My Day At The Polls — Baltimore, Maryland Primary 2006 by Avi Rubin

© 2006 Avi Rubin's Blog

Reproduced under the Fair Use exception of 17 USC § 107 for noncommercial, nonprofit, and educational use.


| EJF Home | Where To Find Help | Join the EJF | Comments? | Get EJF newsletter |


| Vote Fraud and Election Issues Book | Table of Contents | Site Map | Index |


| Chapter 12 — Voting Problems In The 2006 Elections |

| Next — Various News Reports — 2006 General Elections |

| Back — Various News Reports — 2006 Primary Elections |


Tuesday, September 12, 2006 — I don't know where to start. This primary today is the third election that I have worked as an election judge. The last two elections were in 2004, and I was in a small precinct in Timonium, MD. This time, I was in my home precinct about 1/2 a mile from my house. We had 12 machines, over 1,000 voters and 16 judges. I woke up at 5:30 in the morning and was at the precinct before 6:00. It is now 10:18 pm, and I just got home a few minutes ago. As I have made it my custom, I sat down right away to write about my experience while everything was still fresh. In anticipation of this, I took some careful notes throughout the day.

The biggest change over the 2004 election was the introduction of electronic poll books that we used to check in voters. I was introduced to these in election judge training a few weeks ago. These are basically little touchscreen computers that are connected to an Ethernet hub. They each contain a full database of the registered voters in the county, and information about whether or not each voter has already voted, in addition to all of the voter registration information. The system is designed so that the machines constantly sync with each other so that if a voter signs in on one of them and then goes to another one, that voter will already be flagged as having voted. That was the theory anyway. These poll books turned out to be a disaster, but more on that later.

Around 7:15 AM, when we had been open for business for 15 minutes already, a gentlemen shows up saying that he is a judge from another precinct nearby and that they did not receive any smartcards, so that they could not operate their election. We had 60 smartcards, and the chief judge suggested that we give them 20 so that they could at least get their election started. As she was handing them over, I suggested that we had to somehow verify his claim. After all, anyone could walk in off the street and claim this guy's story, and we would give them 20 access cards. The chief judge agreed with me. The guy pulled out his driver's license to prove who he was, but I told him that we were not doubting who he was, we just wanted to verify that we should give him the cards. He seemed to understand that. After calling the board of elections, we were told to give him the cards and we did. A little later, several voters who came in informed us that news reports were saying that in Montgomery county, there was a widespread problem of missing smartcards. I could only imagine what a nightmare that was for those poll workers because as it was, our precinct did not have this problem, and as you'll see, it was still tough going.

My precinct uses Diebold AccuVote TS, the same one that we analyzed in our study 3 years ago. The first problem we encountered was that two of the voting machine's security tag numbers did not match our records. After a call to the board of elections, we were told to set those aside and not use them. So, we were down to 10. We set up those machines in a daisy chain fashion, as described in the judge manual, and as we learned in our training. We plugged the first one into the wall and taped the wire to the floor with electric tape so nobody would trip over it. About two hours into the voting, I noticed that the little power readout on the machines was red, and I thought that this meant that the machines were on battery power. I pointed this out to one of the chief judges, but she said this was normal. An hour later, I checked again, and this time, the machines were on extremely low power. This time, I took the plug out to of the wall and tried another outlet nearby. The power icon turned green. I showed several of the judges, and we confirmed that the original outlet was indeed dead. Had I not checked this twice, those machines would have died in the middle of the election, most likely in the middle of people voting. I hate to think about how we would have handled that. A couple of hours later, the board of elections informed us that we should use the two voting machines with the mismatched tags, so we added them and used them the rest of the day (!).

When we were setting up the electronic poll books, I took over because I was more comfortable with the technology, and the others quickly deferred to me. So, a couple of hours into the election, when one of the poll books seemed to be out of sync with the others, the judges came and brought me to have a look. It appeared that this poll book was not getting synced with the others. I tested it by waiting for someone to sign in with a different poll book, and then a few minutes later trying to sign in that voter on the one in question. The voter was shown as having not voted yet. I repeated this test for about 20 minutes, but it never registered that voter as having voted, and the poll book was falling behind — about 30 by then — the other poll book machines. I suggested rebooting that machine, and we tried that, but it did not change anything. I pointed out to the chief judges who were huddled around me as I experimented, that as time went by, this poll book was going to fall further and further behind the others, and that if someone signed in on the others, they would be able sign in again on this one and vote again. After a call to the board of elections, we decided to take this one out of commission. This was very unfortunate, because our waiting lines were starting to get very long, and the check-in was the bottleneck. The last few hours of the day, we had a 45 minute to an hour wait, and we had enough machines in service to handle the load, but it was taking people too long to sign in.

The electronic poll books presented an even bigger problem, however. Every so often, about once every 15-25 minutes, after a voter signed in, and while that voter's smartcard was being programmed with the ballot, the poll book would suddenly crash and reboot. Unfortunately, the smartcard would not be programmed at the end of this, so the poll worker would have to try again. However, the second time, the machine said that the voter had already voted. The first few times this happened, we had some very irate voters, and we had to call over the chief judge. Soon, however, we realized what was happening, and as soon as the poll book crashed, we warned the voter that it would come up saying that they had already voted, but that we knew they hadn't. Then, the chief judge would have to come over, enter a password, and authorize that person to vote anyway. Then we had to make a log entry of the event and quarantine the offending smartcard. Unfortunately, the poll books take about 3 minutes to reboot, and the chief judges are very scarce resources, so this caused further delays and caused the long line we had for most of the afternoon and evening while many of the machines were idle. Another problem was that the poll book would not subtract a voter from its total count when this happened, so every time we had an incident, the poll book voter count was further off the mark. We had to keep track of this by hand, so we could reconcile it at the end of the day.

At times, the remaining two poll books were way out of synch, but after a while, they caught up with each other. When the lines got really long, we considered the idea of trying to use the third one that had caused problems, but we all agreed that we would feel very stupid if all of them started crashing more. I was worried that synching three of these on an Ethernet hub was more complex than 2, and in fact, they were crashing a bit less often when we had only 2. The whole time I was worried about what we would do if these thing really died or crashed so badly and so often that we couldn't really use them. We had no backup voter cards, so the best we could have done would have been to start letting everybody vote by provisional ballots. However, we had two small pads of those ballots, and we would have run out quickly. I can't imagine basing the success of an election on something so fragile as these terrible, buggy machines.

Throughout the early part of the day, there was a Diebold representative at our precinct. When I was setting up the poll books, he came over to "help", and I ended up explaining to him why I had to hook the ethernet cables into a hub instead of directly into all the machines (not to mention the fact that there were not enough ports on the machines to do it that way). The next few times we had problems, the judges would call him over, and then he called me over to help. After a while, I asked him how long he had been working for Diebold because he didn't seem to know anything about the equipment, and he said, "one day." I said, "You mean they hired you yesterday?" And he replied, "yes, I had 6 hours of training yesterday. It was 80 people and 2 instructors, and none of us really knew what was going on." I asked him how this was possible, and he replied, "I shouldn't be telling you this, but it's all money. They are too cheap to do this right. They should have a real tech person in each precinct, but that costs too much, so they go out and hire a bunch of contractors the day before the election, and they think that they can train us, but it's too compressed." Around 4 pm, he came and told me that he wasn't doing any good there, and that he was too frustrated, and that he was going home. We didn't see him again.

I haven't written at all about the AccuVote machines. I guess I've made my opinions about that known in the past, and my new book deals primarily with them. Nothing happened today to change my opinion about the security of these systems, but I did have some eye opening experiences about the weaknesses of some of the physical security measures that are touted as providing the missing security. For example, I carefully studied the tamper tape that is used to guard the memory cards. In light of Hursti's report, the security of the memory cards is critical. Well, I am 100% convinced that if the tamper tape had been peeled off and put back on, nobody except a very well trained professional would notice it. The tamper tape has a tiny version of the word "void" appear inside it after it has been removed and replaced, but it is very subtle. In fact, a couple of times, due to issues we had with the machines, the chief judge removed the tamper tape and then put it back. One time, it was to reboot a machine that was hanging when a voter was trying to vote. I looked at the tamper tape that was replaced and couldn't tell the difference, and then it occurred to me that instead of rebooting, someone could mess with the memory card and replace the tape, and we wouldn't have noticed. I asked if I could play with the tamper tape a bit, and they let me handle it. I believe I can now, with great effort and concentration, tell the difference between one that has been peeled off and one that has not. But, I did not see the judges using that kind of care every time they opened and closed them. As far as I'm concerned, the tamper tape does very little in the way of actual security, and that will be the case as long as it is used by lay poll workers, as opposed to CIA agents.

As we were computing the final tallies towards the end of the evening, one of the Diebold machines froze. We had not yet printed the report that is used to post the results. One of the judges went to call the board of elections. She said she was transferred and then disconnected. We decided to do a hard reboot of it after we closed down the other machines. When we finished the other machines, we noticed that the problem one had somehow recovered, and we were able to finish. Strange because it was frozen for about 10 minutes.

So, this day at the polls was different from my two experiences in 2004. I felt more like an experienced veteran than a wide-eyed newbie. The novelty that I felt in 2002 was gone, and I felt seasoned. Even the chief judges often came to me asking advice on how to handle various crises that arose. Several other suggested that I should apply to be a chief judge in the next election cycle, and I will probably do that. The least pleasant part of the day was a nagging concern that something would go terribly wrong, and that we would have no way to recover. I believe that fully electronic systems, such as the precinct we had today, are too fragile. The smallest thing can lead to a disaster. We had a long line of "customers" who were mostly patient, but somewhat irritated, and I felt like we were not always in a position to offer them decent customer service. When our poll books crashed, and the lines grew, I had a sense of dread that we might end up finishing the day without a completed election. As an election judge I put aside my personal beliefs that these machines are easy to rig in an undetectable way, and become more worried that the election process would completely fail. I don't think it would have taken much for that to have happened.

One other thing struck me. In 2004, most voters seemed happy with the machines. This time around, many of them complained about a lack of a paper trail. Some of them clearly knew who I was and my position on this, but others clearly did not. I did not hear one voter say they were happy with the machines, and a dozen or so expressed strong feelings against them.

I am way too tired now (it's past 11 PM) to write any kind of philosophical ending to this already too long blog entry. I hope that we got it right in my precinct, but I know that there is no way to know for sure. We cannot do recounts. Finally, I have to say a few words about my fellow poll workers. We all worked from 6 AM to past 10 PM. These volunteers were cheerful, pleasant, and diligent. They were there to serve the public, and they acted like it. I greatly admire them, and while the election technology selection and testing processes in this country make me sick, I take great hope and inspiration from a day in the trenches with these people.

Posted by Avi Rubin: at 11:18 PM




Ping said...

Thank you so much for writing this up. It's eye-opening, alarming, and extremely informative.

9/13/2006 12:12 AM

Joseph Lorenzo Hall said...

I wish we had 1,000 or so of these stories from many poll workers in Baltimore... or better yet, across the Country.

Avi, we will make things better.

9/13/2006 12:21 AM

Anonymous said...

What I found the most enlightening in your write up is that:

1) the cost of electronic voting in terms of "human services" is obviously just as large if not larger than "hand voting," so there is NO economical justification for this.

2) having two machines that are not "certified" (not the right serial number) in the lot, and being told to use them should invalidate the vote, even if in this particular case there was no tampering, accepting this makes it too easy to tamper another time (what would be the difference?)

3) having voting both where the voting gear is not ready (no or not enough smartcards) should also be a reason to stop this election.

4) many dedicated people you included are spending a lot of time and effort to "make this work," and then "validate" this vote when they actually KNOW it can not be trusted.

So obviously there is no democracy in the US anymore, since even the most "democratic" (not necessarily as in "democrats" ) citizen do not have it in themselves to "stand up" and refuse to validate this "show."

What should have happened is that all voting booth using diebold products should just have refused to use them and ask for postponement and the usage of paper ballot.

Incidentally this also show the "microsoft" effect, we are so used to have clunky buggy crashing software and "no choice" that we make all the effort to "use it anyway."

Its a sad world we leave to our children, and an insult to the fights of our parents.

9/13/2006 5:18 AM

Beth said...

Diebold must be held accountable for hiring people who know nothing about the machines and it appears, elections. Meltdown, train wreck, whatever you want to call it, it is here.

We have been sold the emperor's clothes and now we are going to pay the price with widespread disenfranchisement.

9/13/2006 8:59 AM

Steve Tate said...

I checked in this morning because I wanted to see what your experiences were — excellent write-up, especially being at the end of what must have been a very long day.

I am fairly confused about these devices that were used to check in voters. You said they were hooked together through a hub, but were they hooked up to the outside to sync to other precincts? That might be an understandable goal, but if it's just within a single precinct then it looks like a "solution" to a non-problem. Like most places (I guess) we use paper books with voter's names, which everyone signs. Only one slot, so they can only sign once. Load balancing is easy (if imperfect) because there are different books for sets of last names, and people get in the right line. Best of all, it's pretty technically fool-proof. Of course, the books have to be printed, so I suppose the voter lists can't be updated in the few days before the election, but is that really a problem? This just seems like another example of people throwing new technology at something that works fine as it is, and screwing it up in the process...

9/13/2006 9:17 AM

granny6x said...

I worked at the polls for the first time this Primary election. It turned out to be a nineteen hour day. I served as Assistant Chief and was responsible for the provisional ballots. My training was minimal, and the woman training me for the 45 minutes was visibly exhausted. My training took place the day before the election. I was told to help set up the voting machines at the election precinct that night. Only three of the six election judges showed up for that. Ok we did do that by reading the manuals together. But it was tiring and we knew we had to be back at 5:30 the next morning. At that time we could finally open our individual packages of materials and instructions.

But getting the machines themselves set up took all of us, we were very short again of judges, and I did not get to study my job materials until much later after the polls were open.

We had a very, very small turnout and this saved us I think from all the frustrations at other precincts. With a serious turnout it might have been different.

The entire system is so sloppy and risky. That was very apparent to all of us.

But my particular precinct did not experience voters being turned away. The poll books did have a problem in that cards were crashing it, until someone realized that we were supposed to be cleaning them regularly, and after we did that it was much better.

We lost one touchscreen, but that was because the elderly gentleman who was setting it up in the early am was using the instructions for closing it down, so he did. We did not use that machine until the end of the night when the Chief Judge attempted to tally up the vote and it froze. A call to the Board resulted in the instruction for her to just pack up the machine as is.

What probably saved us from irate voters was the miserable turnout.

Frankly I did not miss an opportunity to talk up paper ballots, and to a man each voter that used my provisionals left knowing that their vote was recorded by THEM, not by a mysterious black box.

I wish I had something useful to offer. The fact is I have never thought glitches should be the focus. OK we get it. The organization of the process is terrible. But the very nature of voting on touchscreens is what I feel we should hammer home. They are not to be trusted, nor is the privatization of our vote.

Even if it went "smoothly."

9/13/2006 10:45 AM

Anonymous said...

Dr. Rubin,

Thank You. I cannot express how fortunate you have made us feel in Connecticut.

Here we have a Secretary of State who has insulated us from the experiences you recount so well. She has done this, so she would have the State believe, by entering into a $15.7 million contract with a New England based firm, LHS Associates, to provide AccuVote OS Op Scan units to replace — belatedly — our venerable Lever Machines.

This creative solution to a problem with the reality you describe now allows the Connecticut voter to be assured that he/she does NOT have to be concerned with the hazards of using a Diebold Electronic Voting Machine.

Yes, the Connecticut SOS is a candidate on this November's ballot.

From the land of the "Wooden Nutmeg"

9/13/2006 10:57 AM

Richard Welty said...

I work with a portable scales that have built in batteries (gel cells) not unlike the cells in the voting machines, and have some comments on how I handle battery powered equipment.

First, these cells have a lifespan, typically 4-5 years. This lifespan may end up shortening a bit if they aren't regularly run down and then recharged from nearly dead. What I do when I have good built in batteries is run on battery, but have an external battery backup fully charged and plugged in, and if the low battery indication comes up, plug the battery powered equipment into the external backup to recharge. The battery backups are present in the chain in case of a power outage or in case someone trips over a cord (big win is that the backup units start beeping if the cord gets pulled out; otherwise you might not notice until a battery runs down).

The cells built into electronic voting machines obviously shouldn't be allowed to reach the point of being completely unusable (that is, won't take a charge/work for a reasonable span of hours), but I assume that in some places, this will happen. In these cases, you want to have external battery backups around and run them with the machines plugged in.

9/13/2006 10:58 AM

Dave Bell said...

One thing I don't get a feel for is how complicated this event was. How many different races?

Some things seem so obvious, and so easy to check. It is almost beyond belief that nobody seemed to know what the power indicators meant, and there was no checklist. Were spare power cords available?

But that's one of the small problems, easily fixed if it can be reported through channels and acted upon. Will that happen?

9/13/2006 11:24 AM

LiebermanForLieberman said...

Isn't it interesting that Diebold ATMs seem to be as solid as a rock, while their voting systems are a complete joke?

Obviously, Diebold is capable of making a competent electronic voting system. They have "elected" not to.

9/13/2006 11:30 AM

Ben said...

When I voted yesterday evening in Montgomery County, I requested to vote by provisional ballot, so I could ensure that my vote could be counted if a recount were necessary. My request was denied, and I told the election judge I did not trust the machines to correctly cast my vote. She laughed, and then said "I can't believe you wouldn't trust these machines," as other judges and election officials joined in. There was a good 2 minutes of poll workers joking about how reliable the machines were, how they hadn't opened until 10:20 that morning because the machines were so reliable, and of course "no, we still can't let you vote by any other means except this machine." Next time, I'll vote absentee.

9/13/2006 11:32 AM

Anonymous said...

I voted in Maryland yesterday. On the screen at the end where you can verify your votes, there was a block that said "vote for no more than." This legend took up the whole space allotted, and the number that was supposed to follow was nowhere. (For a different office it was pushed to another line, but for this one it was lost.)

This is, of course, a programming error of the most banal kind, one that all of us programmers have no doubt made sometime. It is hard to see how it could matter much.

On the other hand, I have trouble imagining a quality-control system that would provide the kind of assurance we have been hearing that the hard problems are really under control, at the same time it lets stuff like this pass.

Tom Permutt

9/13/2006 11:57 AM

mattplygop said...

I don't understand what goes on in the rest of the country! In the town I live in CT we have two precincts with about 3000 registered voters each. There are 5 machines at each location and 1 judge. Never in my almost 20 years of voting has there been a problem. In my 10 years of being involved in local party politics has one party claimed the other has cheated, which appears to be happening with all too much frequency around the country. The solutions are simple:

1. Go back to the old mechanical machines. They work. Chips don't crash, no re-booting, and votes are recorded. They are simple to use. What's not to like?

2. Stop listening to every loser that complains about voter fraud. I can't believe it is that rampant. The registrar at the polls is my neighbor and I still have to show my ID. Am I to believe that other checkers are so stupid they let anyone without an ID vote?

It really is simple. We just need to stop having simple people make decisions.

9/13/2006 12:24 PM

Joe in Wynnewood, PA said...

I think it'd be great if you turned this post into an op-ed piece and submitted to MD papers and perhaps a monthly or 2 as well.

Also, you're comment about the heroic efforts of poll workers brings the 5 level Capability Maturity Model (CMM) to mind. As I'm sure you know, most IT shops are usually in the 0.5 to 1.5 range where there is a dearth of effective processes and procedures that ensure effective software development and testing and the routine production of quality software. I think it would be safe to say that Diebold would fare rather badly in a CMM audit and the voting process utilizing their products would fare no better.

9/13/2006 12:26 PM

James said...

Thanks for the information. I am a Hopkins student and I was working outside of the polling location near Homewood for a few hours, and heard about all of the problems through osmosis.

I had read your work in the past about these machines, which is why it felt kind of futile standing there "doing my part" when I knew it could easily have been rigged in the first place.

I'm registered in another state to vote absentee, but I still don't think that's an adequate solution, because who knows if my vote will even be counted anyway, or if they'll just rely on the huge numbers of these "reliable" electronic votes.

Anyway, it makes me feel good to see someone doing something about this. I'd love to take a class with you at some point, but I haven't noticed any on the course list yet for undergraduates. Hopefully I get the chance.


James Freedman bulk AT

9/13/2006 12:43 PM

Anonymous said...

I live in Cuyahoga county OH and am fairly computer literate. I vote at a school nearby and go early, about an hour after polls open. I noticed many new poll workers' faces, as usual oldsters who are statistically not computer users. The Diebold machines had arrived only two weeks previously, and none of the workers had actually seen them before 6:00 that morning. The machines were set up so their ports were against the walls of the gym. A lot of effort would have to be made to tamper with them. I looked for IR ports, just in case and saw none. I spent the next several hours showing the poll workers and some voters how to operate the machines, especially how to work the printer and check the printout. Later, I there were reports of up to 70 missing memory cards, later reduced to 12, negating hundreds of voters' intentions. About 20% of poll workers never showed, perhaps terrified by having to use computers, a common oldster phobia. BTW, I'm 65. The weather was not a factor. Our bipartisan election board split on canning the elections chief and his asst. Whether the votes were reliably counted is still a mystery.


9/13/2006 1:24 PM

Anonymous said...

ATM machines and voting machines are not exactly analogous. The key is that ballots are secret and your checking account balance is not. You and your bank can both see your statement and figure out what the balance is. Unless you want the government to know both your name and how you voted, it's hard to get the same degree of transparency.

9/13/2006 1:25 PM

DrBB said...

LiebermanforLieberman gets at what bugs me about this whole Diebold debacle. I'm looking at the human effects in your account. Not only endless headaches for the poll workers, but long lines and breakdowns — equipment freezing, rebooting, etc. — for the voters.

You build glitchy, kludgy, shoddy equipment, then maybe it mostly works or maybe it doesn't but you ceaselessly train voters that a visit to the polling place is an ordeal and you drive home the implicit but unmistakable message that the people who built the infrastructure didn't think it mattered whether they did a decent — or even halfway reasonable — job at it.

If a voter came away thinking the whole thing was a cheap charade put on by people too lazy to even fake it very well, he or she would be entirely justified.

They could certainly build better equipment than this if they wanted to. Maybe their motivations aren't suspect — maybe — and maybe they didn't deliberately create a system that would be easy to hack. Maybe. But they certainly are aiding and abetting the overall downward trend in voter participation. Who benefits from that trend?

9/13/2006 1:30 PM

michaelweaselo said...

I worked as a check in judge in Carroll County, MD yesterday. I noticed all the same problems as Mr. Rubin. In fact, the problems were just as bad in my precinct. We had 2 chief judges who had never worked an election before. They were confused and stressed. We had the machine crashes checking in voters. During a lull in the voting, from about 2 until 5 PM, I even looked through all the screens on the voting poll book computer. There was a stats page. One tap on the screen and a second stats page came up. It listed the total number of voters issued cards broken down into 4 categories. One of which was listed as "cancelled cards." The number listed was 72. Now, I had no idea what the cancelled cards category meant, but there was another category called "cards cast." No one could explain to me the semantic difference between a card cast and a cancelled card. The Diebold tech (who was very good, extremely helpful, and also a one day hire like the others) didn't know either and called his boss to find out. The boss told him to tell me to shut up and mind my own business as I had no authority to look at those numbers. Well, since it was my job to use the machine and verify that "those numbers" were zeroed out before voting took place in the morning, I think it was my job to look at them.

The Diebold people were rude and awful.

We even had election judges not following the manuals. I was forced to tell one voter who was listed as a Democrat but wanted to vote Republican that she had no choice but to vote Democrat. Clearly, the rules in my manual stated that she could be given a provisional ballot. They refused it. But they did give her a form to change her political party affiliation for the next election.

The machines did crash and list voters as having already voted. The cards were supposedly still valid and we had them fill out a paper instead of having it print out to sign. Apparently the voters didn't have any problems voting with the cards after the machines crashed. I did see them use their passwords and issue additional voter access cards to people. But mostly, we just let the machine reboot and start over.

I noticed the crashes happened more often with certain voter access cards. When we stopped using the ones prone to crashing, we had better success.

If it had been a general election, there would have been many upset voters.

9/13/2006 2:07 PM

Stan Klein said...

I wonder if the Diebold technician had been vetted to be sure he did not have a criminal record. I doubt it, given the speed with which he was hired and trained. Convicted felons can't be hired to carry bedpans in Maryland nursing homes or to work with children. How do we know that Diebold isn't hiring "technicians" that work cheap because someone else is paying them as part of a scheme to do insider tampering and steal an election?

This provides an indicator of the importance placed on the integrity of the electoral process by which the "consent of the governed" is determined.

9/13/2006 2:10 PM

Jim Jones said...

Thanks for the post Dr. Rubin. I have been following your experiences since your original report.

In fact, you have inspired me and my wife to try and work in the polls. We figure that the system may be flawed, but that people that are comfortable with technology, as you have shown, are probably the only way to make it lurch towards a completed election.

Thanks so much. Keep up the good work. You are making a difference in this world.

9/13/2006 3:18 PM

Michael G. Thompson said...

Dr. Rubin:

I was also an election judge cum computer geek in this election, at 08-13, Timonium Elementary in Baltimore County. You said you were at a Timonium site in the past, was it this one? I agree with your assessment of the electronic poll books and the vulnerabilities in the AccuVote TS system. We too had the poll books crash repeatedly and have sync problems. In my estimation, the problem may have been linked to the pollbook's sensitivity to transient voltages. Power demand is at it's highest when encoding a VAC, I suspect, and there is no power conditioning of any sort for the AccuVote units, or the poll books. We've suggested that they supply us with surge suppressors (or UPS with surge suppression) for the general. Please keep up the good works, sir. Electronic voting will be safer and surer for it.

9/13/2006 3:29 PM

zappymax said...


Those first hand comments about the reality of electronic registration or transmission (to which place?) of the vote information emitted in polling places are interesting.

using internet links to a command center somewhere (where?) induces the implementation of any kind of subtle, even secret feedback from those receivers, so to control, eventually, part of the way those are working, freezing, registering, transmitting, hubbing, bypassing, forgetting, transforming, a.s.o, the results of the vote process.

Who has access to those machine software, and the way those might transfer information, and to where officially — and unofficially...

remembering (the) Florida, (or) Ohio ballots in the US, and more recently difficult Mexico voting sessions (even if presidential elections in Mexico are not yet electronically performed, just electronically digested at "some" upper level — but where we still don't know exactly, here), I wonder if electronic machines will not kill democratic process by subtle or brutal manipulation of which nobody really knows — or will reveal — the origins.

Progress in voting? machines for voting "registration" are just an easy way to blur and adapt the voting results to some discreet interests. including the diebold or their likes voting machines producers getting public money or funds to bypass the voting process in accordance to executable forces they even finance to stay elected.

9/13/2006 4:52 PM

Anonymous said...

I live in montgomery county and there was either terrible turnout or my wife and I got lucky and arrived at our voting place during a lull, so we didn't have any waits. However, it was obvious to both of us that the check-in process would produce delays if a large crowd showed up.

Everything went very smoothly and as I exited, I made a comment to one of the election officials about "still no paper trail, huh" and a comment that the diebold machines don't have a very good reputation. In response to my paper trail comment, the official said no, there was a paper tabulation printed internally for each machine. He pointed in the direction of the exit and said the printouts showing each machine started with a zero vote count were on the wall. Heading towards the exit, I find that the printouts for each machine (with original ink signatures) are taped up on the OUTSIDE of the entrance. I wondered what inconvenience would be caused if someone grabbed a handful of these and walked off with them. Considering these printouts were proof the machines started from a known, zero count state, and had the original ink signatures from the officials on site, it seemed pretty stupid to tape these up outside the building.

9/13/2006 5:20 PM

Andy Miller in St. Paul, MN said...

Thank you for your terrifying account of how the process worked(?) in the presence of a technically savvy election judge with a deep understanding of electronic voting issues.

I am worried though that your posting obscures the motivation behind the drive to electronic voting: why are we (the people) trying to make electronic voting work at all? What exactly is wrong with paper ballots? And what do we have to do to get our locals to punt on the whole electronic thing and simply do what works?

9/13/2006 6:08 PM

michaelweaselo said...

They print two copies of those tapes from the voting machines. One gets taped up publicly and the other is saved and brought back to the election board after the election.

9/13/2006 7:30 PM

MyStro said...

I totally agree. I'm in Australia, and paper and ink still work the same as ever, even for elections.

Comments about requests for provisional ballots being turned down are truly scary, as is the story of the voter not being given a choice as to whom she voted for! Isn't that what elections are all about?

9/13/2006 7:45 PM

Anonymous said...

Scary, as is the story of the voter not being given a choice as to whom she voted for! Isn't that what elections are all about?

It's not quite as bad as that. It's a primary election — to choose the party candidates for the general election. To vote in it, you have to be affiliated with a party. Some jurisdictions allow that affiliation to be made or changed at the polling station, but others require it to be declared in advance. The choice she was denied was not of which individuals to vote for, but of which party's primary to vote in.

9/13/2006 9:43 PM

Kevin Mark said...

If there is not some cyber location already, there should be a location where electronic voting issues should be aggregated so that people can see them. If not in a real time blog, maybe a wiki, or site that posts users submitted stories/articles (maybe by state?).

anyway, keep up the fight for fair elections!

9/14/2006 4:29 AM

Petronius said...

I am an election Judge in Chicago, and earlier this year we had our first experience with some electronic devices. Our precincts are much smaller than yours (600 registered in mine). I don't remember who built our units, but we had problems all over town. The single voting machine worked pretty well, but the little gadget that activated the smart card was terrible. It had softie-keys like a Kiddie Komputer, and would sometimes beep when the number I was entering didn't. Or vice versa.

The other issue was during totalizing. Our readers had two slots, one for the OCR scanner machine's cartridge (more about that later) and a smaller one for the electronic unit's chip. Everybody, myself included, put the chip in upside down; there was no ridge or bump to prevent improper insertion. Fortunately, our units did produce a paper tape, so nobody's vote was lost.

We did use the OCR reader system for most of the ballots, and I like it. Our problem, though, is that every other election out here has something called a judicial retention race, with 95 names on it, plus the President, Senator, etc. The ballot would be the size of a Rolling Stone's poster.

My plea to all reading this: volunteer yourself. We need people who understand technology in all areas. At 55 I'm the youngest judge in the district. Help!

9/14/2006 10:11 AM

Anonymous said...

Seems to me a lot of complaints about the machines should be directed at inadequate training.

9/14/2006 10:22 AM

Seth said...

[The "electronic poll books that we used to check in voters"] each contain a full database of the registered voters in the county, and information about whether or not each voter has already voted, in addition to all of the voter registration information... These poll books turned out to be a disaster.

When I voted I personally experienced no problems, but I did overhear someone talking with a poll judge and the chief judge at the polling place. It seems that person was showing up as already having voted. Their solution: vote on a provisional ballot. Either that means anyone can vote multiple times so long as they use provisional ballots, or it means that the provisional ballots are meaningless and never get counted even in the case of a contested/close election.

9/14/2006 10:24 AM

Anonymous said...

What happens if the electricity goes out? Nobody can vote, right? With the old fashioned, lever voting booths, people can vote during a blackout. Technology, schmechnology!

9/14/2006 11:41 AM

Tolbert Feather said...

I worked as a provisional ballot judge in Frederick County, MD, for the 06 primary election. This is my second time as a judge. We had the same problems Dr. Rubin described with the tablet computers holding the polling lists. These machines were out of sync, but seem to catch up over time. They also crashed. This system either needs to be worked on or scraped.

We had 9 provisional ballots cast in the precinct, 1 was listed as absentee on the poll book, 7 were not listed as registered on the poll book, and 1 was listed as a republican but claimed to be a democrat. He got a democratic provisional ballot.

I wondered about the tamper tape too. I think it is designed to make the election officials in the main office feel better rather than really being able to show if anything was tampered with.

We had a little over 1000 voters cast ballots. I assisted maybe 50-100 of the voters. Two or 3 asked about a paper receipt. One was angry there was no paper trail. More education needed here so voters are aware of the pitfalls of these machines. It is certainly not for lack of good work on Dr. Rubin's part.

I also assisted a woman who had just become a citizen, along with her sister and mother. All were voting for the fist time and were very happy. Actually the majority of our voters seemed happy to be there.

It is a long day both physically and mentally taxing. My fellow judges are great people. We left the polling place at the end of the day planning how we would organize ourselves and the equipment for the November general election. It looks like we are all hooked on being a judge for a day. Frederick County had a shortage of judges for this election, as I am sure other counties in MD did too. If anyone is interested, they may volunteer.

9/14/2006 12:17 PM

auros said...

Since somebody mentioned Diebold ATMs, I should note that I have seen one crashed before.

In any case, thanks for all your work — I remember seeing your first report in 2002. I'm in Santa Clara County, CA, and while we're stuck with the electronic machines, we at least have printers (for recounting, and so voters can verify their votes in the paper record before committing them), better physical security (our tape seals are quite good — you'd have no trouble telling when they've been tampered with), and in the times I've worked the polls, the closest thing I've seen to a departure from procedure was using a machine on which the seal over the polls open/closed switch had gotten stuck to itself when being applied, so the "void" message (which is repeated all over the tape, and is extremely easy to reveal if you're not careful) was visible. And we had all four clerks observe the situation and sign for it in the exception log, and called it into HQ. And honestly, given that on our machines, you can only open and close the polls once each (after which the recording cartridge is locked — you can't put any more votes onto it), having that switch sealed is more a convenience thing than security — you don't want some mischievous partisan to interfere with a precinct by closing the polls on a couple of the machines, while nobody's looking, and having to peel off the seals to get at the switch (they're quite sticky), or get out a knife to cut them or something, would make it hard not to be observed...


RM "Auros" Harman

JHU '99, Computer and Cognitive Sciences

9/14/2006 1:11 PM

Anonymous said...


Did I understand right that the Chief Judge put back voided tamper tape? Why did he not replace the voided tape with new tape (which should have been available)? Was this recorded in the Voting Systems Integrity Report (both in Part 1 and Part 2)? Your report is disturbing from a technical standpoint, yes, but what hope do we have if people can't follow the rules?

9/14/2006 1:42 PM

Anonymous said...

Thank you for this great write up. I wonder what you think about requiring the companies that produce these machines to post the code to their software so voters can verify its accuracy. Why should we trust a company that makes its profit from the claim that its machines and software are good and accurate enough to handle one of the most important steps in the democratic process without being able to verify those claims? I even think that the code need to be open source.

9/14/2006 1:59 PM

Ed Terry said...

There is no excuse for the polling issues that took place in Maryland this past week. While the focus was Montgomery County, there were also polling issues in Prince George's County, Baltimore City and other parts of the state.

The solution is state-mandated standards for conducting an election. This would mean having the polling place certified as "ready" twelve hours before the polls open. Missing cards and power cords is simply ridiculous mistakes that should no longer be tolerated without consequences. Of course, this raises the issue of machine security, but if that means human security — then deal with it. It's part of the cost of doing business.

No voter should be turned away — no matter what. Using Provisional Ballots is not the remedy for lack of preparation. There is no excuse for lack of preparation.

Here is where Congress must come in — there is no guaranteed right to vote in America. There must be explicit legislation to equate the right to vote to be a basic, unalienable right for those qualified voters. With this, voters can seek civil redress from being denied their right to vote.

9/14/2006 2:32 PM

SilverSpring said...

Dr. Rubin:

Thank you for your report, and thank you heartily for getting into the trenches with the rest of us judges. I will temper my comments about pointy headed professors from now on.

In Montgomery county, where I served as a Chief Judge, we encountered most of the problem you did, and then some. The failure to provide access cards has been reported widely (worldwide, I'm told. This was an appalling, incomprehensible, and inexcusable human error, but a human error nonetheless. Under the previous punch card system, the equivalent error would have been to distribute the punch machines and forget the cards. Everyone would have said that was stupid, but they wouldn't question the integrity of the equipment, just the competence of the Board of Elections.

Your comments on unsynched poll books and tamper tape are interesting and I will pass them on to our Board. The comments of some of your readers betray a fairly common ignorance of the details of how the system works, and for those in small towns a lack of appreciation for the realities of big city life where a precinct can have 3000 registered voters and 1000 actual voters on election day.

I may comment in the future on some of the specific issues raised by your and your readers, but for now I'll just say that I intend to be a regular reader of your blog.

9/14/2006 5:40 PM

Expatriated Freedom said... why is this better than writing down your name on a slip of paper and putting that into a locked box?

9/14/2006 6:37 PM

Jefferson said...

As a Maryland voter, a question I've had about the "tamperproof" tape, ever since Linda Lamone announced it as some sort of panacea, is: what happens if an election worker discovers that the tape has been tampered with? What's the contingency plan? I asked my election judge this question in 2004 and she had no real answer other than that they would log that it had occurred.

It's an important question because a targeted denial of service could be made trivial by their contingency plan. It would be quite easy to simply tamper with the tape (in a way that would be evident). So would they throw out votes counted on that machine? Could a conspiracy swing an election by tampering with tape on machines at heavily leaning precincts?

It's very clear how completely clueless Maryland's election planning people are with respect to technology — they've never heard of an easter egg, apparently, and they don't understand the first thing about IT security. I and many others are very grateful to you for trying to enlighten them, and being committed enough to the cause to actually work in the trenches.

9/14/2006 10:19 PM

Gary said...

Thanks for the report. I am an election judge, here in Texas we call them judges and clerks.

Tamper tape — I have never seen anyone but me check them and it can be impossible to tell if they are tampered with.

Voter ID, even if the state provides free ID cards — constitutionally required since the outlawing of poll taxes, many people have no birth certificate which is required to obtain the ID and it is long process to get a replacement. This is a GOP ploy to depress the vote of more Democrats than Republicans.

Machine paper tapes, the voter never sees them but they are run off at poll opening and poll closing. They cannot catch programming errors but only errors or fraud at the central tabulator PCs. They only do that if a candidate requests a specific and longer recount which includes auditing the machine tapes, which most candidates don't even know about. I have also heard of this being denied before being agreed to at over ten times the price of a normal recount.

No one has mentioned machine storage — in my Harris county the polling machine tabulators are stored overnight at the chief election judges home for each precinct.

Overall election supervision — the people who run the elections are appointed by elected officers who are usually extremely partisan. If you control a county or state there are many opportunities to make numerous decisions that favor your party over the other, from location and number of voting precincts and voting machines to the training, to appointments, to state level decisions about the process of voting. The Ohio and Florida Secretary of States in 2000 and 2004 are obvious partisan examples.

9/14/2006 11:31 PM

Woody Pidcock said...

My concern with the effect of having such a glitchy in person voting experience is that most people will now opt for absentee ballots instead.

This means that exit polls will become meaningless.

In the 2004 election, we had potential evidence of election fraud because the exit polls were so skewed from the reported election results, especially in battleground states and where diebold machines were used.

With absentee ballots, we will not have this evidence. I doubt if we will have a paper trail for diebold machines by 2008. But even if we do, what protections are in place to ensure that absentee ballots are not tampered with?

9/15/2006 12:56 AM

phil said...

By using Electronics, Digitized Data, and Networks in our elections, our elections can NO LONGER BE VALIDATED!

They have cost us trillions of dollars and killed thousands of people now.

If your going to REQUIRE a VALID ID, then your going to have to REQUIRE your STATE to provide FREE STATE ID CARDS FIRST!

9/15/2006 3:52 AM

Shannon Williford said...

Dr. Rubin,

I was a poll worker here in Nashville, TN for our primary last month. We used "new" ES&S equipment (the company actually sent us used stuff, even though Davidson County signed a contract for new...), and I was serving in my first election.

We opened in the morning and I was asked to work the voter registration books, even though I had been "trained" to work the machines (though I'm fairly computer-illiterate...), so I observed as others daisy chained 6 machines and then opened each without incident by inserting a cartridge until we tried to open the last machine, which would not work. As the last machine's opening was supposedly crucial to opening the whole precinct — we used a different cartridge in it after it was up and then were to print (on a new, yet ancient-technology-looking-grocery-tape printer) out a tape certifying the machines to have no votes on them. So, without being able to print, we called our downtown election commission bosses, who told us to open anyway, letting the voters vote on the machines that were open. The commission sent a tech who got the other machine up and printed the tape by 9 AM.

The rest of the day was more routine, but I wonder how the opening problem affects the vote count. I couldn't tell from my vantage point...

Shannon Williford

9/15/2006 11:24 AM

Paul said...

I, too, was an election judge in Maryland on Tuesday. I served as a chief judge in a Montgomery county precinct, and we had electronic poll books (EPBs), too.

Our EPBs wouldn't sync, either. But since I and the other chief judge were pretty computer savvy, we worked with them a bit on Monday, when we were setting them up.

In the Network area, the EPB shows the other EPBs it is syncing with. One EPB showed all three EPBS (itself and the other two). The other two EPBs only showed each other, not the third one. We tried every combination we could think of: rebooting them in different orders, attaching the single printer we had to different machines, trying just two with and without the hub, etc. One knew about all EPBs while the other two didn't.

Most of the time one EPB was more than enough, but late Tuesday when we had a small rush, we used an additional pollbook. Later we realized it was the odd one. Sure enough, it showed its voters and the main EPB's voters, however the main EPB never knew about the odd one's voters. We tried to be very careful...


9/15/2006 3:58 PM

concerned voter said...

Thanks for posting this very informative blog.

I would recommend to everyone to watch a 10 minute video posted yesterday titled, "Princeton University Exposes Diebold Flaws," posted on

Regardless of whether you are Democrat or Republican, the ease with which a virus can be installed on these machines to rig an election should be a matter of grave concern.

9/15/2006 7:33 PM

Pete said...

Hello. I'm sorry to hear so many bad experiences regarding electronic polling.

I've been involved in the electronics and IT fields for quite a few years. I have a solution for keeping the count straight and being able to verify and conduct recounts up until the following election. It's very simple: Use paper ballots and keep them until the next election.

You might have trouble with silverfish or mold, but crashes and viruses? Never.

And it will be a whole lot cheaper than the prices I've seen quoted for the electronic marvels that are being complained about.


9/15/2006 9:08 PM

David Smith said...

Avi: You nailed it. None of the protections work, the technology is unreliable at best and the protocol for problem resolution is, to be generous, bull manure.

There's nothing wrong with a #2 pencil. Even State Senators can figure out how to make the letter X next to their choices.

Worst case? Use Mark-Sense ballots, so you get the speed of a near instant count, but have a paper trail that you can check in a judicial recount.

Or if a disgruntled poll worker decides to pour a 32 oz. Code Red Slurpee into the technology, you can re-run the Mark-Sense paper ballots on another machine.

9/15/2006 11:05 PM

Rebecca said...

I serve as a Chief Judge in Prince George County and can corroborate all of the problems already noted here, including crashing and out of sync e-poll books and voters' party registrations being listed incorrectly, which only seemed to happen to Democrats and usually listed them as unaffiliated, including one woman who said she has been a registered Democrat for 50 years!

I'm wondering if any of you election judges on this blog tried out the Visually Impaired Ballot Station (VIBS) to see if it was working correctly? There was an entire chapter in the election judges' manual devoted to serving voters with disabilities, but it only discussed etiquette and said nothing about how to actually operate the equipment. This was not covered in my training as a chief judge, and the equipment set-up in the manual merely said to plug in the headphones and keypad into any DRE machine. But when I asked one of our judges to vote on it when he cast his ballot that day, all he could hear was a bunch of high-pitched screeches and blips.

In 2004 when we programmed our Voter Access cards with small handheld encoders, we could select "VH" for a "visually hidden" or audio ballot. But this year we programmed those with the e-poll books. Our technician, who was a very conscientious young woman, searched through her training manual to try to find out how to do it, and even called the roving tech and the county BoE, but no one knew. Did any of you other election judges figure out how to do it?

As has been noted several times in these blog entries, MD, like most other states, suffers from a shortage of election judges (or whatever poll workers are called in your state or county). I urge you to consider serving as one in November.

To learn more about becoming a poll worker, or even if you are already serving as one, please consider signing on with Poll workers for Democracy at:

The purpose of the program, from their website:

"Having a flood of informed citizens like yourself take part in running elections locally is key to recovering a more transparent and accountable democracy, in alliance with the elections officials who want to see votes counted accurately, and voter confidence restored. The observations you make on Election Day will be collected through an online survey, and will be part of a national Election Incident Reporting Database supporting informed reform, media outreach, and other efforts where needed."

Poll workers for Democracy is a collaboration between Working Assets,, and Mainstreet Moms, and will be reporting incidents to the national Election Protection Coalition's database.

—, working for Secure, Accessible, Verifiable Elections in Maryland

9/18/2006 2:08 PM

Anonymous said...

I also was an election judge, a book judge in Anne Arundel county where we had similar problems with the electronic poll books not being in sync, though they would catch up (slowly) after rebooting. The rebooting happened often enough because they crashed so regularly. We tried plugging them in separate outlets but we didn't know if they were actually separate circuits or whether all our electronics, voting machines and poll books, were on one circuit.

Most interesting, to me, my husband who had applied for an absentee ballot and had cast it already, was listed as not having voted (or applied for absentee) in the poll book. I was told that the list of absentees was not up to date in the machine and if he HAD come in and voted again, they would throw away his absentee vote at headquarters. Hmm

9/18/2006 9:19 PM

Anonymous said...

I am an election judge in Northern Virginia who shares your concerns, Dr. Rubin, if not your expertise. My jurisdiction uses all the same Diebold machines you have critiqued, and I likely received the same worthless Diebold training that you did. My 'chief' is a retired CIA officer. As a Northern Virginia professional, I am not without intel community connections of own, so I know BS when I see it. Any suggestions, red flags?

9/18/2006 11:18 PM

Anonymous said...

Excellent report.

I'm also an election judge (specifically, a provisional-ballot judge) in Maryland (Anne Arundel County).

While the problems in Montgomery County have received widespread attention, the less extreme problems Rubin reports seem to have been common in our county, from what friends tell me. Our small, rural precinct (where the electrical power fluctuates at the best of times) had trouble with the synchronization of the poll books, for example. We had a Diebold rep who came through very late in the afternoon. He couldn't do anything to fix the problems — his response, according to my chief, was along the lines of "Keep doing what you've been doing."

The training we received this year was virtually useless. For example, we were told, "Any day now the electronic poll books might be approved for use," but I don't know whether anyone ever got any hands-on training on those "books." Certainly our judges who were responsible for them never did. The rushing into use of this system, which — as I think Rubin noted on a radio broadcast, seemed to have been created to "solve" a problem that didn't exist — caused most of the hassles on our primary day. The other big time-sink was setting up and, especially, taking down all the machines. Our training was definitely geared toward avoiding litigation-prompting back injuries in lifting the machines, etc., but the real problem was with all the crap inside the machines: all that paper tape, requiring rolling and counting and unsealing and signing and other multiply taxing tasks when we'd already been on our feet and working for 14 hours with scarcely a break.

As Rebecca, the Prince George County chief judge who posted above (my experience was very similar to hers), noted, the training also spent rather a lot of time on the etiquette of dealing with a diverse population...I'm an old lefty, and I don't think "political correctness" should be a bad word, but, damn, this wasn't half as much of a problem, in the trenches, as keeping the machines in sync, to name only one problem.

And our manuals were horrible. They used to offer clear checklists and complete procedures. One of the first instructions I read in setting up the provisional area was "Verify the ballot numbers on the upper right corner of the provisional ballots." Verify against what, exactly? My chief couldn't tell me, and anyway he had about 427 more pressing things to do, so I never got an answer. I've worked in this precinct and county before, and I never had weird little problems like this.

A few more things about our procedures in Maryland:

Voters in Maryland do not have to show ID — not even the voter card, which they should have received in the mail. (Many of my provisional voters remarked that their having not gotten voter cards in the mail should have tipped them off to problems with their registration.)

But — and I don't think anyone has mentioned this — I have no idea what sort of investigation, if any, I, as a poll worker, have undergone. I signed an oath when I volunteered to be an election judge, but I don't think I ever showed anyone, anywhere, any sort of ID. I'm not nuts about the notion of having my privacy invaded by an investigation, but it does seem a little weird. There have been instances — I've been party to them — where people have shown up on election day saying "The election board told me to come work here," and it can be tricky to verify that fact in the midst of everything else.

We are paid $25 for the training session we receive (which is mandatory, I believe) and $110 for the day as a judge. The chief judge gets another $30. There are two chief judges, one from each major party. The other judges are likewise supposed to be split among Democrats and Republicans, though staffing problems mean that that's not always what happens. And we're supposed to be assigned specific functions (e.g., mine is handling provisional ballots), but that doesn't generally hold true for the full day. Not having many provisional voters, I spent a good bit of time helping voters with the voting machines and a little time with the electronic poll books.

Our materials are at our polling place when we arrive. We folks working at your local precinct are not the ones to blame if the voting cards didn't show up; it wasn't our responsibility to get them there. (I thank God I don't live in Montgomery County anymore.)

We arrive at the polling place at 5:45 AM, and we're not allowed to leave until all the work is done. Then the chiefs have even more work: delivering the votes, in a secure fashion, to headquarters.

We aren't allowed cell phones or newspapers; we probably weren't supposed to be hearing about the problems in Montgomery County, but of course our voters told us. We're allowed diversions like books, but we don't get long enough breaks to do any reading. We have to provide our own food and drink. We aren't supposed to leave the building — although of course we had judges taking ciggie breaks out back of the school (on what was the only day with lovely weather that we had in a week).

I'm less concerned about the technological and security problems than I am about how their effects are trickling down to the poll workers and voters. When the system leaves us, in real time, looking like bumbling idiots, of course the voters aren't going to trust it. And when we have to hassle with problems that might have been avoided by better organization, that hundred bucks is gonna look less and less worth it. I heard stories of poll workers walking off their jobs on Election Day. And I wonder how many we'll lose before the much busier general election.

One final note in this long post: Unless I missed it in the reams of distracting paperwork provided by the election board, nowhere is there a form or other mechanism given for us, the people who actually dealt with voters and machines, to provide feedback on how the day went. You kinda get the idea that the folks at HQ just don't want to know.

Avi, maybe we Maryland judges should start us a committee, hon?

9/19/2006 10:21 AM

Anonymous said...

I worked as a chief judge in Anne Arundel County this year. I strongly agree with everything Anonymous, the provisional judge, said. I haven't had time to write to the Board of Elections about my concerns, and I'm not sure they will be read or heeded if I do, but I will do so anyway. At least I'll have tried.

This afternoon, after reading the story in the Annapolis Capital ( I seriously thought — for about five minutes — about quitting before the general election. Looks like they're making the workers the scapegoats. We do our best to work with malfunctioning equipment with window-dressing security, and when the system is shown to be a complete mess, what does Linda Lamone do? Blames us, if that story quoted her correctly. It's mostly human error! Maybe so, but I think she's looking a little far afield for the humans who are to blame.

Great — so all of us chief judges in Anne Arundel have to do another training session because in one precinct (or maybe two) — out of how many? — forgot to take the memory cards out of the voting units. I wonder whether the person who decided that made sense was the same one who devised the feeble excuse for a training session I had in July. It lasted three hours. I hope they schedule at least four or five hours this time, because I plan to go in with a long list of questions and remain there until I get at least some semblance of an answer to each of them.

My main beefs about the training session in July are that it was unfocused and lacking in a clear sense of priorities and that we were made to spend far too much time setting up and dismantling the voting units and no time at all on most of the things we chief judges actually spent the rest of the election day doing. They could make sure the unit judges are trained in setting up the units and let them direct that part. When training time is limited, it's idiotic to spend the chief judges' training session on that. They could also drop the inane PowerPoint presentation. All that stuff is in the manual, and I can just as easily read it there. Lots of the things we need to know are not in the manual and could be provided by a couple of experienced chief judges and not a couple of people who, as far as I know, spend election days at headquarters.

Furthermore, the security stuff the Board of Elections is so concerned about (not that it *is* all that secure, as Avi demonstrated) is presented as no more and no less important than locking the legs of the voting unit correctly. They know what they need us to know (maybe) but they don't seem to have any idea what *we* need to know.

It was my first time as Chief Judge, so I wasn't aware of how much wasn't covered in training until I was on the job on primary day and found myself woefully underprepared and having to wing it. I like to think I did a decent job, but I have suffered under chief judges who didn't — and it was their bad example that made me decide to take the chief job when offered, so that I would not have to be managed by dithering incompetents. (I feel a bit more sympathy for the ditherers now, since they probably had lousy training too.) God knows I didn't take the job for the extra $30 and the opportunity to spend two and a half more hours, after the polls closed, driving to Glen Burnie and waiting in line and driving back after being on my feet since 4:30 AM.

I won't quit before the General Election (unless they piss me off too badly at that extra training session). I feel I made a commitment to do the work in 2006 and, more important to me, our precinct had a fine group of poll workers, and I feel responsible to do my best for them, as well as for our voters, in November. But I'm very discouraged by how all this is playing out. It's a grueling day, and the main reason I do it is to contribute to my community, but I'm not sure whether I can go on feeling that I'm contributing by participating in such a botched process. If I don't have any faith in the integrity of the process, how am I to reassure concerned voters?

9/19/2006 11:24 PM

Anonymous said...

Linda Lamone says she's the 'boss' and the 'buck stops with me,' Let's hold her accountable for her admission.

9/20/2006 6:04 PM

Anonymous said...

RE: Here we have a Secretary of State who has insulated us from the experiences you recount so well. She has done this, so she would have the State believe, by entering into a $15.7 million contract with a New England based firm, LHS Associates, to provide AccuVote OS Op Scan units to replace — belatedly — our venerable Lever Machines.

Clearly, The strategy by Diebold Election Systems Incorporated (DESI) to use distributors has worked for one Connecticut resident. The comment above makes it clear this confused resident thinks the $15.7 million dollars is NOT for Diebold equipment but for LHS equipment. I suspect it will be a bit of shock when he/she walks into his/her polling place in November and sees the word "Diebold AccuVote" on the front panel.

Part of the fault for this confusion lies with the Connecticut Secretary of State. She has studiously avoided any mention that Connecticut has bought equipment from Diebold on behalf of the whole state by referring to LHS (the distributor of Diebold with whom the contract was written) as the vendor and by referring to the equipment by model only (AccuVote OS) and never as Diebold AccuVote OS.

9/21/2006 10:27 AM

Anonymous said...

I'm reading your blog 8 days after the election, and I am still incensed. I served as an election judge (check-in) in Montgomery county. We experienced similar problems and additional ones to what you described. I am very concerned about elections all over the country, where similar equipment is used. Every citizen, every voter, every elected official, every candidate should be very concerned as well.

June J.

9/21/2006 2:47 PM

Anonymous said...

Anonymous in Montgomery County said,

"Everything went very smoothly and as I exited, I made a comment to one of the election officials about 'still no paper trail, huh' and a comment that the diebold machines don't have a very good reputation. In response to my paper trail comment, the official said no, there was a paper tabulation printed internally for each machine. He pointed in the direction of the exit and said the printouts showing each machine started with a zero vote count were on the wall. Heading towards the exit, I find that the printouts for each machine (with original ink signatures) are taped up on the OUTSIDE of the entrance. I wondered what inconvenience would be caused if someone grabbed a handful of these and walked off with them. Considering these printouts were proof the machines started from a known, zero count state, and had the original ink signatures from the officials on site, it seemed pretty stupid to tape these up outside the building."

It wouldn't cause much inconvenience — just that voters would not have those printouts to check. Those are copies of the zero report. There is another zero report inside each voting machine, left on the printer, also signed by the judges, and the totals report is printed at the close of polling such that it is continuous with the zero report left on the printer, and it too is signed. Those two connected reports are then removed and turned in to election headquarters along with the memory card.

I agree it's stupid to post them outside of the polling room. We tape ours up near the exit to the polling room, but inside the room, where we can monitor them.

9/22/2006 11:39 PM

cmoose said...

Commendations to Dr. Rubin for not only his engineering analysis and expertise, but also for working as an election judge and experiencing the problems at ground level.

I will describe my experience as a Chief Judge below (boring to most); but I would first like to attempt to find fixes for the system as far is possible for the November elections, as we are apparently stuck with the Diebold system at this point — let's try to determine the scope of some of the problems and see which can be reduced using better physical security, and eliminate some by fixing software bugs. (Our greatest problem was lack of judges — two for 11 voting machines in two precincts!) I have been hearing proposals to switch to other technology — but I believe an attempt to switch methods at this late date would likely be even more disastrous — once we got started and gained some experience, the Diebold system worked smoothly except for the E-poll book crashes. Although the system CAN be compromised, let's minimize the odds that it will be.

MACHINE TRAY KEYS: I've heard this laughed at on talk shows, but consider it a minor issue. The (inadequate) TAMPER TAPES are supposed to be the primary protection of the memory cards — no one expects a simple lock to be more than an inconvenience to a determined culprit. It is unreasonable to have a different key for every machine, and even for each machine in the same precinct. It would be a logistic nightmare to manage a large variety of keys.

TAMPER TAPES: I would expect that a better security tape can be found before November. If not, and in either event, give election judges detailed instructions for careful examination of the tapes, and impress on them the importance of logging the numbers and inspecting them during the day. After the final loading the programs for each precinct, they should be sealed and logged, transported securely (ours were delivered by a policeman), checked and logged before opening polls, and new tapes put over the tray locks before the first vote, and their numbers logged in. Verify the tape numbers and integrity upon closing the polls. Maintain physical security of the memory cards until delivered to the Board of Elections (BOE). (Police signed for ours and delivered them while we were completing poll duties.)

VIRUS PREVENTION: The Princeton report shows how easy it is to spread a virus which can redirect votes from one candidate or issue to another without disturbing the totals. But I believe such a virus should be effective for ONLY the machines at one precinct (or at least within one district). As the program for each precinct contains ward and precinct I.D., and the memory locations for candidates would differ by district, it seems that a program to introduce a virus would need to be customized for each precinct. Use of a checksum should make it possible to assure the program for each precinct had not been tampered with (though I am not sufficiently familiar with the Windows O.S. to know the process of creating checksum) and which would be verified by a Chief Judge on election morning, matching one in his packet with each voting machine.

E-POLL MACHINE CRASHES: This was a major problem, but should be easy to fix. Our precinct had two E-poll books. A message on a red background said "unable to read card" and said to remove the card. After the first crash I set the VAC card aside, and were relieved to see that the E-poll automatically rebooted, but discovered the page showing the votes cast for each party was blank. Later the second pollbook crashed, and I believe the bad card had been inadvertently picked up in the recycled VACs. (I later wondered what it would take for a "smartcard" to go bad — a strong magnet?)

After more crashes, we learned it crashed whether the card was removed or not. This appears to be simply bad programming for error recovery. If the program can recognize that a read instruction has failed, then go on to notify the operator, it should be able to proceed with a reasonable recovery. It was obvious that it flagged that voter's record as having voted although the card had not been written and verified. This should be easily fixed by not marking the record until the card has been written for that voter and verified, then setting the flag and relaying it to other Ebooks in that precinct. Diebold should be able to correct this bug and verify the results promptly.

Commendations for our technician: after exhausting his book to find how to recover the breakdown of party votes so we could complete the hourly reports and count checks, mid afternoon he was able to contact a Diebold engineer and learned that we could go through a hierarchy under Statistics and locate the counts by party.

SYNCHRONIZATION OF E-POLL MACHINES: Synchronization of the poll books within a precinct did not occur to me until during the election. It was not discussed in my training class — we were told the technician would set up the voting machines the night before and the E-poll machines on election morning. Apparently he did not know to connect the Ethernet cables nor anything about synchronization. After recognizing the sync problem (after a crash) we used one book exclusively unless a long line built up; then used the second machine (the first one to crash) only as needed, thus minimizing the synchronization problem. The E-machines were much faster than the old card-lookups, and we could usually keep up with the voting machines. As long as we were able to keep our seven voting machines busy, there was no point in queuing up their lines just to shorten the check-in line.

Recently, I searched the judge's instruction manual; the only reference to synchronization I found was p. 5.6 saying the technician would connect the poll books and on p. 11.2, reference to Ethernet cable to connect e-poll books "in the same polling place together." I found no other reference to synchronization, the process nor even the use of the word.

From Dr. Rubin's blog and links, I learned an Ethernet system is supposed to synchronize the E-books within a precinct. (I heard one report that an effort was made to connect more than one precinct in the same room, but there is no point in that, as a voter can vote at ONLY his assigned precinct, and would mess up the vote counts if it were possible.) A VAC should be accepted only on voting machines within the precinct of the E-book which created it. And I doubt the vote counting program at the Board of Elections can process and correctly tabulate votes from the wrong precinct's voting machines.

But the E-book is excellent in locating voters showing up at the wrong precinct or location! (I heard the E-books include all the voters in the state — and I verified one in a different county!) We directed many voters who did not know their precinct to the other table in our room.

Is it not feasible to have software which could display, for a chief judge, the status of the operation of the network? On my home computer, Windows has software to report status and can even attempt to "repair" a faulty wireless network — I would think Ethernet should be even easier to troubleshoot and repair, or report if there is lack of continuity in a cable or connector.

PRIVACY: I had a couple of complaints by voters who felt that anyone walking (or standing in line) behind them could see their votes over their shoulder. I decided it was too risky to reverse the machines to face a wall because of the daisy chain connections, anticipating problems if we tried to disconnect and reconnect them during the voting. Later I realized that reversing them would still require voters to walk behind other voters, as the daisy chained power cords would preclude walking between the machines, and if you had five machines connected, a voter might have to pass by two (if both ends were open) or up to four other machines to get to the last one in line. Any suggestions, besides maintain a queue of waiting voters at an appropriate distance behind voters until a machine is vacated?.

ELECTION DAY REPORT: I had been a Chief Republican Judge for over a decade at this location where there were typically 10 to 12 judges. I arrived before 6 AM and found only the technician and a new Chief Democrat judge. A policeman arrived with the bags of materials, but without the requisite two judges (Republican & Democrat) for each precinct we did not sign for nor open the bags. We arranged the voting machines nearest the locations used in previous years to minimize confusion of the voters, but where had been three precincts there were now two. By 6:20 we began calling Board of Elections (BOE) for help, and with none there by 7, we were directed by the BOE to open with the two of us signing as the other party judge at the other's precinct.

The technician had left for another precinct where they had problems with power outlets, and we were almost stymied by lack of the power cords to daisy chain the voting machines — we finally found them in a blue metal box on the cart which had been used to transport the voting machines, so we could begin connecting our seven machines and doing the necessary paperwork, verifying tamper tapes, etc.,etc. Another judge arrived around 7:30 and the tech returned, helping us open the polls — which we did a little after 8:30 AM. The technician served as an election judge at my precinct and the other judge helped at the other precinct. (I don't know what party he represented.) One of their four voting machines appeared to malfunction, so they used three.

Once we worked off the long opening line, the system seemed to operate smoothly, except our being so overloaded caused some things, like maintaining the hourly counts, to slip thru the cracks; it is not easy to get the hour counts and party breakdown when the machines are in use and voters queued up, as the count of E-poll VAC cards issued will be greater than the vote counts on the voting machines. At the end of the day, our vote tallies on the seven machines were two less than the VACs issued according to the e-poll book totals (apparently two were issued cards and failed to use them, or the E-poll added a count when a crash occurred but a card was not issued).

Most of the voters seemed to be comfortable with the voting machines — a poll worker/coordinator from another county helped instruct voters to insert their VAC, and to use the touch screens, and several voters commented they were easy to use. Each voter's signed printed slip was placed in an envelope taped to the back of the machine used by that voter.

The E-machines were much faster than the old card and poll book lookups, and we were able to find the voting precinct for those at the wrong table. Some voters' I.D. cards showed the wrong precinct due to the elimination of one, and to some boundary adjustments.

As the load increased approaching noon, I recruited the poll worker, swore him in and filled in a payroll requisition; he stayed until we were packed up about 10:30 PM. And the voters were very patient with us, in spite of the late opening. They were used to seeing about 10 to 12 judges at this location and sympathized with our predicament.

We concluded that the crashes were being caused by bad "smartchips" on the VAC cards, forcing us to go to Provisional Ballots for those willing to endure the tedious process of completing them. We had one voter with a 2005 voter ID card, who was not in the E-pollbook — BOE said could find no reason why he was not in the data base, and directed us to use a Provisional ballot for him as the E-book could not write a VAC for him. Unique case.

After the crashes I became aware of the synchronization problem

After noon I requested BOE provide some help, at least for the closing of the polls — mid afternoon three workers arrived — two helped at my precinct and one at the other.

We began to hear of problems in Montgomery County and Baltimore City, and that polls might be kept open an additional hour. There was discussion by some workers who said unless they were officially notified, they would shut down at 8 PM. I called BOE and was told that the City polls would be open until 9 PM (using provisional ballots after 8), and that police would deliver a notice requiring the judges to work an extra hour and would bring a kit for the extended day, including additional provisional ballots. We kept our machines operating until 8:30 when the police arrived at out location, began shutting them down and continuing until 8:55 when the last Provisional ballot was cast. A few who left before we were able to open voted after 8 that evening.

The closing procedures were confusing — I verified the voting-day tamper tapes appeared intact, and was trying to find the color-coded envelopes for the printouts to determine how many sets were needed. I said I knew we needed at least three or four, and to keep the zero count attached to the first final tally. By the time the envelopes were found and I realized we needed at least four sets of printouts, some of the voting machines had been shut down with only three sets, so we were unable to post our final results outside the building. (I taped up the set from the other precinct's three machines).

I waited until printouts of results were completed before removing the memory cards; I was concerned that removing the memory cards might disturb the printing, though I believed the printers worked from main memory. The memory cards were logged in and turned over to a policeman for delivery to BOE.

We two chief judges and remaining workers signed all sets of printouts from all 10 machines and some of the final report forms for both precincts.

It was impossible to complete all the forms and to organize the material in accordance with the instructions for turn in — for example, we never did learn where the envelopes with the voter authorization printouts should go. And we didn't have time to check count of the voter slips behind each machine with the vote count for that machine (The maintenance man was threatening to turn out the lights on us after 10 PM.) We left about 10:30.

At BOE the other chief judge and I "ran the gauntlet" hoping to be able to locate the correct items for each check point. I was told I needed either the machine memory cards or the receipt signed by the policeman — I had to return to my car to get that, but otherwise got through the check-in without much difficulty, about 11 PM. A LONG day.



1) I have no reason to believe that the results produced at our precinct were not accurate. If most of the problems above are solved, both technical and physical security, I think the Diebold machines can be used in November with confidence, if the facts and precautions can be presented to the public — "perception is reality?" And trying to switch to a different system and retrain judges by November could be catastrophe. If large numbers of voters used absentee ballots or provisional ballots as has been suggested, I expect it could be days or weeks before the results were known, and the human error in counting such votes would likely cause challenges and recounts in close races. And provisional ballets take much longer to complete with the accompanying information required — the number per hour would be MUCH lower. I have heard no reports that there were any serious or suspicious discrepancies, which would suggest fraud or serious errors.

2) Has a breakout of the machine results by precinct been compared with the totals on the printouts for a spot check for a match? I realize that will not catch vote changing, but might catch other possible bugs in the processing.

3) For several elections I have theorized that the most likely way to defraud an election would be to use the main computer which tallies the votes election night. That is where a vote switching virus would be most effective. I asked once if it was possible to see the program software used for the main tally, but was unable to find anyone at BOE who could arrange it.

4) Those proposing use of more provisional ballots are probably unaware of the length of time it takes to complete the paperwork for each, and lack of tables or work areas where they could be used with privacy, and how much they slow the process!

5) As the severe shortage of judges was our most serious problem, I would like to know the basis and scope of the problem. Was the number of judges recruited reduced on expectation that the new machines required fewer (even on their first use)? Were there many judges who took the training classes but failed to report on election day, or was the BOE extremely short of judges trained and failed to warn us of that fact? There was one theory that there was deep resentment that Baltimore City had to scrap the lever machines which were owned and paid for, denied a waiver to use them once more in 2006, and forced to use expensive leased electronic machines, and judges may protested by refusing to participate. More likely in my opinion, there were many elderly judges who may have expected an hour or so for instruction on the new machines, and were overwhelmed by the complexity, thick manual and nearly four-hour class, and still felt unprepared to do the job and decided it was not worth the effort on what was likely to be a disaster.


I have some suggestions for the BOE staff regarding improving training, instructions in the manual, and procedures, which I will offer to them later.

Charles Moose

9/26/2006 12:15 PM


About Me— Avi Rubin


I am a Professor of computer science at Johns Hopkins University in Baltimore, Maryland, U.S. I teach courses in computer security and privacy. My primary research interest right now is the security of electronic voting, and I am the director of the NSF ACCURATE Center. I'm also an election judge in Baltimore County. Besides my faculty position, I am also the president and founder of a computer security consulting firm called Independent Security Evaluators. My hobbies are photography, soccer, tennis, golf, billiards, high tech gadgets, and playing with my kids.



| EJF Home | Where To Find Help | Join the EJF | Comments? | Get EJF newsletter |


| Vote Fraud and Election Issues Book | Table of Contents | Site Map | Index |


| Chapter 12 — Voting Problems In The 2006 Elections |

| Next — Various News Reports — 2006 General Elections |

| Back — Various News Reports — 2006 Primary Elections |


Added September 30, 2006

Last updated 6/14/09