Records: 95 Detroit poll books missing for several days; five never given to canvassers
Detroit elections officials waited several days to deliver nearly 100 poll books to Wayne County officials charged with certifying the presidential election, newly released documents show.
County clerk officials on Thursday released a memo to State Elections Director Chris Thomas that said 95 poll books from the 662 precincts weren’t available at the start of the canvass, which began the day after the Nov. 8 election. Five of those poll books, which contain the names of voters and ensure the integrity of elections, were never delivered to county canvassers and presumably remain missing.
The revelation comes atop other irregularities that have prompted a state audit. Among other issues, The Detroit News reported this week that voting machines registered more votes than they should have in one-third of all city precincts.
“I’m not happy with how Detroit handled this election at all,” said Krista Hartounian, chairwoman of the Wayne County Board of Canvassers, which certified the election.
“We had been seeing improvement, but this one was different. This one was off.”
Canvassers compare poll books with printouts from voting machines to ensure the number of people who signed in to vote match the number of ballots cast.
In heavily Democratic Detroit, the numbers didn’t match in 392 of 662 precincts or 59 percent. The discrepancies emerged during a statewide presidential recount that began last week and ended Friday following a decision by the Michigan Supreme Court.
Democrat Hillary Clinton overwhelmingly prevailed in Detroit and Wayne County, but Republican President-elect Donald Trump won Michigan by 10,704 votes.
Statewide, 10.6 percent of precincts in the 22 counties that began the retabulation process couldn’t be recounted because of a state law that bars recounts for unbalanced precincts or ballot boxes with broken seals.
Hartounian said Detroit officials were still delivering poll books to canvassers on the Friday and Saturday after the Tuesday election. By law, the canvassers have 14 days to certify general elections.
“The canvass was extremely pressed for time,” Hartounian said. “There was so much pressure. It was so tight, and Detroit was still delivering information until the very end.”
The memo from county elections official Jennifer Redmond to the state also shows poll books in 101 Detroit precincts were not delivered in sealed envelopes, as the law requires.
Separate county documents obtained by The News show that poll books in 17 precincts were missing seal numbers from ballot boxes, as is required by state law.
Another Detroit precinct contained 50 of the 306 ballots listed in a poll book, according to an observer for Trump.
Detroit Clerk Janice Winfrey and city elections director Daniel Baxter did not return several messages seeking comment.
Blaming old machines
Last week, Baxter placed much of the blame on what he called outdated, decade-old voting machines, saying 87 broke on Election Day. The city had a two-page ballot, and frequent jams led to inaccurate counts when workers failed to reset counters, he said.
A Detroit News analysis of statewide recount tallies, however, found other counties that used the same optical scanner performed just fine.
In fact, the counties that used the machine, the Election Systems & Software M100, had fewer unrecountable precincts than those who used two other voting machines in Michigan.
Excluding Wayne County, counties that used the M-100 had 6.2 percent of unrecountable precincts, compared with 11 percent for the Sequoia Optechs and 10 percent Premier Accuvotes, according to The News’ analysis.
The M100s, which are used in 55 percent of Michigan’s precincts, have a spotty reputation. In 2008, then-Oakland County Clerk Ruth Johnson, who is now secretary of state, urged federal officials to investigate after the optical scanners improperly counted 8 percent of ballots during testing.
“The same ballots, run through the same machines, yielded different results each time,” Johnson wrote to the Election Assistance Commission, an agency that administers federal payments to states to buy voting machines.
Johnson wrote that vendors blamed dust and debris inside the machines. A Michigan Senate Fiscal Agency report noted the machines still run on the Windows XP operating system, which Microsoft has not sold since 2008 and for which it stopped providing support and security updates in 2014.
A spokesman for Election Systems & Software did not return a message Thursday.
Gisgie Gendreau, a spokeswoman for Johnson’s office, said in an email the secretary of state is confident in the accuracy of the M100 voting machines used in Detroit and other parts of the state “assuming they are properly coded for the particular election.”
There were fewer than 30 machines that required Election Day maintenance, she said. The department pays for half of the maintenance costs in Detroit and other jurisdictions with electronic poll books.
Gendreau told The News the state is not aware of any vote count irregularities in Detroit connected to the kind of machine “logic errors” Johnson highlighted eight years ago.
“The issues we have seen from Detroit are issues with too many or not enough ballots in the ballot container, few of which were actually reconciled either in the county canvass or in the recount, meaning they may or may not be out of balance,” she explained. “The machine logic software (firmware) would have no bearing on this.”
State Rep. Fred Durhal III, D-Detroit, said the state “has a responsibility to take a closer look” at Detroit.
“If you have faulty equipment, it can definitely cause an issue in Detroit, because you’re counting more votes than any other city,” said Durhal, who pushed for more funding for new voting machines. “I don’t think fraud is involved, I want to make that clear, but I think we’ve got to get to the bottom of what happened.”
It’s unclear how many extra votes were counted in Detroit. That’s because tallies were off by five or more votes in 52 Detroit precincts, but county officials will not release the exact number of discrepancies in each precinct. The News has filed a Freedom of Information Act request for the information.
But the county records do show exact numbers for precincts that were off by four or fewer votes.
The partial records indicate machines tallied at least 388 more votes than they should have. That’s 0.16 percent of the 248,000 ballots cast in the city that voted for Clinton 95 percent to 3 percent over Trump.
Voting machine replacements set
The Secretary of State’s office is planning to replace voting machines across the state by the 2018 gubernatorial primary and is expected to announce contract details in early 2017, said spokesman Fred Woodhams.
The state has roughly $30 million in federal funds leftover from the 2002 Help America Vote Act, he said, and state legislators appropriated another $10 million combined in the two most recent budgets.
But replacing voting machines across the state is expected to cost closer to $45 million to $50 million, said Bureau of Elections Director Chris Thomas, so the state will require local governments to shoulder a portion of the costs to replace equipment, likely $1,000 or $2,000 per precinct.
For Detroit, it could mean a tab approaching $1 million.
“Local communities will have ample time to prepare and budget for that,” Woodhams said. “Some may be ready before others, but we want to have new machines in place for August 2018, which is really the next statewide election.”
Michigan last bought new voting machines in 2004 and 2005, and those optical scanning devices are now effectively obsolete, Thomas said.
“The more tech they’ve shoved into these things, the more they’re like your PC,” he said. “You have a PC that’s 10 years old, you’re probably frustrated.”
In Washtenaw County, of 150 precincts, 15.3 percent, were not recountable. The county uses the Premier Accuvote voting machine.
“It’s passed its useful lifespan, and hopefully it will be replaced,” said Ed Golembiewski, the county’s elections director.
Staff Writer Chad Livengood contributed.