Complicated voting machines!

Alternate title: My first experience of voting on a Direct Recording Electronic or DRE device.

On Nov 7, 2006, upon arrival at the polling place with completed sample ballot in hand, unfamiliar voting machines were visible from the doorway. I asked the poll workers if paper ballots were available. They were! They asked me if I would like one. No, I decided, I wanted to experience the computerized system first hand.

Most elections I’ve participated in over the years have used the punch card machine, the one made infamous by the hanging chads of the presidential contest of 2000 between Bush and Gore. Since that time, our district of San Diego County has sometimes used a paper ballot and pen that was optically scanned as a last step before the voter left the polling location. This time, and for the first time, our district used DRE machines made by Diebold.

I don’t intend to vote using one again.

The process of entering one’s sample ballot selections into the machine was easy and straightforward enough, but auditing ones results was a study of frustration. While I didn’t think to measure the total time used to vote, I estimate it took at least four times longer than any other method I’ve used in the past, and most of this extra time is spent to insure the machine has recorded and will cast the vote as intended.

Double checking never seemed an issue with the punch card machines after the pin had been pressed into the correct hole, though evidently chads not fully punched are an issue. The other system, a paper ballot and pen, can be doubled checked when it’s initially filled out while the sample ballot is open to the contest page in question.

Here are some things I noticed as problems when using the Diebold DRE with a voter verified paper trail:

The initial vote screens seemed to match up with the pages of the sample ballot until about page six, where one contest from the sample ballot appeared on another page of the DRE device. This problem seemed somewhat minor, but it does seem somewhat confusing and perhaps slightly time consuming.

Once all votes had been cast, the computer built a summary page view. This was arranged into three columns. The voter must go back to page one of the sample ballot unless they have all the contest choices memorized. Another flaw is that all candidates’ full names were not always visible in the summary view, in some cases only their first names were presented. One must scroll downward to complete column one’s sequence, then one must scroll up to find the top of column two, followed by scroll downs to complete column two, then a scroll up to the top of column three, followed by scroll downs to complete the summary audit. The arrangement didn’t match the sample ballot at all, and it seems fewer scrolls could have easily been designed by sequencing the contest data in rows instead of columns.

One time I accidentally touched the summary screen while using my fingers to help track a particular line while my eyes darted between the screen and sample ballot. The machine decided to take me back to the initial vote casting screen that correlated to that contest, presumably so I could change it. I had to press the summary button again, and the machine’s screen said something to the effect of “building summary file”. Whoops.

I asked a poll worker if that meant I had to go back and double check the answers I’d already checked, and he said no (all the poll workers were different this year, and predominately of the younger, computer-literate generation) This advice I promptly judged as naive or computer illiterate at best, and likely coached at worst. The computer was clear, it had built a new summary file, so I needed to start back at the beginning of column one of the summary view and page back in my sample ballot again, unless I was willing to gamble that the computer hadn’t made any errors.

I can only imagine how much time voting might have taken had there been errors that needed correcting. The summary view was user unfriendly. There was something wrong with the scrolling, so after the device had completed the scroll, finding the next line to focus upon was an issue.

When the summary self-audit was completed, there was a paper printout that I could not touch but could see through a small glass window. This printout appeared to be thermal paper with dark print. I can’t tell you how stupid I think that decision was, if the paper was in fact thermal paper, given that the voter-verified printout is intended as a permanent record of each voter’s intent. There have been years when I’ve had receipts printed on thermal paper that I’ve had to discard from my end-of-year accounting because I could not read them, as time and or heat had rendered the print unreadable on darkened paper. All some election fraud perpetrator would need to do is leave the paper ballots in a warm enough location, such as a closed vehicle parked in the sun light on a warm day, or perhaps stored inside a non-climate controlled facility, and the thermal paper record’s usefulness is likely destroyed.

The printout was much shorter than the summary screen, two lines for each contest. Many of the lines were self-explanatory, unfortunately, the judicial section was not. Our election had about fifteen choices for judicial positions. On the paper printout, every line that titled twelve of those contests appeared precisely the same, they all read “Associate justice court of AP…”, that line was followed by another line indicating the yes or no vote: No candidate name appeared in any of this judicial sequence. The only way I was able to discern what candidate those lines correlated to was to count the lines and assume the ordering was the same as that presented in the sample ballot. Great, I need to assume something for a self-audit? To make matters worse, all fifteen contests did not appear at once under the limited glass size. The first presented readout needed to be accepted somewhere in the middle of the fifteen choices, afterward the next page printed and scrolled. The scroll itself caused issues with this counting to determine the contest, several of the bottom lines of the previous printout remained at the top of the glass screen. If I had not been watching very closely during the scroll itself, this would have caused me to get lost in the counted sequence that I correlated to the sample ballot ordering.

Is this judicial obfuscation deliberate? How can this possibly provide a paper trail of the voter’s intent if the name of the candidate isn’t printed on the paper trail?

The only good thing I can say is I did not experience any problems with the machine changing my vote, or making the voting for any of the candidates I intended to vote for hard to select, so far as I could tell, and as has been reported in the news at times.

In past years, once contest choices had been decided and the sample ballot marked, I remember spending no more than 2-5 minutes zipping through all the choices on the ballot at the polling place, though standing in line and waiting for a voting booth added to that time. I conservatively estimate that I spent at least 20 minutes actually voting on this DRE device. Since the process of voting on these machines takes about four times longer due to all the self-auditing and matching up of different presentational formats (that I truly felt needed to be checked), it’s no surprise that some voting districts that use these machines are reported as having long lines. The reported move by some other districts towards voting super centers in favor of the familiar neighborhood polling places will only exacerbate this problem. The mixing of formats in the summary view and separate printout view lead to a definite lack-of-clarity: another term for that could be obfuscation, certainly the failure to provide any of the candidates’ names on the judicial section of the paper trail is some type of obfuscation.

The process of filling out the ballot has been rendered much more difficult and complicated with this particular machine than it used to be with other ballot types.

Prior to this experience, I was open to the idea of electronic voting on a DRE machine so long as it had an unalterable paper record of the vote cast, this is seen in a willingness to try the system once our polling place offered it. Now that I’ve seen and experienced its implementation first hand, the thermal alterable paper that also fails to show my complete intent, the formatting confusion between the initial vote screens versus summary view versus the paper printout, even the polling place worker who claimed that I didn’t need to re-check when the machine built a new summary file, never again will I choose to vote with such a machine.

Of all the types of voting systems I’ve experienced, paper and pen seems the clearest, simplest, and fastest. Why make voting apparatus more complicated than it needs to be?

Randy Wooten, a mayoral candidate in Arkansas found out that his vote for himself wasn’t counted, official results showed him with a vote count of zero. Reports seem to indicate the voting machine was not from Diebold.

I have never once observed or experienced the actual counting of the votes. Counting the votes openly seems even more important than the type of voting system used, and how the votes are counted in machines using proprietary code owned by private corporations can never be known by the majority of citizens.

It seems to me that we need not only paper ballots and pens, but local hand counting by citizen volunteers at the polling locations and observable by concerned voters, with results posted immediately, publicly, and openly.

Until then we can kiss our democratic republic farewell. I was wrong to ever believe that computerized voting could be a good thing.

The punch card systems were counted by computers, except when they needed close examination. Paper ballots marked by pens and followed by optical scanning are essentially computers counting the vote, but once again the ballots might be counted by hand if there’s suspicion surrounding the results. DRE and their complicated system is perhaps the epitome of computerized vote aggregating, if equipped with one, their paper trail also can be recounted if necessary, but then there’s the little problem I noted above of the paper trail not always including all candidates’ names, and a potential temporal record issue if thermal paper was used. Lawyers seeking access to these paper records for purposes of recount have reportedly been experiencing cooperation difficulties.

For my entire life computers have been involved in counting votes. The computers apparently run proprietary, private code under the control of a few corporations, therefore it is not surprising that over the longer term such a ‘democracy’ has found itself living as a corporatist system.

One thought on “Complicated voting machines!

  1. The voting process in any country *must* be transparent. That should have been a spec of the system when bids went out. In fact, if it were, I’m sure we wouldn’t have seen these types of problems in the first place.

Comments are closed.