Website Search

Tuesday, June 24, 2008

Instant Runoff was a disaster in Cary North Carolina

Instant runoff voting was not so great in Cary North Carolina

Even "one of the best Boards of Elections in the state of North Carolina" had trouble counting IRV ballots in the Cary NC in Oct 2007 and provisional ballots weren't counted until after :

1. It was difficult to count just 3,000 ballots correctly. Officials had to manually tally the IRV results for the Cary, NC “instant runoff”. There was confusion during the counting and ballots were miscounted and not properly allocated to the candidates. Friday, the day after the "runoff" or count of the 2nd round, the election director performed an audit, according to the media. Errors were discovered and the audit extended into a full blown recount:

Oct 12, 2007 Recount widens Frantz lead in Cary Matthew Eisley, Raleigh News & Observer A double-checking of votes today in Cary's razor-thin District B Town Council election showed that Don Frantz appears to be the unofficial winner after all.... But Elections Director Cherie Poucher said today that an audit of the votes found math mistakes: several votes for Frantz had been missed, and a group of 24 one-stop ballots had been counted twice for Maxwell. The new tally is 1,392 for Frantz and 1,339 for Maxwell, giving Frantz a 53-vote lead.

One small error cascaded into a miscount that had to be corrected at another date. Oct 30, 2007 Critics Take Runoff Concerns To Elections Board ..."What IRV does is violate one of the basic principals of election integrity, which is simplicity," said Perry Woods, a political consultant in Cary. He says a small glitch threw everything into turmoil. Basically, someone counted the same group of votes twice; the error was caught, and corrected after an audit. Wood says his problem is with how they conducted that audit. "In this case, they ended up recounting all the ballots again and calling it an audit," said Woods. "I felt like if they were doing that, the public should have been involved, so no doubt is there."

A Knightdale resident said. “If the best board of elections in North Carolina had this much trouble counting 3,000 votes, this is too dangerous to try statewide.” (January 22, 2008 Opinion mixed on Cary's instant-runoff trial )

In "Instant Runoff Voting – 17 Flaws and 3 Benefits" a description of the Cary count:

According to Chris Telesca who observed the IRV counting in Wake County, NC, to hand-process a little over 3000 paper ballots (after the first choice votes were counted on the op-scan machines) when there were only 3 candidates plus a few write-ins for the Cary district B, single member town council seat, and the counting went only two rounds it took 6 sorting stacks for each of 12 ballot groupings or precincts (8 precincts plus absentee by mail in Cary, board of elections one-stop site, the Cary one-stop site, provisional ballots- Cary, and possibly some transfer votes from another county which were eligible to vote in the Cary IRV contest) or 12 times 6 stacks = 72 stacks. Wake County officials decided to put each stack in a separate plastic bag to keep track. This would not be possible if there were more than one IRV contest because each contest requires independent sorting and stacking to count. The procedure was very complicated, but it was there in print. Even so, the Wake Board of Elections (BOE) didn’t follow it. There was no overhead projector so that observers could follow the process. Non Board members were sorting the ballots into stacks which was hard to follow. Nonetheless, observers and the Board came up with different totals at the end of the day. The next day, the different totals were determined to be caused by a calculator error that was discovered in an “audit” – that also discovered a few missing votes.

The “audit” – which had to have included going back into the previously sorted/stacked and counted ballots – was not done in public. It took 3.5 hours minimum to do the first expedited processing of the 3000 ballots, not including the non-public “audit”. If you proceeded at the same pace for a county commissioner election in 2008, it could take three teams of counters 350 hours to sort/stack and count 300,000 ballots for just one race. That is just ten hours short of nine weeks –more time than it would take to hold a runoff election 4 to 6 weeks later.

IRV is complicated to count, because each individual ballot has to be considered when deciding which ones advance to the "next round".

Why IS instant runoff so hard to count? Because IRV is not additive. There is no such thing as a "subtotal" in IRV. In IRV every single vote may have to be sent individually to the central agency (1,000,000·N numbers, i.e. 1000 times more communication). [Actually there are clever ways to reduce this, but it is still bad.] If the central agency then computes the winner, and then some location sends a correction, that may require redoing almost the whole computation over again. There could easily be 100 such corrections and so you'd have to redo everything 100 times. Combine this scenario with a near-tie and legal and extra-legal battle like in Bush Gore Florida 2000 over the validity of every vote, and this adds up to a complete nightmare for the election administrators.

The State Board of Elections issued 3 pages of instructions on how to count IRV elections. Instructions on counting optical scan IRV ballots are on pages 1- 3, and sample ballots are on page 5 (provided by the Rocky Mount Telegram)

2. The Wake BoE did NOT count the voters' provisional ballots before going to the 2nd round. I inquired and was told that these were later added into the totals. (considering how IRV can only be counted by considering each individual ballot, how were these added back in?)Tuesday, October 16, 2007 at the public canvassing of the election, the provisional ballots were approved and added back into the election(s).

Oct 17, 2007 Frantz wins Cary runoff Cary's new 'instant runoff' procedure settles a close Town Council election ...Because dozens of provisional ballots had yet to be verified, the narrow margin meant that Maxwell might still win.

3. Overvotes were not reported by the voting machines, because our voting machines cannot see/read the 2nd and 3rd choices, so an important protection was lost. Especially considering that one of the Cary candidates was telling voters to rank him 1, 2 and 3 - many voters' 2nd and 3rd choices would not count in a "runoff'.

4. IRV creates new costs for every election. IRV requires specialized voting machines and software, increased ballot printing expenditures and voter education, something North Carolina does poorly.

North Carolina’s voting machines cannot count IRV ballots. According to the State Board of Elections, “There are no provisions on ES&S equipment to tabulate IRV.” ( January 7, 2008 email from Keith Long, NC State Board of Elections Voting Systems Project Manager.

Wake County spent $9,000 on voter education for the Cary experiment, with advocacy organizations donating the rest. This figure is on the low end and amounts to about 8.5 cents per registered voter , with Cary having 76,258 voters. This is the cheapest I have ever heard of being proposed for IRV voter ed. We know that IRV proponents heavily augmented the voter education efforts.

Contrast that to San Francisco, one of the few IRV jurisdictions in the US, which spends approx $1.87 per registered voter on voter education. San Francisco had 421,094 registered voters in 2004, spent $776,000 on IRV related voter education , or $1.87 per registered voter, with $210,000 specifically allotted to the community organizations for their efforts. 700 public outreach events in one year were held.

How much would it cost to educate the entire state's voters about IRV?

If done for the cost of a first class stamp (.42 cents), with 5,810,420 registered voters, that would be $2.44 million at least. This will have to be repeated each time, and many may ignore the mailer. If done for the low ball unrealistic amount of 8.5 cents per registered voter, that amounts to approximately $500.000 or half a million dollars. Just for the most meager voter education for a very foreign way of voting.

Then there's the cost of printing ballots. Ranking candidates takes up a lot more of the ballot, what normally would have fit on half a page may end up taking 3 ballot pages.

Instant runoff voting would make a difficult process even more complicated and less transparent - we need to focus on counting ballots the plain old vanilla way:

On May 6 primary night, 15,000 ballots in Wake County and 2500 in Mecklenburg were double counted ( May 8, 2008 Mecklenburg, Wake find vote flaws News 14 Carolina, NC )

and 4,000 were omitted in Onslow County.” ( May 9, 2008 Thousands of votes missed in Tuesday tallies Jacksonville Daily News, NC )