members of the public can perform independent verifications of the election results, and several have done so. The city also explained on its website what data would be produced in each stage of the processing and how to independently verify all of the data.

HB>; We know that the CD ROM does not contain “all of the election data”.  In particular it does not contain the museum question’s official results, or any Accuvote results, or the Accuvote’s audit logs, or the ballot inventory logs and any evidence of chain of custody and it does not contain the created ballot images. Members of the public can verify election outcome only based on trust of the interpretations provided by True Ballot.  This requires dependence upon the audit of the interpretations which is featured in the comments below.

It was also conducted honestly, with ample public notice of the rules and procedures that would be in effect, full observation of all steps, and full release of election data. And yes, when we made a mistake in the IRV tally for mayor — a mistake that affected a small number of ballots only after a candidate had already reached a majority of votes and therefore been elected — the city made that information public. But more importantly, the data necessary to verify the IRV tallies was already public, so if there had been any error affecting the outcome, it would have come to light.

HB>; Again a personal testament to honesty about which Mr. Kleppner can only testify to his own. However I am not aware of any evidence for dishonesty.  On the other hand, ample public notice was not provided for a number of critical procedures.  In particular, the software parameters used were changed without notice on election eve after testing results were obtained and this is a gross exception to best practice.  There was not “full observation of all steps”- in particular the bulk of the hand counting of the test ballots for logic and accuracy testing was conducted by a city attorney behind closed doors. This is also a topic which Mr. Kleppner has little credible direct information.  The presumption that publication of interpretations is sufficient to allow the completion of any truly independent audits of election outcome in timely fashion sufficient for an election contest is hopeful, but not protective of the public interest in the election outcome. While it is important that this data be made public as it was, it is also important that the election procedure include a full and independent public audit of its outcome prior to the end of the contest period. This was not done in Aspen for the May election.

And if all that wasn't enough, the city conducted a post-election audit that far exceeded anything required by law. First, the city randomly chose 10 percent of the ballots to make sure that the rankings on those ballots matched the electronic information stored. There were no discrepancies. Then the city performed independent verification that every ballot was tallied correctly in the IRV tallies. The independent tally for mayor is posted on the city's website, and again, there were no discrepancies in the mayoral and council tallies.

HB>; The city-conducted audit (which ought to have been conducted by a citizen managed board such as the Election Commission) did not “far exceed” anything required by law. In fact it fell short of reasonable auditing practice. While 10% of the ballots were touched for the audit of interpretations, these were split between two of the three contests on the ballot, and therefore it can only be stated that 5% of the ballots maximum were chosen for audit for any contest.  10% is an opportunistic exaggeration of which the City Council has already been made aware.  These 5% were not at all randomly chosen. For each contest they were in fact chosen from two City Clerk selected precincts and they were selected in a sequence in the order in which they were scanned, taking only a different starting point in the stack as a hopefully “randomizing” variable.  None of the early or mail-in ballots were audited, and these were subject to substantially different conditions of voting to the ballots in the group which were audited.  Again, a gross departure from best practice.  Mr. Kleppner says “the city performed independent verification”.  This phrase is an ironic paradox.  “Independent” means that parties other than the city are verifying.  Yet in this election not only did the city manage the partial and limited audit of interpretations, but the city’s contractor- True Ballot apparently conducted, belatedly, the remainder of the crucial audit,  off site, without public observation or City direction, determining that the election results in the mayoral race were defective.  This was also not really an “independent” verification.  Independent verification of this tabulation audit was possible, but not assured by the election procedure.  But no “independent” verification of the interpretations was done, nor is it possible for this to take place under the constraints enforced currently by the city.

You really can't have more transparency and verifiability than that, unless you make the actual ballot images public. Right now, state law doesn't permit the city to release the ballot images, but if it did, I'm sure it would happily release them. Of course, the city already publicly verified a random sample of 10 percent of the ballots, which should give rational members of the public pretty high confidence that all of the ballots were recorded correctly.

HB>; My comments above may have discredited the first statement here.  We have not seen the state law which “doesn’t permit the city to release the images”.  The audit actually conducted by the city would give no “rational member of the public” confidence that all the ballots were recorded correctly.

Whether you like IRV or hate IRV, Aspen's election was a model of transparency and verifiability, and American elections would be improved if they incorporated elements of Aspen's election.

HB>; I do not agree that Aspen was a “model” but I do agree that American elections would benefit from  more transparency and verifiability such as what Mr. Kleppner’s company provides.  For Aspen to become a model for a superior election, it is necessary to look at the details and come clean about the defects of this innovative election and to avoid making these mistakes again, whether you like or hate IRV.

Caleb Kleppner is a vice president at TrueBallot Inc., which has run elections for municipalities, labor unions, associations, state Democratic and Republican parties, and others over the past 15 years.

HB>; Harvie Branscomb is a Board Member of Coloradans for Voting Integrity, a Trustee of the Colorado Voter Group, and appointed by the Democratic Party to be Eagle County Canvass Board member.

No comments: