Iowa Democratic Party to release partial results later on Tuesday after manually verifying data against paper backups.
Nearly 1,700 precinct caucus sites, tens of thousands of people and 10 Democratic candidates for the US 2020 presidential elections.
The start of Iowa’s Democratic caucuses on Monday night appeared to go off without any major issues. That was, however, until hours went by without results, forcing the Iowa Democratic Party (IDP) to announce there was a problem.
Nearly 48 hours after the caucuses began, only partial results were out. With 86 percent of the precincts reporting, the IDP said former South Bend, Indiana, Mayor Pete Buttigieg led in the state delegate race. Senators Bernie Sanders and Elizabeth Warren were in second and third place and former Vice President Joe Biden trailed in fourth place.
It is unclear when the final votes are expected, but IDP Chair Troy Price apologised for Monday night’s errors and delays.
Here’s what we know about what went wrong.
What was the issue?
The problem, according to the IDP, was with the code in a mobile phone app that precincts used to report results.
Price said in a statement that the delay in reporting results was not due to a hack and that independent cybersecurity consultants had tested the systems in preparation for the caucuses.
“We have every indication that our systems were secure and there was not a cybersecurity intrusion,” Price said.
But he added: “While the app was recording data accurately, it was reporting out only partial data. We have determined that this was due to a coding issue in the reporting system.”
He noted that the issue had been fixed, and the “application’s reporting issue did not impact the ability of precinct chairs to report data accurately”.
Why was an app being used?
The Iowa caucus is not a ballot system – Democrats vote by actually showing up to designated gyms, churches, mosques and other locations throughout the state and, for the first time, around the world to physically stand together in groups to show their support for their preferred candidate.
They go through two rounds. In the first round, each candidate must have 15 percent of people’s votes to be viable for the next round. If their candidate is not viable in the first round, people can walk over and join another group or if close to viability, try to convince others from other non-viable groups to join their side.
Many precincts were expected to enter the results of these rounds into a new mobile app, specifically designed for the caucuses. But, as one precinct chair told KCCI Des Moines, the app would not let her enter the number of participants in the second round.
Waukee Precinct 2 Chair Tina Weber told the local news channel the app had “helpful” directions. “It’s really slick,” she said. But when she got to the second round in the app, she said she was not able to enter the number of participants who supported Andrew Yang. Because of this issue with the app, she said: “I ended up with more people than there were actual people there.”
The app was developed by Shadow Inc, a Washington, DC-based company headed by CEO Gerard Niemira, who previously worked on Hillary Clinton’s campaign.
State campaign finance disclosures show the IDP paid Shadow two instalments in November and December 2019, totalling $63,000, for web development and travel.
Shadow Inc said in a statement on Twitter that it regretted the delay in the reporting of Monday night’s caucuses, but “the issue did not affect the underlying caucus results data”.
It added that the company had “already corrected the underlying technology issue”.
Was the app properly tested?
The IDP maintains that its “systems were tested by independent cybersecurity consultants” in preparation for Monday’s caucuses.
But Douglas Jones, computer science professor at the University of Iowa, told Al Jazeera he had researched the app in recent months and grown increasingly worried about it.
He said the IDP had been secretive about the app and said political parties must be more transparent, given that their role in the election process is set out in law and the decisions they make can alter the structure of government.
In the run-up to the Iowa caucuses, he anticipated that it would be impossible to properly test the software before the real vote. “The risk of failure with a ‘live test’ is very high, and without a realistic test, I can’t imagine trusting such software,” he said.
With nearly 1,700 precincts in Iowa, he knew that even in precincts operated by extremely well trained and prepared workers, they would likely make mistakes.
“That tests a system far more thoroughly than any planned testing can do. It means that simulating that environment without a real election would be extraordinarily difficult and expensive,” he said.
“So, when you deploy software that’s going to face that kind of test, you have to be extremely careful. And it appears that they were not always careful,” he added.
“I think they were overconfident and under-resourced, and that combination is dangerous,” he said of the IDP. He added that it would have been cheaper and more reliable to hire phone operators to do the same job.
He said that, in general, there will always be a long list of security concerns with apps involved in caucuses and elections. When a user downloads an app, she must trust the source she is downloading it from. And once it is on her phone, the app is vulnerable to all other free apps she downloaded that could be monitoring her, for advertising purposes or other nefarious reasons.
Any app that can be downloaded would be communicating with a server, he explained, so someone with bad intentions could download the same app and attack that server.
Ultimately, whether it is due to human error or an outside attacker, if a vote is misreported, it can lead candidates to mistakenly withdraw from the race. If that happens, it is too late for that candidate to recover, Jones said.
“The entire primary process, from the Iowa caucuses all the way through to the National Convention, has that risk associated with it because the unofficial results reporting at each stage in the process is used by the candidates to make important decisions about their campaigns in the next round of the race,” he said.
Jones said he was nervous about any other caucuses or elections using apps to record or report votes. “Voting by cellphone, we’re just not there,” he said.
What else happened?
Some precincts also had trouble reporting results by phone.
Shawn Sebastian, caucus secretary for Story County Precinct 1-1, tweeted that it took him more than two hours to report his precinct’s results by phone. He clarified to Al Jazeera that his precinct did not use the app to report results, only the phone, saying that there were also initial problems downloading it.
“Give the problems, we planned on just reporting by phone,” Sebastian told Al Jazeera.
His call finally went through a little after 11pm local time.
“It took almost 20 minutes to report,” he tweeted. “Starting to understand why I was on hold for so long.”
These are the results that I've been trying to report for about 2 hours now. I was on hold for about 90 min, CNN wanted to interview about my experience, and in that moment I got off hold … but couldn't get on fast enough and they hung up on me.
I'm back on hold.
— Shawn Sebastian 🐺 (@shawnsebastian) February 4, 2020
The IDP has not commented directly on the issues with the phone line. The party had not responded to Al Jazeera’s request for comment at the time of publication.
How are the results being verified?
As the IDP investigated the issue, its staff executed a backup plan and entered the data manually. “This took longer than expected,” the IDP said.
The IDP investigation determined that the underlying data collected by the app was “sound”.
“Because of the required paper documentation, we have been able to verify that the data recorded in the app and used to calculate State Delegate Equivalents is valid and accurate. Precinct level results are still being reported to the IDP.”
The IDP is using tech systems, photos of results and a paper trail to validate the results, IDP communications director Mandy McClure added in a statement.
The Associated Press reported that volunteers and Democratic party staffers were being sent across the state to gather paper results to verify the results.
What happens next?
It is unclear what time the final results will be announced.
Meanwhile, the issue in Iowa’s caucuses has many worried that a similar problem may present itself in Nevada’s caucuses, scheduled for February 22.
The Nevada Democratic Party said it was confident the same issue would not recur.
“NV Dems can confidently say that what happened in the Iowa caucus last night will not happen in Nevada on February 22,” William McCurdy II, Nevada’s state Democratic chair said in a statement.
“We will not be employing the same app or vendor used in the Iowa caucus,” McCurdy II said. “We had already developed a series of backups and redundant reporting systems, and are currently evaluating the best path forward.”
What about New Hampshire?
The next Democratic contest is the New Hampshire primary on February 11. Similar to Iowa, the way New Hampshire votes can have a huge effect on the rest of the race.
But the difference is that the Iowa caucuses are run by party officials who do not have as much experience as state and local elections staff who will run the New Hampshire primary, ProPublica pointed out. Due to the concern about poorly run caucuses, the 2020 Democratic contest has fewer caucus races and more primaries compared with 2016.
Monday’s chaos has re-energised the debate about whether Iowa should hold primaries instead.