Navigation
Topics
Log In
Log Out
:
Special Search
New Today
New This Week
Advanced Search
Tree View
Your Account
Edit Profile
Register
Forgot Password
Tools
Help/Instructions
Policies
...
|
| (TN) 4/12 - SHELBY COUNTY PERFORMED U... |
|
| Author |
Message |
   
Bev Harris Board Administrator Username: Admin
Post Number: 11591 Registered: 12-2004
Best of Black Box?  Votes: 11 (A keeper?) | | Posted on Thursday, April 19, 2012 - 6:25 pm: |
|
By Bev Harris SUMMARY: Details on Shelby County, Tennessee alteration of its Diebold voting system. They inserted a completely different program (not from Diebold at all) which alters the way the system works. Documents show that fully half of White suburbs are now processing information in a different way than the rest of the county, and that this selective and unusual processing has been taking place for at least four years. The alterations made to the system are illegal almost everywhere in the USA and automatically render their $600,000 voting system decertified, although they will claim that they don't need to use a certified system. One man, who does not work for Shelby County, has been given almost unprecedented access to the system. The alterations allow them to customize how party affiliation is coded, creating different party identifiers in White, Black, and mixed counties. I have posted a few documents, like the work order, in the article and will post several more, including a little bit of related source code, this weekend. SHELBY COUNTY'S VOTING SYSTEM ALTERATIONS MAKE IT UNCERTIFIED AND UNAUTHORIZED UNDER ANY FEDERAL STANDARDS In May 2007, Shelby County executed a work order to alter the way Shelby County's voting system works. Computerized voting systems are tested and examined to (supposedly) show that they function to record and count votes correctly. Federal approval for these systems is based on all the working parts. You can't change anything and still call it a certified system. The whole certification thing is pretty bogus to begin with, but at least it means that SOMEONE has, at least theoretically, examined the voting system. According to systems developer John Washburn, "Both NASED and the EAC consider the voting system to be the specific constellation of software and hardware. There is a system number which is a particular constellation of hardware and software at specific revision levels. If you make a change or take a piece out or put a piece in that system is no longer certified." This was not a minor modification. Shelby County not only altered its voting system in a major way, but documents I have obtained reveal that it processes information differently in some precincts than others, and this is almost certainly related to the change they made in how their system works. Fully 50% of White suburban precincts are affected. WHEN IS THERE EVER AN ELECTION WHERE THE SAME POLITICAL PARTIES WOULD NEED TO HAVE DIFFERENT PARTY IDS IN DIFFERENT PRECINCTS? The alteration to Shelby County's voting system is especially troubling because it enables precinct-specific political party codes. Let me explain, and I'm sure you will see how inappropriate this is: The Diebold/ES&S/Dominion "GEMS" system assigns a number to each political party. For example, Democrat = "1", Republican = "2", Nonpartisan = "3". The system then knows that a ballot or candidate tagged "1" will be a Democrat, "2" a Republican, and so forth. Imagine, then, the fun you could have if you could tell the system, "but in THIS precinct, let's switch those numbers so "1" is Republican and "2" is Democrat. These party ID numbers are never seen by the voter or by any poll worker. They are internal ID numbers which tell the computer how to identify each political party. In the Diebold GEMS system, the county IT guy sets this ID number just once. It is a global value, and is automatically the same for every precinct in the county. In fact, let me just show you what the party ID table looks like for the whole county: Democrat - 1 Republican - 2 Nonpartisan - 3 See how simple this is? Imagine now, that you take a completely separate program that didn't even come with the Diebold system, and was written by someone else, and instead had it do this: Precinct 1 - Democrat = (assign your own number) Precinct 1 - Republican = (assign your own number) Precinct 1 - Nonpartisan = (assign your own number) Precinct 2 - Democrat = (assign your own number) Precinct 2 - Republican = (assign your own number) Precinct 2 - Nonpartisan = (assign your own number) That's what Shelby County has done. Why would there ever be an election where you use a different identifier for Democrats in one precinct than another within the same county and in the same election? Shelby County has hundreds of precincts, and has altered their system to allow each precinct and sub-precinct (called "splits") to assign party identifiers that can vary by precinct. But more to the point, it doesn't matter why they say they did it. "All really good pieces of malware do what they claim they're going to do, ... eventually." FUN WITH PRECINCT DEMOGRAPHICS Shelby County has especially racially-charged precinct demographics. Many precincts are almost all Black. Some precincts are almost all White. A handful of precincts are racially balanced. This alteration in Shelby County's voting system allows party identifiers to be set one way in White precincts, a different way in Black precincts, and yet another way in mixed precincts. Why would that ever be good, or even allowed? Answer: It's not. This manuever automatically decertifies Shelby County's $600,000 voting system. Shelby County also has an unusually strong correlation between Black votes (96% Democrat) and White votes (over 90% Republican in precincts outside of the city of Memphis). With the GEMS system intact and its countywide party ID code, to create a plausible vote flip (or even to realistically shave a percentage of votes), in Shelby County, due to the polarized racial and political demographics, you'd get caught. It might help immensely to be able to customize internal Party IDs by precinct. But GEMS won't let you do that. But now, the alteration enables it. THE WORK ORDER Richard Holden was on the Election Commission when the work order was put through, and is currently the chief elections administrator for Shelby County. The work order, attached in the next post below, was executed with Cathi Smothers representing Diebold. Based on memos between Shelby County elections IT guy Dennis Boyce and the programmer for this, the work order appears to have been carried out by Curt Wolfe, who lives in Fayetteville, Arkansas. Unbeknownst to Shelby County's 600,000 voters, one man, who doesn't even work in the elections division, has almost unprecedented access over what is now an uncertified voting system. Not since Jeffrey Dean had his way with King County's voting computers have I seen such a high level of access by one outside contractor. Curt Wolfe may be a lovely man and honest as a preacher, but what Shelby County has done to order up these changes is inexcusable. MIDDLEMAN PROGRAM The alteration to Shelby County's system inserts an entirely different program called ESM, originally written by Accenture and later customized to act as a go-between amidst GEMS and the voting machines. The system normally works like this: GEMS -> VC Programmer -> Voting machine - GEMS is used to configure the election, assigning candidate names, Party IDs, Precinct Identifiers and so forth. - GEMS sends information to VC Programmer, which creates a voter card. - You put the card created by VC Programmer into the touchscreen voting machine. The card tells the voting machine which ballot to show on the screen and you cast your vote. SHELBY COUNTY'S PROGRAM NOW WORKS LIKE THIS INSTEAD: GEMS -> ESM software created by someone else -> VC Programmer -> Voting machine - GEMS sets the party identifiers, precincts, candidates etc - ESM takes this information from GEMS, and in ESM the party identifiers are changed from one county-wide set to hundreds of precinct-specific party codes, which can be changed precinct by precinct by whoever has access to ESM - The information from ESM then goes to VC Programmer, which creates each voter card - VC Programmer's voter cards then provide instructions to the touchscreen WHAT ARE THE ACTUAL INSTRUCTIONS SENT BY VC PROGRAMMER TO EACH VOTING MACHINE FOR EACH VOTER? VC Programmer happens to be a component that's been in trouble. It has proven dangerous enough for some states (Florida) to disallow its use at all, and other states (California) to decertify the whole Diebold touchscreen system when Secretary of State Kevin Shelley caught Diebold making unauthorized program changes to it. One internal memo written by Diebold programmers frets that "If people find out" what they are doing with VC Programmer, "We'll find ourselves on the front page of USA Today." VC Programmer was blamed for directly disenfranchising over 40,000 California voters in California's 2004 presidential primary. Diebold's attorneys, from the Jones Day law firm, submitted a perjured interrogatory to the California secretary of state claiming that VC Programmer was not programmed by Diebold, and was "commercial off the shelf software." Imagine the panic when Sec. State Kelley paid a surprise visit to the plant where technicians were installing custom-coded VC Programmer patches. Diebold hastily sent all the technicians home, but one of them, James Dunn, showed up with a signed affidavit at a hearing and it was all over. Shelley decertified the system and banned its use in California in 2004. Therefore, a missing but probably very interesting component to what has really been going on in Shelby County is this: What instructions did VC Programmer actually send to the touchscreens after being visited by its customized ESM middleman? Whatever it is they say all this is doing, and whatever Tennessee says is allowed (if the state has even been notified of this high-risk alteration at all!), this customization is full-out illegal just about anywhere in the USA. WHAT'S UP WITH THE WHITE PRECINCTS? THE SYSTEM IS PROCESSING THEM DIFFERENTLY. Documents I have obtained show that the list of "who voted" and what party ballot they chose in the primary is processed differently in fully 50% of the White precincts in Shelby County's suburbs. After GEMS configures the election and sends it to Accenture's Wolfe-customized ESM software, which then sends customized precinct-specific party ID codes to VC programmer, which creates a voter card inserted into the voting machine at the time of voting, the information is relayed back to the ESM program, which then produces the list of "who voted" and (in primary elections) which party each voter chose. So here comes an interesting anomaly: In the 2008 presidential primary, and in the 2010 primary/general, half the White suburban precincts reported voters two or three times. But only in these outlying precincts, almost all of them White, and never in the city of Memphis. (Come again?) For example, in Memphis precincts you get a report like this: Participating voter list Mary Brown Alex Johnson Shane Rodman But in half of the White outlying precincts, you get this: Martha Smith Tom Williams Brad Nelson Martha Smith Tom Williams Brad Nelson If the election is a presidential primary, where Tennessee has only two party options, each name is doubled. If the election is a primary/general, which has three options (Democrat, Republican and nonpartisan), each name is tripled. A Microsoft Access file I have from August 2010 shows that IT manager Dennis Boyce then performs his own customized filtering process to weed out the extras before publishing the Participating voter list. At least Dennis Boyce and Richard Holden have clearly known about this issue for years, though they apparently have not briefed the Election Commission. In 2009, ES&S (now servicing Shelby's Diebold contract) informed them that no, they could not customize a whole new electronic pollbook to weed out these duplicates for them. I hope you're not thoroughly bored by now, because the point here is that the list of who voted, and which primary party their vote got attributed to, is one of the most crucial records in the whole election. It's a record you need to fudge if you are cheating and a record that has to match up to the number of votes you report. What produces these duplicates and triplicates, and why does this keep happening over and over, and why always the same demographics (identical demographics, but not always the same precincts)? THE PLOT THICKENS In August 2010, an election controversy erupted in Shelby County. An informant surfaced, referred to in community meetings as "Deep Throat." He had worked in the elections division, and he reported that certain electronic pollbooks, in certain outlying precincts, were being treated differently than all the other electronic pollbooks. They were "corrupted" or something, he said, and workers were told not to load them with the same data all the other epollbooks were receiving. This is especially interesting because in Shelby County, electronic pollbooks are only used on Election Day. Early voting goes through the customized ESM system and VC Programmer. But on Election Day, the voter card is produced by the epollbook and the list of who voted, and which primary ballot was chosen, comes from the epollbook. So here we have a DIFFERENT system used on Election Day, the ExpressPoll electronic pollbook system, again being reported as treated differently but just in some precincts, which happen to be a set of outlying mostly White precincts. SOURCE CODE AND OTHER STUFF I mentioned earlier that I will be releasing some files, including some of the source code related to this issue from Curt Wolfe. I'll poke around among those over the weekend and upload a bunch for you. (watch for them at http://www.blackboxvoting.org) ESM SYSTEM ALSO USED IN WISCONSIN By the way, the ESM Accenture tables (before Wolfe's customization) probably apply to other states, like Wisconsin, as well. They had an ill-performing voter list system that got them in a lot of hot water, with some states like Colorado kicking them out and squabbling over their bill. Apparently they no longer service Shelby County either, and Shelby County executed a contract to buy their source code. More documents this weekend. I'm really quite pooped, and I'm sure you are too if you have finished reading all this! PERMISSION TO EXCERPT OR REPRINT GRANTED, WITH ATTRIBUTION / LINK TO HTTP://WWW.BLACKBOXVOTING.ORG
|
   
Bev Harris Board Administrator Username: Admin
Post Number: 11592 Registered: 12-2004
Best of Black Box? N/A Votes: 0 (A keeper?) | | Posted on Thursday, April 19, 2012 - 7:08 pm: |
|
Work order:
|
   
Bev Harris Board Administrator Username: Admin
Post Number: 11593 Registered: 12-2004
Best of Black Box? N/A Votes: 0 (A keeper?) | | Posted on Thursday, April 19, 2012 - 7:11 pm: |
|
More:
|
   
John Washburn Voting Rights Forum Participant Username: Johnwashburn
Post Number: 3 Registered: 6-2011
Best of Black Box? N/A Votes: 0 (A keeper?) | | Posted on Thursday, April 26, 2012 - 3:37 pm: |
|
RE: ESM SYSTEM ALSO USED IN WISCONSIN I wish I could confirm or deny that statement, Bev. But when I asked for the Entity RelationshipDiagrams to the Wisconsin Statewide Voter Registration System (SVRS), I was told that such information is exempt from disclosure as an open record because the ERD is a trade secret owned by Accenture. More evidence that with Electronic Voting: The Truth is Trade Secret. I will upload the denial letter from the GAB for the ERD and post a link. |
|
|