Based on preliminary reports from the team of experts hiring New Hampshire to investigate an electoral discrepancy, it appears that a build-up of dust in the reading heads of optical scanning voting machines can cause paper wrinkles (possibly over several years) to interpret lines in postal ballot papers as votes. At a local competition in a city preliminary reports suggest this resulted in four Republican candidates for state representative being withdrawn about 300 votes each. That didn’t change the election result – the Republicans won anyway – but it does show that New Hampshire (and other states) may need to maintain the accuracy of their optical scanning voting machines by paying attention to three things:
- Routine risk mitigation audits to identify inaccuracies if / when they occur.
- Regularly clean the dust from reading heads with optical scanning; Pay attention to the calibration of the optical scan machines.
- Make sure the machines that automatically fold postal ballot papers (before sending them to voters) do not place the fold lines over the ovals for the voting targets. (The same goes for poll workers who hand-fold ballot papers.)
Everything I write in this series is based on public information from the state of New Hampshire, the City of Windham and the Tweets from the WindhamNHAuditors.
In the general election on November 3, 2020 in Windham, New Hampshire, the race for the state representative was very close – 24 votes difference – so that one candidate asked for a recount. The recount showed she had lost hundreds of votes – not just 24. The hand-counting of the optical scanned ballot papers matched to a shocking degree the numbers reported by the optical scanned voting machines. New Hampshire Citizens Wanted To Know: Were The Op Scan Machines Hacked? Were the devices configured incorrectly and were you reading marks in the wrong place on the paper? Did wrinkles in the postal ballot papers cause the op-scan machines to misread votes? Was the narrative itself wrong?
The Secretary of State said there was no provision in New Hampshire law that would allow him to recount or inspect the voting machines for hacking. For example, a Republican legislature tabled a forensic examination bill to find out what had happened. This bill passed the legislature unanimously:
… [T]its law authorizes and directs an examination of the ballot counting machines and their memory cards as well as the hand tables of the ballot papers in relation to the parliamentary elections on the 3rd.
An electoral forensic review team will be formed to complete the review described in Section 3 and will consist of: a person nominated by the City of Windham, a person jointly nominated by the Secretary of State and the Attorney General, [and one person to be chosen by those two auditors]. The results of the test shall not alter the official results of the Rockingham County’s House of Representatives race under the decision of the Voting Rights Commission dated November 25, 2020, while maintaining the re-count of that race.
Governor Sununu signed Law SB43 on April 12, 2021. On April 26th, the city of Windham voted Mark Lindeman the Verified Voting Foundation as auditor, and the Attorney General (on the advice of the Secretary of State) elected Harri Hursti, a well-known expert on the cybersecurity of voting machines and ATMs. These two were then selected Philipp Stark, University of California at Berkeley, as third examiner.
This is really a “dream team” of experts. They know what they are doing, have experience with electoral testing and voting machine forensic examination, and know how elections work. New Hampshire couldn’t have found anyone better prepared to get to the bottom of it.
What happened in Windham?
The town of Windham has a single polling station where 10,006 ballots were cast: 6,697 personally and 2,949 machine scanned and 80 hand counted votes (because the overseas “UOCAVA” votes have a different format than the machines do not accept).
Postal ballot papers are processed in New Hampshire at the polling station, on election day. That means there is a preprinted electoral roll. When a registered voter appears in person to vote, his or her name is found on the list and crossed out. A postal ballot envelope is processed by comparing the name of the voter (from the envelope) to the same preprinted voter roll and strikethrough it; then the postal ballot is removed from the envelope and scanned in the same voting machines which are used for personal voting.
This procedure is a way of checking whether or not the same person is voting absent more than once both absent and personal: your name only appears once in the list and can only be ticked once.
In Windham there were four voting machines with optical scanning, model “AccuVote OS” from Global Election Systems on November 3, 1998 and 2000. The 6697 personal and 2949 scannable postal votes were fed into these four devices during the day. It’s like putting a ballot in every machine every 10 or 15 seconds, all day long!
At the end of the election, each machine printed its result on a “result tape” – thermal receipt paper that recorded how many votes each candidate received. There were four “result bands”, one per machine; Here’s a small part of just one of them for the State Rep competition:
In that race, 8 candidates ran for 4 seats, and each voter could vote for 4 out of 8.
The electoral administrators handpicked the four voting machine result tapes (along with the 80 hand-counted ballot papers) on a paper worksheet to create an official vote return form, which was signed by the city clerk.
In the race for the state representative Kristi St. Laurent was in 5th place out of 4 seats, only 24 votes behind the 4th place candidate Julius Soti. So she asked for a recount. Under state law, these are done manually by the Secretary of State’s office in Concord (the state capital).
The results of the recount were:
|Kristi St. Laurent (D)||4,456||4,357||-99|
|Henri Azibert (D)||2,787||2,808||+21|
|Valerie Roman (D)||3,415||3,443||+28|
|Ioana Singureanu (D)||2,764||2,782||+18|
|Julius Soti (R)||4,480||4,777||+297|
|Mary griffin (r)||5,292||5,591||+299|
|Bob lynn (r)||4,786||5,089||+303|
|Charles McMahon (R)||5,256||5,554||+298|
As you can see, something is grossly wrong – either machine counting or hand counting or both. The people of New Hampshire and their lawmakers wanted to understand how this happened – and how to prevent it from happening in the future.
The audit team considered
- the voting machines (and their ballot programming and performance on the type of heavy use they saw last November);
- the ballot papers (and the associated notes) as kept by the Foreign Minister from the November recount
- and the context learned from polling officers in Windham and other New Hampshire cities about polling procedures.
What caused the discrepancy?
The audit team has not yet prepared its official report, so all I am describing here are only the preliminary results as described in their tweets.
The auditors monitored a careful recount by teams of 5 people (callers, checkers, 2 tallyers, flaggers). Members of each team included Democrats, Republicans, and Independents. The hand count was within 0.05% of the state’s official recount for each state representative candidate (state did not recount races for president, senator, etc.). So basically the official recount of the state was correct, the voting machines were wrong; why?
There is no evidence that the voting machines were hacked. Forensic research shows that the software in this device matches the reference device provided by the State Department (the audit team will continue to investigate this software to make sure the copy of the SoS is correct) and there was nothing unexpected on the memory cards.
Fold in the voting slip. Ballot papers marked by voters at the polling station were (usually) not folded. Postal ballot papers are printed identically; they were folded in thirds before being mailed to voters; Marked by the voters at home, folded again and sent back to the town clerk and then (on election day) scanned by the same voting machines as the ballot papers in the polling stations.
What if a fold goes through one of the “voting targets” (the ovals that voters are asked to fill in with a pen to indicate their choice)? The voting machine is not allegedly interpret a fold as a mark. However, the audit team noted that in many cases a fold was read as a marker:
The fold lines for postal votes were not always exactly in the same place, but on hundreds of ballot papers they crossed the target of Democratic candidate Kristi St. Laurent. Now consider what happens if a voter marks all 4 Republican candidates (and none of the 4 Democratic candidates) by blackening their ovals with a pen. When the optical scanner Even if a Democratic candidate interprets the fold line as a marker, then the machine “thinks” that there are 5 votes in a vote-for-4 contest; and that’s an overvaluation, well All of these votes are not counted.
Lots Voters voted all-party: all R or all D. As a result, hundreds of votes for the Republican candidates were not counted. A purely democratic vote with a fold line through a blackened target would not be affected.
In the recount where people looked at the votes, the votes were counted accurately. People are unlikely to interpret a fold line as a voice! So the official Windham results (after the recount) are trustworthy.
But many cities in New Hampshire use the same AccuVote optical voting machines. We can legitimately wonder whether some elections in other cities (which were not counted by hand) produced an incorrect result. I will discuss this question in my next post.
In my next article, I’ll examine:
- What do national voting machine standards say about how voting machines should distinguish fold lines from intentional voting markings?
- Could a buildup of dust cause the voting machine to misinterpret wrinkles rather than markings?
- What are election administrators in other states doing to avoid this problem?
- Should New Hampshire throw away its voting machines and buy new ones, or throw away its voting machines and count the votes by hand? Or are there any steps they could take to use the same machines in a trustworthy manner?
- What circumstances led to this issue in Windham 2020, and how could these corrective actions prevent something like this from happening again?
- Could improved technology in optical scanning voting machines be less prone to this problem?