Fwd: Interview with Georgia Diebold Election Machine installer

From: Arthur Keller <arthur_at_kellers_dot_org>
Date: Fri Aug 08 2003 - 05:54:41 CDT

>Hi all,
>If anyone wants to know what happened in Georgia prior to the Nov '02
>election, you gotta read this interview Bev Harris did with a technician
>working on the Georgia voting machines. This is UNREAL!!!
>Have your vomit bag ready. That whole election should be canceled
>and re-run. Punchcards would have been better.
>This is longish and a bit techie, but any damn fool can see that any
>pretense at believability of the election results is missing. Control
>of the US Senate hung on just this kind of election.
>What really happened in Georgia?
>INTRODUCTION: This is a story about an odd folder found on an open
>web site, called "rob-georgia.zip." It contained program updates for
>the 22,000 voting machines used in Georgia, illicit patches
>installed without examination or certification. Because of these
>unexamined files, no one will ever know quite what programs were on
>those 22,000 voting machines. (If you are a computer programmer, go
>to the forum at this site, where you can examine the rob-georgia
>patch and discuss your findings with others).
>Dr. Brit Williams, referred to in this interview, is the official
>voting machine examiner for the state of Georgia. To see what he
>claims is the procedure he followed, click here:
>If you like obfuscation and plain old-fashioned lying, also see the
>"official versions" of this story here:
>But if you want the straight story, meet Rob:
>Interview: March, 2003:
>Harris: What was the FTP site for?
>Rob:One of problems we had was an issue with the GEMS database. They
>had to do an update to it, so they just post the update to the web
>Harris: What was rob-georgia?
>Rob: I believe what that file was for, I did a -- well, there were a
>ton of holes with the programs on those machines. When they all came
>into the warehouse, I did a quality check, this was something I did
>on a Saturday. I found that 25% of the machines on the floor would
>fail KSU testing --
>Harris: "What is KSU testing?"
>Rob: "Kennesaw State University. We knew basically what they would
>be testing and the trick was to make sure the machines would pass
>the testing. So I went and checked a pallet and found it was bad.
>And I checked another, and another, and I knew we had a problem."
>Harris: "Was that both you and James Rellinger?"
>Rob: "James dealt with the network, but I was dealing with the
>touchscreen machines themselves.
>Harris: "What kind of problems were you seeing?"
>Rob: "…One of the things we had wrong was the date wasn't sticking
>in the Windows CE. The real time clock would go to check the time on
>the motherboard, and it would have an invalid year in it, like 1974
>or something, and basically the machine would continue to keep
>checking. Every time it checked, it saw that the date was not right
>and this put it into a loop.
>"They had to do an update in CE to fix all those dates. So the way
>we did that in the warehouse was, they would post whatever the
>update was on the FTP site. James would go get the file and put it
>on the [memory] cards. Because you load everything through the
>PCMCIA cards. You boot it up using the card and it loads the new
>"This was done in the warehouses -- once the machines were sent out
>to the county, these updates were done just to make sure the
>machines were running correctly. I went over to Dekalb [County]. We
>updated 1800 machines in basically a day and a half. I still
>remember ol' Rusty, down at the warehouse, we ended up touching
>every single machine off the pallet, booting 'em up, update it, we
>had a couple hundred machines done when in comes a new update over
>the phone.
>Harris: "You mean you used a modem or they called you on the phone?"
>Rob: "No. A phone call. They'd say 'Oh no no, the way we had you do,
>that's not going to work, here's another thing to do. Okay, we just
>did a few hundred machines, now we gotta do it this way -- But we
>got it done.
>Harris: "Did you personally ever download anything at all from the FTP site?"
>Rob: [it was] mostly James.
>Harris: "Did you work for Diebold, or James Rellinger?"
>Rob: "I worked for ABSS. So did James."
>Harris: "What about the rob-georgia file?"
>Rob: "I think they put it out there for me when we were doing the
>Dekalb thing, but I was busy managing the whole crew so, I had my
>laptop out, and one of the engineers used my laptop -- or maybe it
>was James -- one of them had to go in and get it from the FTP, put
>it on a card, make copies of the cards and then we used them to
>update the machines."
>Harris: "So one of the people downloaded the patch and then made
>copies of it?"
>Rob: "They use my laptop. It was not secure, either. They just used
>the laptop to repro the cards. Diebold never gave us anything with a
>PCMCIA slot, then they'd tell us, 'Go download this,' so we'd have
>to get out our own laptop to do it."
>Harris: "Who instructed you about the FTP site? Was it a Diebold employee?"
>Rob: "It was Diebold."
>Harris: "Was it the people in Ohio or the people in Texas?"
>Rob: "The people in McKinney [Texas]."
>Harris: "Who were some of the Diebold people? Do you remember any names?"
>Rob: "Ian. I remember one of the guys, Ian, I can't remember his
>last name. One of the main guys we dealt with was a guy named Ian.
>He was actually involved in the design of the motherboard. He was
>very much involved in trying to figure out how to fix the problems.
>So they sent us upgrades, but then after we did it KSU still failed
>a ton of machines."
>Harris: "As I understand it, they send the system to Wyle labs for
>certification, and also to Ciber to test the software. But from what
>you are describing, I can't understand how the machines got through
>what they are telling us is 'rigorous testing.'"
>Rob: "From what I understand they ended up figuring out that the
>cards that we were loading that fix that Diebold provided for us,
>well they were never tested, they just said 'Oh here's the problem,
>go ahead and fix it.'
>Harris: "So what is your opinion about the certification testing?"
>Rob: "No, it's not just that. NOBODY even tested it! When I found
>that out -- I mean you can't not test a fix -- I worked for a
>billing company, and if I'd put a fix on that wasn't tested I'd have
>gotten FIRED! You have to make sure whatever fix you did didn't
>break something else. But they didn't even TEST the fixes before
>they told us to install them.
>"Look, we're doing this and 50-60 percent of the machines are still
>freezing up! Turn it on, get one result. Turn it off and next time
>you turn it on you get a different result. Six times, you'd get six
>different results."
>Harris: "Can you give me an example of different results?"
>Rob: "Meaning the machine does something wrong different each time
>you boot it up. One time and it would freeze on you, next time it
>would load the GEMS program but have a completely different type of
>error, like there'd be a gray box sitting in the middle of it, or
>you couldn't use a field."
>Harris: "Was this all due to the clock?"
>Rob: "I don't know for sure. They [the machines] were not originally
>doing it. Then they fixed the real time clock, and it was supposed
>to make it work normal. It fixed the clock problem -- the clock
>problem had caused it to come up and not show the battery at one
>point. It was supposed to say either 'low battery,' 'high battery'
>or 'charging.' But when the real time clock was messed up, you'd
>boot the machine and it would say 'No battery!' I mean, you don't
>have the machine plugged in, you boot it up, and it starts, and says
>it 'has no battery.' That's like saying, 'this morning I got out of
>bed and I stood up and I had no brain.'
>"And that's how they ended up finding it, the problem. What it was
>doing was it was checking for the right time, and kept going back
>trying to get a better time, and while it was doing that, it was
>supposed to get the battery status but it was still busy trying to
>get the time.
>"And then when we loaded the software to fix that, the machines were
>still acting RIDICULOUS!
>"I was saying, 'This is not good! We need some people that know what
>this stuff is supposed to do, from McKinney, NOW! These machines,
>nobody knows what they're doing but Diebold, you need some people to
>fix them that know what's going on! They finally brought in guys,
>they ended up bringing in about 4 people.
>"When they left, they still did not know why it was still sporadic.
>My understanding is, after I was dismissed, they came back the
>following week. That's when they figured out what the real problem
>was. But they'd already had us do their 'upgrade' on thousands of
>machines by then."
>Harris: "How did this work? Did Dr. Brit Williams get the machines
>first and do acceptance testing, or did you guys get them first?"
>Rob: "When the machines came in, they came to us first. They were in
>the warehouse. We assembled them. They'd come in a box with a
>touchscreen, and another box with the booth. We assembled the
>machine and we ran it though series of tests. We'd check the power
>cord, boot up the machine, check the printer, bar code it, update
>Windows CE, then send it on to Brit. He did the KSU testing the L&A
>[Logic & Accuracy] was done at the county level, right before the
>Harris: "So…the L&A was not done at acceptance testing?"
>Rob: "It got so there wasn't time. They did it before the election."
>Harris: "How long does it take to do a Logic & Accuracy test?
>Doesn't it take like, 15 minutes per machine?"
>Rob: "When we did the updates in Dekalb, they kept saying it would
>take a really long time. But they don't think about the different
>overlapping things. You can update a bunch of machines
>simultaneously. Same thing with an L&A test. You have a whole group
>of cards, they have to touch every machine. What we had done before,
>we had 10 material handlers throw the machines up there, use the key
>to open it up, stick 10 cards in, boot 'em all up which installs the
>Harris: "But what about the L&A testing?"
>Rob: "The L&A testing -- You would just enter, like, one vote and --
>you just choose one -- you don't need to be specific on which one.
>When they did this L&A testing, that's when they did the FINAL
>update to the software."
>... Harris: "So the touchscreens came and had to be assembled?"
>Rob: "Of course you have to have the touchscreens assembled in the
>warehouse, and do some testing. It turned out that there were a lot
>of problems that needed to be dealt with, and they simply weren't
>dealing with them."
>Harris: "How long did you work there?"
>Rob: "They let me go only one month into it. The Project Manager let
>me go. He didn't like my management style. I'm very matter of fact.
>If this is wrong, fix it. I'm a simple person -- if something is
>broke, do you stand around and talk about why its broke for a month,
>or do you solve the problem?"
>Harris: "After your experience with Diebold, how confident are you
>that the machines count votes accurately?"
>Rob: "If you were to ask me to tell you how accurate I thought the
>vote count was, I'd have to say 'no comment' because after what I
>saw, I have an inherent distrust of the machines.
>"I was absolutely astounded that they functioned at all in the
>election. Here's me, I'm at the polling place looking around,
>waiting for someone to get frustrated...
>"I took this because of James, who is my friend, and because I'm
>A-plus certified. But when I came in there was a bunch of internal
>bickering. They had no inventory control in the warehouse. I
>guarantee you that the state of Georgia can't accurately reflect
>where each machine is.
>"Diebold was impressed with what I accomplished, and asked me if I
>was available for some other states they'd be doing...
>"The problem, what they were doing with the inventory on the machine
>was this: Inside the case is the serial number. They would hand
>write the serial number on a post-it, stick it to the front of the
>machine, and there would be a sheet hand-written from that list.
>Now, you've got 20 machines sitting on a pallet. The guy making the
>list would look at the post-its and he'd record all the post-it
>numbers on a list. Look, if you're writing numbers by hand, twice,
>by two different people, there is a real good chance you'll
>transpose some numbers.
>Then, they used the list for bar codes, but I would say probably
>1-2% of the machines are incorrectly bar coded. They couldn't track
>them in the Access database, because they'd punch in and it would
>say 'that number's already been used.' Then they'd check the
>machines, and they had the right number, so the wrong bar code was
>sitting on some machine that had already been shipped out to the
>"Ironically, they would send a spreadsheet of all the numbers of the
>machines that they shipped straight from the factory. This was from
>the same computer that generated the labels. They had copies of it
>all along. I said, 'Hey guys, if you check these when they come in
>the door you'll never miss a label.'"
>"I was very down on Diebold, because they were very sluggish and
>didn't move well. I worked there from mid-june to mid-july. The
>whole time they were upgrading the software and doing some sort of
>fix to it. This was supposed to be prior to KSU testing."
>Harris: "What about the program patches begun in August?"
>Rob: "Aug 20, they started to put these teams together and go out
>and update the machines. You have to understand that the patching
>all started when I did the first quality check that Saturday. They'd
>never have done it. They had shipped us 6,000 machines and NO ONE
>had ever done a quality check. I'd come in on a Saturday, I had two
>of my sons with me, and I thought I'm going to just look. And it was
>"Then first thing Monday morning I raised the question, I said, 'Hey
>guys, we've got a problem -- there's 20-25% of the machines that are
>palletized that are failing, and then they had a new update come out
>and I was doing an update, and then they sent a new one. I updated a
>whole bunch of machines. Then they finished about the time I left.
>But later they put in another one, I guess. In August.
>"You've gotta go take care of this JS [junk shit] equipment, I told
>them. Finally, I raised it as high as you go, I raised it to Bob
>Urosevich, he's the head of it. I told him personally, 'This is bad,
>I don't see us putting an election on with these machines!'
>"That's where they finally assembeld the teams. They got some big
>ol' vans we loaded up as many people as could fit in.
>"They were actually swapping parts out of these machines that were
>on site. They'd cannibalize a machine with a bad printer or
>whatever, they'd grab the screen off of that to put on another
>machine with a failing screen, they'd retest it. They were not just
>breaking them down, they were taking pieces off and putting it back
>"Even the machines that are updated, that had the right release of
>the software, exactly like the company wanted it, you'd boot it up
>and all kinds of crazy things would happen. That led to my belief
>that when voting took place, there would be problems."
>Harris: "Do you remember what release number it was? What version of GEMS?"
>Rob: "Release -- I don't remember the number because what they did
>was it was always the date. I had to take it to the level of these
>testers, they knew that the machine either did pass the test or
>didn't. We'd check the date to make sure it was the right version.
>"The date was…let me see…June 28. No, the last one, the date that
>was supposed to be on there was July 5. (Note: a patch labeled
>Georgia062802.zip is on the ftp site, and when you review it, you
>will see that it contains much more than just the "Windows updates"
>claimed by Georgia officials.)
>"There was about three updates, the CE software, the date that would
>come up would be the last. After that they came up with another fix,
>that's the August one at that point.
>"I told Darryl Graves, the Project Manager, I told everyone at
>Diebold, 'I have zero confidence in the ability of these machines to
>Harris: "I understand that they go through Wyle testing labs and so
>forth. How in the world do so many critical errors get through
>Rob: "When I was handling these machines, they were coming straight
>from a factory in North Carolina. That's where the actual touch
>screen was manufactured. Booths came out of California. We assembled
>the booth with the machine. That's all I know."
>Harris: "What do you know about the ROM chip, or whatever?"
>Rob: "There's the eprom, or the flash as they call it. A lot of the
>fixes they did they could do in the flash memory.
>"If they said they tested it I'm going to tell you right now the
>software that I installed on the machine myself, they found out that
>that was NEVER tested. Okay, I don't want to get other people
>involved, but you should talk to Rellinger.
>"Anyway, that they had never tested it, that made complete sense to
>me, watching what was going on.
>"This is an example we did: We would plug it in, boot it 3 times,
>unplug it, boot it three more times. I wrote a sheet on this. This
>guy came in from McKinney, he was about the second in command. He's
>a good friend of Bob Urosevich. About second to Bob, at least now,
>he got a promotion. Greg? Something like that. He flew in and I went
>to Dekalb and I tested and together we went through, and we wrote
>down every single error, and he booted them himself, and was looking
>at the results and seeing how sporadic they were. and we found out
>of the machines we tested, about 75% of the machines had different
>sporadic things. He was working with me and we were writing them
>down, we literally wrote everything down."
>Harris: "Do you have a copy of that?"
>Rob: "I don't think I have it. I have some email. I'd have to look.
>I know we came back and he copied it and he -- Greg Lowe (spelling?)
>is his name. I drove him out there. Brit was there, KSU was doing
>their testing. They were bombing these machines out left and right."
>"I'm telling him, 'They're all like this.' At this time I was
>working 150 hours in 2 weeks I was there all the time with these
>machines, that's the reality of it. The techs were working overtime
>trying to fix them. We couldn't get enough from the factory because
>so many were bad. You'd get a shipment of 300, but 75 were bad, they
>couldn't put them out fast enough to replace all the defects.
>"It was the software, not the hardware, that's where the problem was.
>"If they're telling you they tested that, well they did NOT test the
>fixes that they did to the windows CE software.
>Harris: "Do you know who was writing the fixes?"
>Rob: "He had a weird name. He came out of Canada."
>Harris: "Guy Lancaster? Josh …Talbot Iredale?"
>Rob: "That's it! Talbot Iredale would actually fix it and say, 'Oh,
>here's the problem,' and stick it on the FTP site we'd grab it stick
>it on the card and make a bunch of copies and use it." (NOTE: You'll
>see the initials "tri" in the source code files. Talbot R. Iredale
>is one of the main programmers, and has been a stockholder.)
>Harris: "So you took the patches right off the FTP site and
>installed them on the machines?"
>Rob: "That's what we did, he'd FTP it, and tell us to grab it, we'd
>put it on a laptop, copy it and when you boot the machine -- it's
>just like a computer that looks at the "A" drive -- these machines
>look at the card and then erase the flash, reprogram with whatever
>they said needed to be fixed -- I say, erase it and reprogram it
>with crap -- and then the whole thing would start all over again.
>"My understanding was that they figured out what was conflicting and
>James told me that Tab, well the team that came out after I left,
>they figured out what was going on, they figured out that when they
>fixed the real time clock problem they had never tested their fix.
>"The only people that that cost was Diebold, who had to pay all
>kinds of extra expenses. The rumor around the office was that
>Diebold lost maybe $10 million on the Georgia thing. I mean, they
>only sold the machines for what, $2,000, or $2,500, and then you
>have to build them and then you're paying people $30 an hour and you
>are out touching 22,000 machines FOUR TIMES -- there's no way they
>didn't lose money on this deal.
>Rob: "You know one of the main things that really just made me so
>upset, they were just like, 'This Brit guy, don't even speak to him,
>it's a political game, you've gotta play the politics.' Well, he
>walks in and says 'What are you guys doing?'
>I said, 'We're putting in an update.' He said, 'Will it change what
>it does?' We said, 'Just do your normal test, we're supposed to get
>the machines ready for you.'
>He tells someone at the office and they freaked out. They were like,
>'What the heck are you doing???'
>"I wasn't supposed to talk to him at all, I guess. The guy had a
>flannel shirt on, he was kicking it and he was very genuine and open
>and there we are in the same room together, but because I actually
>spoke to him I got reprimanded. They said, 'If they ask you any
>question, you gotta say 'Talk to Norma, to one of us.''
>"And then you know, ironically, later on right before I exited, they
>were scrambling for a date, they were trying to get us, the teams,
>into Fulton County to do Fulton County's 1,900 machines.
>"They were in the most horrific spot. The place they warehoused them
>was like 1900 machines in a little office space, there was no way we
>could get at them. The machines are like 58 pounds, and they had to
>bring them in unstack them off the pallet, restack on the pallet,
>talk about labor, talk about wasted money! It's like a warehouse and
>offices off 75, in Atlanta, I'm talking to this guy he's a great
>guy, he's from Fulton County. Him and I were scheduling this,
>figuring it out how to get to these machines and do the update
>before KSU has to test them. We cannot be doing this at same time as
>KSU because there was NO ROOM for that.
>Brit had been down there, he knew this. I'm talking to the Fulton
>County guy. He opens this one last door and here's this huge giant
>empty warehouse. Why didn't they put the machines out here?
>He says, 'Well you see over there's these boxes of county material,
>you can't be out here because there may be some sensitive stuff in
>these files. They don't want anybody near 'em. His name was Barney,
>the only Barney I've met who's black. He said, "Yeah, they were
>talking about putting a fence out here."
>"We could just get all the testing done at once, I thought.
>Whatever. Maybe someone could just get a security guard to watch us
>and make sure we don't get into the boxes. I go back to the office.
>Brit was there, and he says 'What's it look like for Fulton?'
>I said 'There's no way were going to able to get to Fulton County by
>Thursday.' I said we could probably be out there by Friday or
>Saturday. He said 'There's no way we can do it at the same time, you
>know that.'
>Rob: "I think a lot of the problems they had ---- I've worked in
>billing software, and it's common to have this little thing wrong --
>a simple little hardware change, you have to put some little line of
>code in Windows CE to make it work better. But the thing that blew
>me away was when I'm told me they'd NEVER TESTED THE FIX.
>"They produced it and got it to us in 24-48 hours. If I'd known they
>hadn't tested it I simply wouldn't have installed it! My background
>tells me that's a no-no.
>"I went into this Diebold thing with no real knowledge of the voting
>industry. When I left, I not only had a complete grasp, but I had a
>complete disrespect for these machines.
>"And with the folks in the office who were so -- you know, 'I'm the
>political person, you have to know how the system works' -- they
>were so much more concerned about their own self importance, they
>"Because that's what the people in Georgia need. And I'm one of them!"
>Harris: "Who are some of the names working in that office?"
>Rob: "Norma Lyons and Wes Craven -- they're from Diebold, and Keith
>Long. Norma and Wes live in George, Keith was in Maryland before,
>then here, I think.
>"They sat in the weekly meetings on Monday. Norma had been a county
>worker doing voting for 10 years. She knew all these people in
>several counties. She was the liason between Diebold and the
>counties. They [Diebold] would tell you something important, and she
>may or may not tell you because she wouldn't know how important it
>"Wes was the kind of guy who needs to work for Sprint or a big company..."
>Harris: "How secure were the machines, from what you saw?"
>Rob: "I'll tell you something else -- we didn't have badges --
>people could just walk right in and get to the machines."
>Harris: "And that FTP site, anybody could walk right into it also.
>Even Diebold's competitors."
>Rob: "Anybody who's in voting, you leave one company you go over
>there. Ooh yeah, we'll take you on. Someone comes in and says, 'By
>the way, I uploaded the source code, want to grab it?'"
>Harris: "Were there any protections to keep you from duplicating
>memory cards, or to have them serial numbered or whatever?"
>Rob: "The memory cards, you can just duplicate them. You have to
>have the proper info on the card, for the machine to boot up, but
>you can just make copies of the cards."
>Harris: "Were there any passwords on those FTP files?"
>Rob: "No."
>Harris: "Any passwords on the files themselves? Or the site?"
>Rob: "What we got never had passwords. You just pick it up and use it."
>Harris: "Do you still have any records?"
>Rob: "Emails. And James downloaded to his personal laptop, it's
>probably still on his. And probably still on mine too. Diebold
>didn't provide us with anything with a PCMCIA slot so we had to use
>our own laptops to transfer the files when they told us to.
>Harris: "When I asked Diebold if there was anyone named Rob in
>Georgia, they said no. Did they know about you?"
>Rob: "They knew me and they knew me well. I met Bob Urosevich a
>couple different times, and Ian, and then Greg Lowe, he got promoted
>to like almost the DFO, he was basically Bob's right hand man."
>... "If you would have realized the scolding I got for actually
>speaking to Brit. The whole quality control issue, I kept having to
>remind them, I'm the one that pointed this out -- we want this to be
>right -- my goal is to just get it fixed and move on.
>... Harris: "Do you think anybody could have tampered with a
>machine, if they wanted to?"
>Rob: "Well, when we did the quality control check we'd open it up,
>they have a little box for the printer. We would find the key still
>in the printer. Someone could literally take that. We found cards
>left in the machine. I wondered what would happen if the wrong
>person got it."
>Harris: "I understand they did a big demonstration during the
>summer, with the machines."
>Rob: "I was there when they told me I needed 1100 machines for a
>demo. I thought, 'The trick is coming up with 1100 machines that
>actually work.
># # # # #

Arthur M. Keller, Ph.D., 3881 Corina Way, Palo Alto, CA  94303-4507
tel +1(650)424-0202, fax +1(650)424-0424
= The content of this message, with the exception of any external 
= quotations under fair use, are released to the Public Domain    
Received on Sun Aug 31 23:17:03 2003

This archive was generated by hypermail 2.1.8 : Sun Aug 31 2003 - 23:17:17 CDT