A list of questions was sent to both Columbia County Elections Commissioners on Wednesday, seeking answers for the underlying causes of the confusion and delays in reporting results starting at 9 pm, November 5th.
Democratic Commissioner Virginia Martin brushed those questions aside (despite years of touting herself as a champion of transparency and integrity in elections). But her Republican counterpart Jason Nastke unexpectedly agreed to a half-hour phone interview, and provided extensive new details about what’s been going on. Below is an account of what Nastke said.
: : : : :
On Election Night, Nastke said, all voting machines were shut down at 9 pm, as usual: “We have a good process for that; it works. It was the same as we’ve done for my 10 years as a Commissioner.”
GOP BOE Commissioner Jason Naske
At that point, he explained, “I’m not looking at numbers. I’m making sure that everything is sealed, and everyone is doing their job. There’s a whole team taking in numbers. Some are being called in, while all of the machine results are transported down on chips. Those computer chip numbers were entered that night; when I left around 1 am, there were only two or three left to be loaded.”
Asked why some numbers appeared to be removed from the initial Election Day results between Tuesday night and Wednesday morning, Nastke said that the numbers posted should have included that day’s machine counts, and may have had some early voting numbers as well, but an issue arose about the numbers collected from 10 days of early balloting in Hudson, Copake and Valatie, the first year such voting took place.
“Once we realized there was an early vote count issue, we only kept the Election Day numbers up,” he said.
The chips and tapes containing the early voting numbers were waiting at the Board to be tallied once polls closed. However, by law they could not be counted before that, to avoid any results leaking out and influencing other voters.
“That night, we had to open the early voting machines” from the three sites, Nastke continued. “We’d never done that before. It took about an hour and 15 minutes to get the tapes out of the machines. We took the chips upstairs to load that data up.”
Two major new problems then emerged, in addition to a third they already knew about.
“When we first posted the combined early machine votes, the numbers did not match the [printed] tapes. So we took those numbers down. The State was involved. We had to reload everything, because not all of the data had transmitted [from the chips]. I think it was Wednesday afternoon that we looked at that… The data we’d downloaded didn’t match the tapes.”
Nastke described the technical issue as incomplete downloads akin to “when you download a webpage and you don’t have great connectivity, so you get the text but not the pictures.” He said that the second time, the data from the chip matched the printed tapes.
Contrary to some reported rumors, Nastke said there was no “zero issue,” referring to the requirement for election inspectors to make sure machines at their polling places are starting at zero, and are not retaining any data from a previous cycle or tests.
Once this was resolved to their satisfaction, he said “that data was reported independently for each poll site.” But the Board soon realized that in reporting it in this way, it potentially could “reveal a few people’s vote.”
Any County voter could cast an early ballot at any of the three early voting sites, regardless of which local town they live in. But if, hypothetically, only one person from Ancram cast an early vote in Valatie, then the Valatie results for Ancram would reflect only their vote. Someone who took an interest in public data about who voted could then correlate the results to that one person.
Nastke conceded that had the Board just merged the early voting results from all three sites, then the aggregate numbers would have prevented such accidental disclosure: “The problem was—and it was our fault—we did not set up the reporting initially to be merged as one… It’s great data if you’re a political junkie, but it’s not so good for protecting people’s vote.”
“So then we had to figure out how to merge that stuff,” the Republican Commissioner continued. “We had conference calls with the State” to figure it out. This, he said, explains why some early vote tallies were published and retracted several times, as they grappled with these issues.
: : : : : :
A third and even knottier problem had been known to the Board since the very start of early voting on Saturday, October 26th. Nastke gave this run-down of “why it occurred, and how it came about.”
“I was actually the first person to vote in Valatie,” he recalled. “I wanted to make sure it worked. The check-in process worked seamlessly.
“What didn’t work was the scanning machines we’d been using for 10 years. The inspectors restarted the machine—that didn’t resolve it. We tried someone else’s ballot—same issue. I then spoke with Ron Moore in Copake, and they were having the same issue, and we started to figure out it was happening at all three locations.
“I had the vendor on the phone. We reprogrammed the machine, which took two hours—same problem.”
Next up was a call to the State, and a call to the Fort Orange Press, which is a State-approed ballot printer. “I sent an image [of our ballot] for them to look at. When you’re a sanctioned vendor, they’re used to using the software. They had no problem recreating our ballots and printing a new set to use the next morning,” which would have been Sunday.
“So why didn’t the ballots scan? We didn’t know why. I spent a good chunk of the day with a ruler and a tape measure trying to figure it all out. We looked at the PDFs, trying to figure out what was wrong. Was the paper the wrong size? We buy the paper a whole year ahead. It measured up, and it was paper we’d used before. I finally quit in exhaustion that [Saturday] night.”
On Sunday, Nastke remembered, “the Fort Orange Press confirmed again that “our PDFs were perfect; our ballots were fine. Then they asked what type of printer we had, and had it been serviced or moved recently? Yes. It had been moved—due to the flea problem. We had a big debate about who could move it, County facilities or the company who sold it to us, Ricoh. This is a very expensive piece of equipment. So we decided to pay the company who sold it to move it.
“The same way I would expect tire people at the auto shop to know what do with my tires, I expected the same with the printer. But Fort Orange asked if it had been recalibrated by Ricoh, and were all the belts tightened? I didn’t know, that’s why we hired them to do it. So Fort Orange walked us through it on the phone, which was very nice of them. They knew just what was going on.”
Uncalibrated and adjusted after it was moved, Nastke came to understand, “the printer might pull one ballot correctly, but then as it prints each one gets a little more off. So the [voting machine] scanners weren’t seeing the paper properly. We did not have a paper problem. The paper was fine; we had an printer problem.”
What does the Election Law say to do when ballots won’t scan? Nastke says “If the scanners don’t scan, you have to open the emergency ballot box in the scanner, put them in there and they get scanned once it does start to work. That would be great in a normal world… but it was never going to work because of how the ballots were printed on the paper.” The scanner couldn’t handle the misprints, so they would have to be hand counted.
“So the ballots were locked up,” Nastke continued. “The State said by law you can’t hand count those either before 9 pm on Election Day… So all those votes sat there, until Tuesday night, when we brought a team in to hand count them. I wanted a fresh set of eyes who hadn’t been working all day. It’s a tedious process, it takes forever. Each ballot has somewhere between 42 and 50 fields, with little circles that have to be looked at. We didn’t have many fresh eyes who also had been through these counts before, because they had all worked as poll workers all day. So I had an exhausted staff.
“Could more people have solved the problem quicker? Maybe, but I didn’t want to have inexperienced counters. The people we had counted until about 12:30, then went home. It was more work than we anticipated it would be, especially for a tired and understandably grouchy staff, working for weeks with no break.”
The counting then continued on Wednesday, though staff “understandably didn’t get going until 11 am, because they hadn’t gotten home until 2-2:30 in the morning” on Election Night. The count continued throughout the week, while the Board was dealing with the other issues which had bedeviled this year’s election.
“We had the early machine vote problem, add to that the hand count [of unscanned ballots, all of which had to be double-checked, and then the Election Day numbers. So that’s what took until Friday.”
: : : : : :
An unresolved issue: The County Board still has not reported the aggregate results for each candidate, for both countywide and local races, only totaling up precincts for each separate party line.
So if one candidate had the Democratic and Working Families lines on the ballot, and the other had the GOP/Conservative and Independence line, the public and press have to add together each line to figure out who got the most total votes.
Nastke said they are still working on this. “Yesterday, we went through it all. We got all the data inputted, so that is being checked. As far as the totals, we will have those totals up, and I’d certainly like to have that before the absentee ballot count starts. Ideally, if we can, well before that. Our priority right now is to get the hand count checked.”
Reflecting on all that happened, Nastke said “You know, we live in an Instagram world, but counting votes still gets done with paper… The other thing that this has taught me going forward, I’d like to reach out to NTS, which is a vendor for us, to figure out a better way to tabulate results, so we can get them up in a cleaner way. Right now we’re using Access and Excel, which were not specifically designed for reporting results t o the public. If we could have something specifically geared toward reporting, that would be really helpful.”
Asked whether it appeared that early voting had absorbed many of the votes which normally are cast by absentee ballot, which in the past has resulted in protracted counts and challenges, Nastke noted that there are “1,350 absentees countywide as of yesterday. We’ll have data on how much early voting took up absentees, but last time it was over 4,000 absentees. Vote turnout was down a bit, but still that’s a big difference.” He also noted that the party affiliation of absentee ballots received is breaking down overwhelmingly Democratic, with about 900 Dem registrations, 200 Republicans and Conservatives, and the rest Independence and unaffiliated voters (NOP) making up the remainder.
As far as whether he intends to re-up for GOP Elections Commissioner job, Nastke said “That’s a conversation I have with my party every time before the term is up, and we’ll see what happens.”
Asked the same question twice, Democratic Commissioner Martin did not respond as of press time.
