IRV
news
videos
costs
turnout
majorityfailure
minorities
Disabled
currentuse
sanfrancisco
Scotland
votingsystems
vendors
rawvotedata
fairvote
runoff elections
3rdparty
tallying
About
studies
Download
e-mail me



 

Instant Runoff Voting,  Voting Systems and Technology

IRV requires (incentivizes) more complex voting machines/technology that is still in the beta testing phase.

Any jurisdiction that switches to new IRV capable voting machines will be beta testing them.  That is beta - testing millions of dollars worth of computerized voting machines. These machines are considered to meet federal standards as long as they do not exceed a 9.2% failure rate in a 15-hour election day. San Francisco spent over a million dollars to provide IRV capability for their optical scanners, ES&S Optech Eagles.  After three years of beta testing the special IRV software created for them, problems were discovered:  the algorithm was flawed, undervotes weren't being reported in some cases, and also the machines couldn't read all popular types of ink. 

Fair Vote, the main organization promoting IRV,  found that voting machine technology is an obstacle to implementing "Instant Runoff Voting".  Fair Vote believes that pushing IRV will help them achieve their goal of "Proportional Voting". More complex systems must be purchased and are not readily available:

April 24, 2008.  Fair Vote Director, Rob Richie's statement at the April 24 2008 Federal Election Assistance Commission Roundtable Discussion: Richie advised that something needs to be done to make voting systems compatible with IRV. He said: "for instant runoff voting, or preferential voting methods, it often bangs up against the fact that voting equipment isn't flexible enough to handle these voting methods....  It has real life impact. It's also creating havoc for Pierce County, Washington, a county of 800,000 people which have a big county executive race and they don't know if their system is going to be ready."

June 22, 2007. Under the heading of "Building Proportional Voting Infrastructure", Fair Vote's website states: "One major obstacle currently in the way of proportional voting systems in many localities is the difficulty of adapting existing voting machines to new types of ballot.  Many voting machines are unable to cope with more sophisticated ballot designs, and in particular with the ranked ballots which systems such as choice voting and IRV require.  Even when machines are theoretically compatible with ranked ballots, machine manufacturers will often charge huge amounts of money for upgrades to localities looking to put ranked systems into place."  

2007. After 3 years of use, the Secretary of State reported problems San Francisco's IRV tabulating equipment - an anomaly in the RCV algorithm discovered and inaccurate reporting of overvoted and undervoted ballots.  See memo from then Secretary of State Bruce McPherson:

Date: June 20, 2007
 
Re: Response to Board of Supervisors Inquiry #20070515-002: Voting System Certification
 
...The Secretary of State’s staff report from 2006 explains the issues experienced during
testing which caused Secretary McPherson to approve one-time use of the system.
Some of these issues are noted below:
 
Equipment not reading marks on test ballots because the pens provided by ES&S
to mark the ballots contained the wrong type of ink.
 
Reporting issues that do not accurately reflect overvoted ballots cast in polling
places and that inaccurately reflect undervoted ballots for ranked-choice voting
(RCV) contests.
 
An anomaly in the RCV algorithm concerning the elimination of the lowest ranking
candidates who are tied.
 
Breakdown of two machines: one could not communicate with the memory card
that defined the election; the second machine stopped processing ballots and
started to reboot itself.
 
When the system was modified in 2004 to conduct RCV, the system was unable
to receive full federal approval because the system hardware did not meet federal
testing standards developed in 2002. The software source code review and
functional testing that took place to assess the ES&S system used older standards
set in 1990. To have received full approval, however, the system would have
needed to meet standards set in 2002.

IRV incentivizes more complex voting systems. In fact, Scotland switched from hand counted paper ballots to computerized voting machines for the first time when implementing STV in May, 2007. 

Dr. Rebecca Mercuri warned  about Scotland switching to a more complicated vote counting method and to computerized voting: "IRV and other proportional balloting methods have been proven to incentivize the introduction of electronic ballot tabulation in places where none previously was needed or has existed, and they further complicate what has become an increasingly closed process for the determination of election results."

Scottish officials followed the advice of  "experts" and switched from hand counted paper ballots to computerized voting in May '07 to support STV, a form of Instant Runoff:

3.3.1 Experience of counting STV elections elsewhere led the Scottish Executive to examine the feasibility of counting the ballot papers electronically. Following a procurement exercise in late 2005 DRS Data Services Limited (DRS) was selected as the preferred supplier to provide the equipment and support necessary to support such a process. DRS provided the e-counting services in the Greater London Authority / London Mayoral Elections in 2000 and 2004. DRS also involved Electoral Reform Services (ERS) in their proposals – ERS having extensive experience of STV elections. http://www.pkc.gov.uk/NR/rdonlyres/F26CB0C8-EE76-46CD-953A-43B39495261B/0/07271.pdf ("Electoral Reform Services" is related to the Electoral Reform Society, a non profit whose goal is to spread IRV).

The results were disastrous - see May 7, 2007 "Not so much an election as a national humiliation  - Scotland’s voters were treated with arrogance and contempt". 

Other jurisdictions in the US have not implimented IRV because their voting machines cannot accommodate the vote counting method. See "Current Use"  for more detail.   Lawmakers pushed IRV in Maine for awhile, but efforts stalled over implementation costs . Maine would have had to purchase voting machines for jurisdictionst that still hand count their ballots and second and third choice votes would have to be counted at central locations. Vermont's Secretary of State ordered a feasibility study , found that their current voting machines could not handle IRV for statewide contests, and further that operational costs such as postage, ballot printing and voter education would increase. 

Touch screen/Direct Record voting machines have been recommended as the most compatible voting systems for Instant Runoff Voting:

2003. The Los Angeles Voter Empowerment Circle recommended touch screen voting machines for IRV  to the state of California in 2003: "touch screen machines for DRE systems are also better able to ccommodate alternative voting methods such as Instant Runoff Voting.  We therefore believe that DRE systems are preferable to paper-based systems, such as punch cards or optical scans. 

2004. From the San Francisco Dept of Elections report on IRV in 2004.  Beginning in April 2002, the City and ES&S began discussions on how to meet the new Charter mandate. The discussions involved not only how to implement an RCV voting system, but how to properly modify an existing contract that did not contemplate RCV. The uncertainty on how to best proceed in the development, certification, and implementation of an RCV system lengthened the amount of time required to modify and finalize the parties’ agreement. During the initial discussions, ES&S considered the best approach was for the City to move to touch-screen voting systems, ..although there was no funding or widespread support to move away from the City’s paper ballot voting system.

ES&S realized that its current paper ballot system could not provide voters the opportunity to rank all candidates that qualified for the ballot. For instance, if 22 candidates qualified for one contest, the system could not accommodate voters making 22 selections in order of their preference among the candidates. Touch-screen systems could most likely accommodate the ranking of all candidates.

For absentee voting, however, the formatting for the paper ballots for the current optical scan system limits the number of choices. The RCV Charter amendment allows for voters to have no less than three selections for an RCV contest if it is not technically feasible for the system to allow for voters to rank all of the candidates on the ballot. Thus, the City agreed to have its system modified to allow voters three rankings among the qualified candidates appearing on the RCV ballot.

 
2007. North Carolina State Board of Elections circumvents law and develops touch screen "work around".  Instead of making the Instant Runoff Voting Pilot fit North Carolina's election transparency law , a "work around" has been crafted  that will in effect gut key sections of the law to fit IRV.  Because there is no software to accommodate IRV on North Carolina's (new) voting machines, and perhaps to make IRV attractive to the touch-screen counties like Henderson County, an uncertified work around was created to avoid the laborious manual sorting and counting of the 2nd and 3rd rounds. This work around consists of five pages of single spaced directions, and the slightest error could flip or alter election results.  Since Hendersonville did not have to hold a "runoff", this uncertified method was not tested.  By circumventing North Carolina's election integrity law, the State BoE has made touch-screens a more attractive option than optical scanners.
 
Optical Scanners and Manual Sorting Errors. In the Cary, North Carolina "instant runoff" pilot, since the voting machines could not tabulate IRV, the ballots had to be manually sorted for the “runoff”.  A few errors cascaded into a miscount, and ballots had to be recounted.  The election workers count didn't match the candidates' informal count. An "audit" was done (not in a public meeting) and it resulted in the ballots being recounted and a "correction" of the results.  See "Critics Take Runoff Concerns to Elections Board"  NBC 17 
To “automate” the counting process would mean spending millions of dollars on new voting machines

2007. San Francisco's Voting Machine Problems with the Automark:  Why did so many ballots have to be "re-made" for the San Francisco election?

California SOS Debra Bowen ordered that ballots that don't have 3 choices for the mayor's contest be re-made.  (The Sheriff or DA contests weren't affected, there were only 2 candidates for Sheriff, only 1 for DA). This is because the OpTech Eagles fail to count ballots using many popular kinds of ink.
 
From Electionline:
The ballot remaking comes at the direction of Secretary of State Debra Bowen (D) who has expressed concerns that San Francisco's ranked-choice voting (RCV) system by ES&S may not be able to read all ballots cast by voters. Ballots that don't have three choices marked for the mayor's race or are marked with an instrument other than a number 2 pencil or a pen with black or dark blue ink need to be remade.
Mayor Galvin Newsome made it worse
Newsome told voters to "make it simple for themselves by just voting for one choice for mayor."
 

By November 5th, 94% of absentee ballots did not have 3 choices for mayor, had to be remade:

From the San Francisco Chronicle:      And because an astounding 94 percent of the absentee ballots processed by Monday had to be remade because voters didn't list three choices for mayor, it has taken a lot of time. 

According to the Golden Gate Express , though - voting may have been lighter than usual:

As of October 29, with only a week to go before the election, just under 20,000 of the 145,000 San Francisco absentee ballots sent out by City Hall had been returned, according to Department of Elections executive assistant Giannina Miranda. She added that the turnout was well below usual figures that soon before an election.

 
The Electionline article mentions the remarking of ballots that had been marked by the Automark:
"Remaking ballots is already an accepted practice in some California counties that use the AutoMark ballot marker, said John Gideon, co-director of VotersUnite."

But this re-marking is not because of problems with the Automark's ink.  It is because the Automark is not federally certified with that voting systems such as San Francisco's Optech Eagles.

There are at least 3 other jurisdictions in California that combine the Automark with Diebold (machines that haven't been federally certified together) and have to remark ballots:  Marin, San Luis Obispo, and  Santa Barbara
 
 

Jurisdictions using a combination of Diebold and Automark.

 

Marin
Paper
Diebold optical scan
(AccuVote ES-2000)
P
ES&S touchscreen/optical scan
(AutoMARK)

 

San Luis Obispo
Paper
Diebold optical scan
(AccuVote ES-2000)
P
ES&S touchscreen/optical scan
(AutoMARK)

 

Santa Barbara
Paper
Diebold optical scan
(AccuVote ES-2000)
P
ES&S touchscreen/optical scan
(AutoMARK)

 

http://www.calvoter.org/issues/votingtech/currentdirectory.html

Why the Lawsuit?  San Francisco and the SOS filed suit against ES&S because the vendor distributed an uncertified version of the Automark.  That is against the law in California.  San Francisco borrowed certified Automarks from another county to use in this November Muni election.

From the San Francisco Chronicle: 

San Francisco is also demanding ES&S pay for problems with hardware used Tuesday by disabled voters.

In April 2006, San Francisco purchased 565 AutoMARK machines from ES&S for $3.5 million. City officials thought they were buying the Model A100 versions, which were certified as vote-marking hardware for voters who couldn't mark paper.

Instead, ES&S delivered Model A200s, which were not certified for Tuesday's election. City attorneys said ES&S executives didn't notify election officials of the switch — and ES&S included instruction manuals for Model A100.

When officials realized they had uncertified equipment, they borrowed 600 certified machines from Contra Costa County. San Francisco is demanding that ES&S pay the costs of moving the machines, and it's asking ES&S to provide certified machines for the February primary.

Kim Zetter of Wired News wrote about the difference  in the certified Automark and the uncertified Automark:

Winger said the differences between the A100 and A200 systems are significant and easily identified just through visual inspection (see the photos above at right). The size of the motherboard and the types of wiring inside the machines are just two examples of the differences.

 

2007 Report to the Vermont General Assembly by the Vermont Office of the Secretary of State     (Word document)
Instant Runoff Voting (IRV): Administrative Implementation Options and Costs

Appendices to the 2007 Report to the Vermont General Assembly by the Vermont Office of the Secretary of State     (PDF document)
Instant Runoff Voting (IRV): Administrative Implementation Options and Costs