Exit Poll Hacked

The media Edison Mitofsky National Election Pool data was manipulated during the election to bring it into conformity with the vote tallies. Efforts may be under way to eliminate or even outlaw the exit poll.

An Unofficial Guide to the Presidential Election 2004

Created Novmber 18, 2004

Last Updated December 16, 2004 (new info color coded) 

Exit Polls

The Presidential election has been criticized by the Nato Parliament and by Senator Richard G .Luger (Indiana Republican). The New York Times article explains:   "The victory ..., by a margin of nearly 3 percentage points, that was given in official results diverged sharply from a range of surveys of voters at polling places that gave the opposition as much as an 11-point lead."   The senator asserts  "A concerted and forceful program of election-day fraud and abuse was enacted with either the leadership or cooperation of governmental authorities."   Election Abuses  The election was in the Ukraine.

The first complete official exit poll data for all states and DC as of 4PM EST November 2, 2004 direct from Edison / Mitofsky has now surfaced in New Zealand. Exit Poll Data   These numbers differ from most previous partial leaked information available. 42 states show a shift toward Bush and only 9 states show a shift toward Kerry. We will use these official numbers to guide our analysis.

Kerry General Strategy

Kerry currently has 252 electoral votes and needs 272 to be elected. The only states where those 20 votes are available are states where the official exit polls show a Kerry win but Bush is currently leading in the vote counts. These states with their electoral votes are Florida 27, Ohio 20, Iowa 7, New Mexico 5, and Nevada 5. Contesting Iowa, New Mexico, and Nevada would give Kerry only 17 electoral votes. The deadline for requesting a recount in Iowa has passed. Kerry’s only options are Ohio and Florida. He must win one of these two states. The Berkeley study sited below concludes that as many as 260,000 votes may have been changed by e-voting DRE machines.  If true, these votes are probably lost forever and Florida is completely out of reach.  The results of likely court challenges in Florida are completely unknown.

Statistical Analysis by County

Enter the statisticians. The research question of interest is how to explain the big exit poll errors by looking at individual counties testing for a relationship between deviation from expected voting patterns and type of equipment used to count votes for that county.

First up, Ida Briggs, Michigan software designer and database developer. New Hampshire Study    Her study shows that up to 80% of the deviation is associated with counties using Diebold AccuVote optical scan voting systems. Ida convinced AirAmericaRadio host Randi Rhodes who convinced enough listeners to convince Ralph Nader to request a recount in New Hampshire. 

Ida Briggs has written a summary of manual recounts for 11 Counties.  She reports not much change which still leaves unanswered what explains the big difference between the exit polls and vote tallies? Ida NH Update The only people she finds fault with in this process are the Democrats.  Ralph Nader has until Tuesday Dec.7 to ask for additional precints.

From New Mexico a study of ES&S optical scanners shows deviations ranging from 32% to 44% for four counties. New Mexico Study

Florida featured dueling statisticians. Until Berkeley, the most revealing study was the  Florida Study   Diebold, ES&S, and Sequoia each supply equipment for counties highly correlated with the deviations.  This study did not differentiate between optical scan and DRE equipment.

A research team at UC Berkeley Survey Research Office  report that irregularities associated with electronic voting machines may have awarded 130,000 - 260,000 or more excess votes to Bush in Florida in the 2004 presidential election. Berkeley Study The study shows an unexplained discrepancy between votes for Bush in counties where electronic voting machines were used versus counties using traditional voting methods including optical scanners.  This is the first statistical study to directly implicate the DRE e-voting machines in voting anomalies.  This is troubling because if the touch screen units were software modified for the election the original votes could be lost forever. Even if the fraud is uncovered, the actual counts cannot be reconstructed.

Some more numbers from Florida in an interesting attempt to narrow the vote differential by voting machine model number.  William F. Faulkner Study This study includes the percentage of voters in Florida using each type of equipment. This study clearly implicates ES&S in some strange results.

Here are some raw numbers from Ohio by county showing 2000 against 2004 counts and giving the voting technology in use in each county. Ohio 2000-2004 At first blush, these numbers don't show the remarkable and obvious deviations from expectation of the New Hampshire, New Mexico, and Florida studies but a statistical analysis might shed light on a pattern.  It appears that Ohio counts only about 15% of its votes with optical scanners and far less than that with DREs.  It may not have been possible to hack these machines to achieve the necessary result without attracting big attention. Maybe Ohio had to rely on other and older means to steal the election, such as voting non-shows for Bush and double counting.

A study of North Carolina voting where exit polls predicted all statewide races except that for Senate and for President, and where the Senate race was probably stolen has now been done.  This study shows just how complex the rigging of an election can be.  To prevent loss of this study when DailyKos times it out, I have preserved a local copy. IgnatzMouse Study This study heavily implicates ES&S who supplied DRE, optical scan, and punch card tabulation equipment.  The author of this study looked for patterns of deviation between absentee plus early voting and voting on election day November 2.  The author tries to correlate these deviations by type of equipment used and also by vendor.  North Carolina is complex in demographics across the state and uses a complex mix of voting equipment.  This excellent study is an object lesson for those hoping for a simple smoking gun explanation of how elections are stolen.  The mechanisms used may be very complex and very subtle.

 Real World Evidence of Problems

Multiple states, big deviations, optical scanners and DRE machines from Diebold and ES&S. Is there a pattern here? Bev Harris in Volusia County Florida Bev Harris and her BlackBoxVoting group descended on Volusia county Florida on November 16, 2004, armed with FOIA requests. She now has retrieved from the trash, bags full of optical scanner tapes and other material showing mismatches with reported counts in the central tally machine. She has also secured the tally machine media and logs. Bev Harris in Volusia  Bev was in this very same Volusia County in 2000 and found similar problems. Resolving the problems would have required a full countywide recount that was stopped by the US Supreme Court. Bev Harris Volusia 2000

Bev Harris's plan is to hand count the optical ballots (no machine) for several precinct. A hand count is the only practical way to assure that the recount will not be tampered with. She expects to find big discrepancies in the counts as evidence of fraud.  Bev Harris Update Nov 19

Bev Harris reports that Volusia County resident Susan Pynchon, with the help of Volusia County attorney Daniel R. Vaughen, P.A., filed a lawsuit on Tuesday, Nov. 23, seeking to set aside the Nov. 2, 2004 Volusia County election due to irregularities.  Bev Harris Nov 23   Bev also notes the following counties have refused to be held accountable for the 2004 presidential election, by declining to produce basic audit documents until after all election contest periods have lapsed: Palm Beach County, Ft. Myers County, Pasco County, Highland County, Holmes County, Indian River County, Lee County, Levy County. Black Box Voting is requesting citizen audit groups to work with them to take these counties into full audit mode in December.

Black Box Voting filed suit in Palm Beach County Florida Nov. 30. for failing to provide public records under FOIA.  Now Palm Beach county says it wants money $7,000 to produce the records.  Michigan says it wants $125,000 to look at the records. Bev Harris Update Dec 3

The reason I have listed below the various models of central tabulators and their potential communications now becomes clear with the Volusia lawsuit.   This lawsuit indicates that the non secured central tabulator in question shows evidence of remote access during the critical period on election day.  The blog is a call for anyone to come forward who has information about the communications or network connections of any central tabulator in the country.

Bev Harris is also very critical of the Kerry campaign people.  They clearly don't get it. Kerry Blew It

Lawyers in Ohio have challenged the Ohio election. The challenge is based on a large collection of irregularities and evidence of fraud throughout the state. Ohio Lawyers to Challenge Election 

A recount is now taking place in Ohio.  Each county randomly selects three percent of the ballots to count by hand, then runs those selected ballots through a tabulating machine.  To prevent a total recount in the precinct, the manual counts must "match" the machine counts.  As you might imagine, the hackers are working overtime to make sure countires don't trigger the total recount.  This time Triad Voting Systems the largest vendor-maintainer of tabulating machines in Ohio appears to be scrambling to make sure the machine counts match the hand counts.   Professor of Computer Science Douglas Jones asserts that Triad's visits to local voting centers compromises the credibility and integrity of the Ohio recount. Jones Affidavit  Rep. John Conyers (D-Mich) is attempting to get the FBI to investigate. For a blow by blow account of the recount in Ohio see Cobb Press Releases  

The New Mexico election canvass board met November 23.  Kerry still trails Bush by 6000 votes but half the 12,000 provisional ballots were discarded.  NM Canvass Deadline Will a Presidential candidate (there were six) call for a recount? No.

The Nevada canvass certified the vote November 23. Justices Certify Vote

 Why Optical Scan Equipment?

Most fraud concerns prior to the election focused on the new touchscreen voting systems (DRE) which lack voter verifiable paper. These machines can be hacked in such a way that the results are altered without possibility of recounts (the perfect crime).  DRE machines are popular with voters and may have been placed in locations where the voters are heavily conservative or where unusual results might have attracted immediate attention. We have also learned that Ohio did not deploy 200 DRE machines at all contributing to heavy voting lines so maybe there were not enough DRE machines to adequately change the results.  What is now clear is that optical scan voter equipment seem implicated in some statistical anomalies across several states. The Berkeley study now indicates that DRE e-voting machines may be implicated as well in Florida. Illegal software in DRE machines without voter verifiable paper will forever destroy all evidence of the voter's intent.  The search for fraud must initially focus on those machines with paper evidence, the optical scan equipment.  It may be possible to prove that illegal software was used in DREs but the true vote counts will never be known.

Lets look at optical scan models from three suppliers; Diebold, ES&S, and Sequoia Pacific.  There are two primary types; precinct and central. With precinct tabulators, the v oter inserts ballot and receives immediate overvote/undervote notification.  These model are located at the actual voting place and are unlikely to be networked though they may be equipped with modems, either wired or wireless (Wi-Fi). Illegal software would likely be loaded into these models prior to their move to the precinct. With central tabulators, ballots are counted centrally, away from the voters. Therefore the voter receives no overvote/undervote notification. This model is more likely to be networked and are certainly modem accessible with the possibility that illegal software can be loaded remotely just prior to counting.  Further, these models are all PC based.

ES&S has two primary optical scan types.  The ES&S Model 100 (110?) precinct tabulator appears to be PC based and has two PCMCIA slots and RS232 ports for insertion or attachment of wired or wireless (Wi-Fi) modems.  The second is the PC based ES&S Model 150, 550 and 650 central tabulator.

The  Sequoia Pacific Eagle Precinct Optical Scan has an optional modem. The Sequoia Pacific 400C also known as the IV_C Central Optical Scan is a PC based unit that also serves as the vote tabulator machine and can be networked. 

The Diebold AccuVote OS is a precinct optical scan unit and is not networked or modem equipted.  Diebold's central tabulating unit is the GEMS which is PC based and does not include centralized optical scanning.  GEMS is also used to tabulate counts from the Diebold AccuVote TS DRE.

Does the official Ohio State site show real time (updated automatically) counting going on in Ohio? Ohio Election Real Time Counts   This is first evidence I have seen of a statewide network of vote tabulation machines.

Conspiracy to Steal a National Election?

Diebold Hacker 

Image (c)2004 Rand Careaga/salamander.eps

Diebold’s number one critic and enemy in the area of voter machines and potential for fraud was Athan Gibbs founder of TruVote, a competing voter machine producer. On March 12, 2004 Gibbs was killed in a road accident. Whistle-Blower Killed  A a photo and report of the collision is from the Tennessean. Collision Fatal His obituary appears in the Tennessean. Gibbs Obituary  

On July 27, 2004, Jeff Fisher Democratic candidate for congress in Florida’s 16th district was contacted by an individual claiming to be the MIS Director at Bay Point Schools. Jeff Fisher refers to this individual as XXXXX (here XXX) on his web site Jeff Fisher  XXX says he has uncovered a scheme to alter election results by inserting “kernals” of code into optical scan machines and vote tally machines. He says this method was first tested in 1999 and was used in the 2000 and 2002 elections in Florida and is now planned for use throughout the US in 2004. He claimed to have documents and to have given copies of those documents to someone who was killed (Gibbs?). XXX fears for his life and has gone into hiding. Fisher tried to interest various people in XXX’s story without success until the election surprises.

Jeff Fisher also refers to another individual who worked securing  Bay Point School's computer systems in 2003, Dr. Piotr Blass, who won't talk about what he knows.  Blass was a write in independent candidate for US Senate in Florida in 2004.

On November 15, 2004 at midnight the blog site dailykos broke the story. DailyKos Fisher  The bloggers commenting on this story discover that Bay Point Schools in fact exists  Bay Point Schools and is owned by Mel Sembler, Bush’s ambassador to Italy. A quick search on the Internet reveals that Sembler has a history going back to Ronald Reagan, Oliver North, Iran Contra, possible foreign election rigging, and funding Clinton scandal research. He is widely regarded as the likely source for the Niger yellow cake forged documents unmasked by Ambassador Wilson. The chairman of the board of Bay Point Schools is Joseph P. Klock Jr. Attorney for Katherine Harris before the US Supreme Court in 2000. This fascinating blog story was unlinked by 8AM and was locked down entirely a short time later.

On November 16,2004,   Keith Olbermann mentions and dismisses the Fisher story.   I didn't make myself clear on November 18.  The reason I believe Fisher's story is not that the school's owner and chairman have  Neo-Conservative Neocon ties. The reason I believe this story is that in July, before the election, XXX knew that a multiple state fraud using optical scan voting equipment as the weapon of choice would occur.  Everyone I have read prior to the election was obsessing about the DRE (touch-screen) and its lack of voter paper verifiable audit trail.  Our fraud masterminds don't care about the existence of paper ballots because those paper ballots would never be counted.  We observers were shocked on election night when the exit polls were wrong.  We were shocked again when the statistical studies by county pointed to the optical scan equipment in state after state.  Only XXX predicted this to my knowledge.

Practical Nationwide Implementation

XXX outlines a methodology consistent with the statistical anomalies we find and would explain why Bev Harris’s recovered optical scanner tapes don’t add. How would this methodology be implemented on a large enough scale to affect many states? Wired reported in Oct 2003 that, "A former worker in Diebold's Georgia warehouse says the company installed patches on its (DRE) machines before the state's 2002 gubernatorial election that were never certified by independent testing authorities or cleared with GA election officials.” Wired on Georgia 2002  Diebold itself admits to sending 360 professionals to Georgia for the 2002 election. Diebold Professionals in Georgia  Could these professionals have patched the AccuVote systems as well? Is this a practical method for nationwide deployment? Let’s say we want to patch systems in 17 states and it takes 360 professionals per state. That’s more than 6000 Diebold professionals showing up at various sites just before the election. And don’t forget that ES&S needs to do the same so we have another 6000 ES&S professionals running around. They also need to come back after the election and remove their patches. One would think this might attract some attention somewhere but maybe not.

Evidence of illegal software being loaded into ES&S optical scanners and tally machines can be seen in this article. ES&S Whisle Blower

 Now suppose many of these optical scan systems are connected into networks and that the manufactures have access to these networks. Then the patching can be done from a single site (say a school in Florida?). See Single Agent Hypothesis  for a theoretical look at what one person could do given access to an entire network of voting machines. Why would anyone try to change an election when a recount will discover the discrepancy and correct the count? The assumption must have been made that is the election is not close recounts won’t happen. Recounts in elections are rare events and the perpetrators must have felt they could be avoided. We may yet see court challenges to recounts as we get close to the modified counts.

The first principals of machine recounts are 1) all machines involved in the recount must be isolated from all communications.  2) an audit must be made of all software in each machine.  3) each machine must demonstrate the ability to correctly count test ballots. If any of these conditions cannot be assured the result of the recount is as unknown as the original count.  Bev Harris, and we hope Ralph Nader, are going to do hand recounts of the actual original paper ballots to eliminate entirely the possiblity that the machines have influenced the recounts.  The American people will believe nothing less at this point.

In January 2004, Bay Point Schools installed a new computer lab consisting of twenty PCs a network (Internet) server and networking equipment.  Bay Point School Computer Lab This lab in located in West campus at 25005 SW 133rd Avenue, Princeton (Miami), Florida 33032. The network communications rack shown has a patch panel with 48 plugs (probably 100baseT) indicating that up to 48 devices can be attached locally to this network.  XXX told Fisher that Mel Sembler was helping the school acquire equipment needed for the planned election work.  The baypointschools domain was create Feb 1, 2000 and is currently hosted on readyhosting.com servers, not from this lab's server. 

It would be interesting to learn the communications bandwidth of the server's external Internet connection and to find the number of POTS (plain old telephone service) lines at this West campus.  Bev Harris has recommended that election officials disable modems connected to their election systems but the vendors probably have prevailed to leave these modems active for last minute emergencies and bug fixes.  Since there are probably hundreds of these modem equipted systems nationwide, a significant modem bank would be needed to download the "kernals" to all these systems just prior to the start of counting.  Might this modem bank be located in this computer lab?  XXX has indicated the there may be other schools involved but I would opt for the assumption that a single site was used.  This limits risk and a single site is all you need.

The early reports from New Hampshire are that no significant counts differences are being found.  If these recounts involve the suspect software modified optical scan equipment, has this recount equipment been subjected to the above tests? This report appears to imply that hand recounts are being done.  Early New Hampshire Recount

Implications for This Election

The election will proceed in its natural course of counting the provisional ballots, counting the absentee and mail in ballots, and resolving discrepancies in count totals. Where necessary recounts will occur with different types of voting technology including punch cards. Total recounts of votes on DRE equipment will not be possible since they lack voter verifiable paper. We may have been lucky this time assuming we assure that the recounts are done on honest equipment. In future elections widespread fraud involving DRE equipment without audits can result in perfect crimes. The optical ballots exist and can be recounted by hand or by honest optical scanners. We have every reason to hope that the original intent of all the voters can be recovered in a recount. Should major discrepancies exist between previously reported counts and recounts, then after the election investigations can be launched to discover why these discrepancies exist.

If anyone comes across information relevant to this continuing narrative please let me know so I can include it as the story unfolds. 

Send your comments to: Dennis Hall

Single Agent Analysis

Hypothetical analysis of the damage a single agent could potentially inflict on a national election.