11-3-05: The first public examination...

Author Message   BBV Admin Board Administrator Username: Admin Post Number: 2610 Registered: 12-2004 Best of Black Box?  Votes: 2 (A keeper?) Posted on Thursday, November 3, 2005 - 4:36 pm:        San Joaquin County - Nov. 2, 2005: We now have the first public software inventory of a GEMS server. We also have established a precedent that adds to the public oversight options for all citizens. Careful review of your state's public oversight laws for elections, along with a bit of creative thinking, can empower you in unexpected ways. Calif. Election Code 15004 is part of a package of election observation laws. This law happens to be “party centric” in that it gives special rights to political parties rather than ordinary voters. It allows close inspection of voting machines and systems before and during elections. EC 15004: “The county central committee of each qualified political party may employ, and may have present at the central counting place or places, not more than two qualified data processing specialists or engineers to check and review the preparation and operation of the tabulating devices, their programming and testing, and have the specialists or engineers in attendance at any or all phases of the election.” Black Box Voting objects to the idea of special observation rights for political parties. We believe this to be an equal protection violation. That aside, before suggesting reforms we decided to explore what can be accomplished with the current law. The California Libertarian Party collaborated with Black Box Voting to review six counties. The Peace and Freedom Party is collaborating in another county, and in future inspections we may work with any other political parties (major or minor) to open up the “black box” election process. We have a good foundation for future collaborations with any interested political party in any state where there are “party centric” election observation laws. San Joaquin County 15004 Examination – Nov. 2, 2005 County Registrar Deborah Hench suggested Nov. 2, 9 a.m. for the Libertarian Party's inspection. She said that the inspector would be required to provide photo identification. The day before the inspection, Hench was informed that the inspector for the Libertarian Party would be Jim March from Black Box Voting. The Libertarian Party agreed to pay one dollar for this service. On arriving, Hench asked March to prove that he was a “qualified data processing specialist or engineer.” He offered his curriculum vitae, showing 15 years as a data systems professional. Hench declined to accept it as evidence. After negotiations, including a phone call from the California secretary of state’s office, March was allowed to conduct the inspection. Voting System in San Joaquin County San Joaquin County is currently running a Diebold optical scan system, consisting of precinct-based optical scan machines, an adaptation of these for central absentee counting, and the GEMS software for ballot configuration and central tabulation. (San Joaquin ran Diebold touchscreens before they were decertified, and has switched – perhaps temporarily – to optical scan machines.) What we obtained in the inspection: For the first time, we have a detailed set of software inventory printouts for the central server running the Diebold GEMS program, and as it turns out, many other programs. Thanks to the Libertarian Party, we also have the first precedent regarding EC15004 – it can be used to scrutinize voting systems to a degree otherwise impossible. 15004 is not perfect, and its limits are still not clear. (Another organization is suggesting that “examine the programming” means “look at the source code,” a credible position but one that the California Secretary of State’s office may not at this time share.) What we didn’t receive: San Joaquin County refused to release any GEMS database file (extensions .MDB/.GBF). Diebold has written a letter to all of their California customer counties, declaring these files to contain confidential Diebold proprietary information. The letter does not, however, offer to pay any of the counties’ legal bills when they are sued under the California Public Records Act for production of these documents. Black Box Voting considers these to be public records, as does a court in Shelby County, Tennessee, which has already released a complete post-election file as part of discovery in an election challenge. In earlier letters to Jim March, trying to quash publication of certain files on his Web site, Diebold dropped its claims of proprietary privilege and copyright on its San Luis Obispo .gbf and .mdb files. Part One of Inspection Report - Hardware San Joaquin's GEMS server is a typical Dell double width file server (the size of two desktop mini-tower PCs, side by side). This was supplied by Diebold, with a Diebold-installed operating system build. It runs Windows 2000 Service Pack 3 and GEMS 1.18.19. It has an Intel Xeon 1.8gHz CPU. Data connections at the GEMS server are as follows: - One Ethernet port on the motherboard runs to a standard hub of about 16 ports, but only two of the 16 ports are in use right now. The other is used for a local printer connection. If they were still using Diebold touchscreens, the rest of the 16 hub port connections would be used for uploading memory cards from TS/TSx series terminals on election night and programming TS/TSx terminals pre-election, as observed by March in San Joaquin in March, 2004. - A second Ethernet port on the motherboard runs a single Cat5 wire to an Ethernet-to-Serial bridge made by Digi corporation. The serial ports thus added to the GEMS server run to a series of “central count optical scanners” used for absentee ballots. These are NOT the newer “high speed central count” scanners, but are the same hardware as precinct optical scans with three differences: 1. A small paper hopper capable of holding about a hundred sheets or so was added. 2. There is different firmware on the optical scan (ver. 2.0.12, not to be confused with ver. 2.0.12 of the Diebold High Speed Central Count optical scan machine, which is an entirely different system). 3. The memory card slot is not used. Instead, scans are read by serial port straight into the GEMS software and are not recorded on memory cards at all. Note that this configuration produces no poll tape for absentee votes, so any manipulation of the GEMS tabulator, like that demonstrated by Howard Dean/Bev Harris or the script hack done by Dr. Herbert Thompson, is unlikely to be caught in standard canvassing procedures. This puts approximately 50,000 absentee votes at a disadvantage in San Joaquin County. Barnes said that no other Ethernet connections are made or used, and that final results go onto the department’s Internet site “by sneakernet.” He appeared genuinely dismayed by the use of a firewall and router for uploading results in San Diego (July 26 mayoral primary). - On election night at 8 p.m. the modem bank will be plugged in. San Joaquin County plans to take unofficial results via that modem bank. - According to Barnes, the single phone number for that incoming bank (which has “rollover”) was programmed into the voting terminals by county staff, not Diebold, and is kept as high security and changed after each election. Barnes and Assistant Registrar of Voters Austin Erdman could not, however absolutely assure us that no Diebold staff has that phone number. - There is an Ethernet port from a PCI card originally provided by Dell. Barnes said that it is never used. There was nothing plugged into it at the time of the inspection. - A “Digiboard” plugged directly into a PCI slot used a very large connector to go to a bank of additional serial ports, which in turn drive the modem bank. At the time of the inspection, the power cord to the modem bank was removed; Barnes stated that it was completely unplugged “364 days out of the year.” - There are standard mouse/keyboard ports (round, not USB) and video connections. They also have a respectably large UPS. Part Two: Software – We are reviewing the findings and will publish this report after the rest of the 15004 examinations for the Nov. 8, 2005 election have been completed. ------------------------------------------ Additional notes: • Hench admitted that she did not have any written policies regarding production of credentials for a 15004 inspection. She was clearly making it up as she went along, and only a call from the California Secretary of State’s office (much appreciated), salvaged the situation. • We now have the first public software inventory of a GEMS server. We also have established a precedent that adds to the public oversight options for all citizens. Special security note on the memory cards: The results are uploaded first by modem. According to San Joaquin County Elections Systems Administrator Bill Barnes, modem results are unofficial results. Then the results are put into GEMS a second time, overwriting the modem results with official results, which come directly off the precinct memory cards. To do this, additional standard optical scan units are connected via the motherboard serial ports; these optical scanners are used as “card readers.” Black Box Voting has some concerns about the accounting and audit trail implications of uploading the same data twice, overwriting the original data, especially in view of the known tamperability of the memory card used to transfer results. On the other hand, the modem sessions aren't all that secure either. Each iteration of the data should be preserved in it's original form. One way to preserve the data would be to obtain and examine different "snapshots" of the .MDB/.GBF files from different time periods on election night and afterwards...however Diebold has instructed the counties not to release these.   James Zukowski Frequent Voting Rights Forum Participant Username: Jimz Post Number: 156 Registered: 12-2004 Best of Black Box?  Votes: 1 (A keeper?) Posted on Friday, November 4, 2005 - 9:29 pm:        Shouldn't the modem/card reader data be recorded as transactions in the log file? Or is that too much to ask, as it would create a very large file? Then again, with the low price of disk space today, it shouldn't really matter... Peace! James Zukowski The people who cast the votes decide nothing. The people who count the votes decide everything.   BBV Admin Board Administrator Username: Admin Post Number: 2628 Registered: 12-2004 Best of Black Box?  Votes: 1 (A keeper?) Posted on Saturday, November 5, 2005 - 7:09 am:        The log files record the time the data was uploaded and the identity of where it was uploaded from, but contain nothing about the data itself. Diebold either has no concept of accounting or designed it to be easy to embezzle votes. For example, their optical scan system preserves no record of each individual vote. Even at the precinct-based scan level it only retains sums of votes, and those are only retained on the removable replacable memory card. At the central count optical scan (mail-in ballots AND recounts) they don't even have the memory card record of the sums of votes. All you get is what is electronically dumped into GEMS. And GEMS can be tweaked in less than a minute. (An exception is the HSCC - High Speed Central Count Digital Scanner, which retains an individual image of each ballot, offering more protection for the safety of each vote.)   Jim March Voting Rights Forum Participant Username: Jimmarch Post Number: 57 Registered: 01-2005 Best of Black Box? N/A Votes: 0 (A keeper?) Posted on Saturday, November 5, 2005 - 11:13 pm:        The San Joaquin election officials said that they had seen a demo of the high speed central count at a trade show and that it had never been tested or installed at their offices. They said that the HSCC has a habit of going out of calibration more often that the standard central count device. I don't know what they based that on (where the info came from) but I see no reason for them to lie on this. What's puzzling is, with a bitmapped scan processed by software, a calibration problem SHOULD be automatically corrected for, right? What I mean is, the software should look at the positioning marks first and base everything on that...if the paper is "off center" or otherwise not feeding quite right, who cares as long as it all gets scanned. Otherwise it's simply bad software design...this being DIEBOLD we're talking about that's no great surprise of course. I'd bet the problem occurs if it's "rotated" slightly, we've all seen faxes/scans that look like that and since the human eye doesn't mind we just live with it versus re-scanning/faxing. Good software could compensate for rotation same as our eyeballs do. We REALLY want to see those ballot image scans, that's for sure. I should also ask the San Joaquin people who gave them that impression...can't be Diebold unless some low-level tech caught a sudden case of honesty, more likely an election official from another jurisdiction over a trade show buffet table or something...   John Washburn Frequent Voting Rights Forum Participant Username: Johnwashburn Post Number: 236 Registered: 04-2005 Best of Black Box? N/A Votes: 0 (A keeper?) Posted on Sunday, November 6, 2005 - 6:41 am:        Rotation, stretching and compression are allf affine transformations. Given 3 non-colinear points (e.g. registration marks in the corner) it is possible to map the distorted image to an undistorted, perfectly oriented image. I wrote just such transformational code in 1985 for AutoCAD. And this was in FORTRAN and AutoLISP without the aid of the DSP software packages available today. Registration with a digital image should be a non-issue. In Liberty, John Washburn

All original content on this website is Copyright (c) 2008-2009 by Black Box Voting. All rights reserved.
Forums powered by Discus Professional - www.discusware.com.
Original site and logo design is by Andy Markley - art101.com.