Not Rounding Errors - It's Supposed To Be This Way

Because  of the way the math was done at the caucuses, many precincts wound up with an extra delegate that got "assigned" to one of the candidates at the end of the caucus. We were able to look at worksheets from 18 precincts that were posted on Twitter. We found a strange process where numbers below .5 were being rounded up in multiple precincts. See the photos above for more examples.

Six of the precincts had the unusual "rounding up." One seemed to be a tie - which may have been decided with a coin toss. Of the six precincts with the "rounding up"  - 4 were given to Buttigieg, 2 to Biden, and 1 to Warren. 

We posted the results on Twitter, and at that point received an explanation about the math. We had originally called them "Rounding Errors." But it turns out they are not rounding errors. This is exactly the way that the Iowa Democratic Party wants the votes to be counted. 

As seen above in the instructions on page 15 of the Precinct Leader Manual, if there are extra delegates after the 2nd round of voting, the extra delegates are assigned to the candidate with the "highest decimal below 0.5."

There doesn't appear to be any fraud here. Just a bizarre and arcane set of rules.

You might think, no big deal, it's only one or two delegates, but because of the way the math is being done, it's possible there are an increased number of delegates that have been assigned this way. This is why: 

The delegates are assigned by a formula. 

After the 2nd round of voting (the 2nd assignment) the candidates' totals are all multiplied by the number of delegates that will be awarded in that precinct. If you look at the worksheets you can see examples of this. Then that number is divided the number of caucusgoers. 

Because a number of people have left by the end of the night, and some caucusgoers don't vote with a viable candidate, the number of actual votes could be well below the number of original caucus goers. Is this contributing to the precincts (about 1/3 of the ones we looked at) that have one or more leftover delegates that are being assigned by the twisted formula listed above? 

There are reportedly 1678 precincts. If 1/3 of them have an extra delegate that was assigned in this way, that could be approximately 500 delegates. In the official results, Buttigieg is currently leading Sanders by 4 delegates. So will this rule of assigning delegates to the candidate with the "highest decimal below 0.5." determine the winner of the Iowa Caucus? 

I spoke yesterday with Doug Jones, a computer science professor at the University of Iowa who has caucused for close to 40 years. Jones has also been active in the election reform movement for two decades, and is a co-author of Broken Ballots - Will Your Vote Count?  “If you follow the rules as written the math doesn't add up," he said. Looking at Cedar County FM as an example of the way the formula was implemented, he pointed out, "Basically everyone who left early was counted as Buttigieg, because of the way it rounded." His overall assessment was, "The rules as written don't lead to a sensible result.” 

This protocol also violates one of the basic principles of elections: that the number of voters has to match the number of cast votes. The Caucus math has a different number of voters than votes cast on many of the worksheets. 

There are quite a few posts online saying that dividing by a higher number does not affect the totals in any way. That seems strange, since a dividing by a higher number would inherently change the totals. The reason that I noticed this is because in Story County Precinct 1-1 dividing by the lower number of actual voters gave Sanders an extra delegate. But now that I have been introduced to the .05 rule - I understand that he would have lost that extra delegate in the next computation. It's not obvious what would happen in each of these permutations.

This simply isn't any way to run an election. I shouldn't need to stay up all night trying to figure out how some ridiculously archaic rule works - and have thousands of people debating whether it's inaccurate or just obscure. Elections need to be straight-forward and easy for people to follow. Not some sort of byzantine calculation that only the initiated high priests of caucus history are able to navigate. 

What is great about the Iowa Caucus is that all the votes are hand written on durable paper, so we can trust that they represent the voter's intent and ultimately we can observe - and hopefully - understand how they have been counted. People who are watching the results have been able to correct them when they've seen errors. That's great - that's the accuracy, transparency and public oversight that we need in elections. We need durable hand-marked paper ballots in all elections for everyone who is able to use them. And we need secure devices that create durable and easily understood paper ballots for voters with disabilities. Not thermal paper ballots that are read with barcodes or QR codes.

This year the Iowa caucus rules were changed so that we can all see what's going on behind the scenes with the math. That's good. Next year, maybe we could have one vote per person, and the whole election could be run by a non-partisan election commission the way it is in Australia, instead of a party that has been shown repeatedly to have a bias. 

In an email , Professor Pippa Norris of Harvard agreed that political parties running U.S. Elections is a “pure conflict of interest.” Adding that, “In established democracies there are normally ... election commissions,” with, “neutral and impartial professional civil servants.” She cites the UK, Norway and Germany as examples. According to their website, the Australian Electoral Commission, “places special emphasis on political neutrality” to provide Australians with “an independent electoral service.”    

SMART Elections has an ongoing initiative called #CountTheVote that was launched in an op-ed at The Hill. We urge you to get involved and help us monitor and assess the security, accuracy, accessibility and transparency of the vote. This is just the beginning.

Regarding the earlier designation of this math as "rounding errors:" 

In our original post we made it clear that we were trying to understand what was going on with these totals. We reached out to the Iowa Democratic Party, leaving voice mails that were not returned and sending emails, which bounced. We looked on the website for the manual, which it turned out was not under the tab "caucus guide" - the name it is given on the worksheets, but instead called the "Precinct Leader Manual." We even had one of our technical advisors scrape the web and try to find it. Finally, we asked for and received clarification on Twitter. Now that the manual has been produced, and the explanation for the rounding is clear - it serves to raise more questions than answers about the validity of this process.

Note: this post does not seem to load photos on the Safari browser. We will try to get it resolved. In the meantime, please try looking at the post in Firefox or Chrome. Thank you. 

Post by Lulu Friesdat - Co-Founder of SMART Elections. The unusual calculations were first noticed by Teresa Basey, a member of the #CountTheVote team, who put them into a spread sheet.