Patterns of Touch Screen Voting Machine Fraud

Identified and Documented

in Florida, Ohio, New Mexico and Elsewhere in 2004

If you already understand what happened in the 2004 Presidential Election, and you want to start doing something about it, click here to proceed to a coalition Action plan. If you need to see the proof that the Florida election was racked with manipulation and fraud, read on.

Activists investigating the 2004 Presidential election have identified hundreds of preceincts in Florida, Ohio and other states where the voting results do not match the exit polls. These inconsistencies occurred primarily in precincts where electronic voting machines with no paper trail were used. In Florida, these discrepancies appear to have amounted to enough to have been responsible for George Bush's statewide "victory" margin. Many of them were in precincts with a strong Democratic majority. Some media commentators have explained the gap between the exit polls and the final vote counts by claiming that the exit polls were flawed. However, in those precincts where there was a machine that produced a "paper trail," the exit polls almost exactly matched the actual vote and there were few discrepancies giving George Bush extra votes. When a voter casts his or her ballot for someone other than the candidate they intended to vote for, this is called a "misvote." Misvotes in Ohio,Florida, and New Mexico appear to have given George Bush his winning percentage. (Misvotes favoring George Bush reached as high as 40% on some vote machines in some Florida, Ohio and New Mexico precincts. There were also high misvote totals in other states.)

On November 22, the Electronic Frontier Foundation (EFF) and the Verified Voting Foundation (VVF) announced that they had sent letters to voting officials in eight counties around the country urging them to allow independent testing of their electronic voting machines. The two groups were among the 60 organizations in the Election Protection Coalition (EPC), which ran an Election Day hotline and the web-based Election Incident Reporting System (EIRS). The Coalition received 40,002 reports of election irregularities, including 2,242 incidents concerning voting machines. Click here for an analysis of some of these incidents by a team of computer scientists. (This link requires Adobe Acrobat Reader.) According to EFF and VFF, the most serious problems were reported in Mahoning and Franklin counties in Ohio, Broward and Palm Beach counties in Florida in Florida, Mercer and Philadelphia counties in Pennsylvania, Harris County in Texas and Bernalillo County in New Mexico. Florida and Ohio were the big swing states that gave the election to George Bush.

While any form of voting fraud or interference is bad, a malfunctioning voting machine can prevent hundreds of people from casting their votes -- or change the votes of those who do. Most computer experts who have studied electronic voting do not consider the systems used in the 2004 election to be secure or reliable. The state of California has successfully sued Diebold, the manufacturer of one touchscreen voting machine over this very issue, after machines that were purchased for California turned out to be unusable. According to programmers and engineers who have investigated the security of electronic voting machines, touchscreen machines can be set up with a default choice for any candidate that would not be visible to the voters. (The Black Box Voting site explains some of the ways in which this can be done.) Their votes would automatically be cast for the default candidate -- such as George Bush -- unless they could successfully override the hidden default choice programmed into the computer. For example, if a voter deliberately chose not to vote for any Presidential candidate, the touchscreen voting machine would count the non-vote as a default vote for George Bush.

Reports from voters in Florida, New Mexico, Ohio, and elsewhere (especially other swing states) documented that many touchscreen voting machines appeared to have been set with a "Default to Bush". The "Default to Bush" could be changed only if a voter successfully selected another candidate. But it appears that in many cases the voters did not successfully override the "Default to Bush," in some cases because they did not notice the problem and in other cases because it was difficult or impossible to get the machine to accept another candidate. This was a major problem in New Mexico and Mahoning and Franklin counties in Ohio. There were also problems with "Default to Bush in the "Big Three" Florida counties: Palm Beach, Broward and Dade, and elsewhere in Pinellas, Hillsboro, Pasco, Sarasota and Lee. In fact, Florida was the state with the most reported incidents in the Election Protection Coalition/Election Incident Reporting System database. (There are state-by-state links below.)

Election officials had to replace some of the machines in Mahoning County (Ohio) after repeated attempts by technicians to "recalibrate" them failed. This also happened in Florida and New Mexico. The EIRS system also identified patterns of default away from Kerry and the minor party candidates elsewhere. The same pattern was also found in some U.S. Senate races, including the race in Florida, which elected Republican Mel Martinez over Betty Castor.

One of the two most severe touchscreen voting machine errors reported to EIRS involved voters who selected Kerry on an ES&ampS electronic touchscreen and saw their vote changed to Bush on a summary screen. ES&ampS machines are the most commonly used in Florida. Florida's other worst touchscreen problem was with the Sequoia AVC Edge machine where voters saw preselected default choices presented to them. (Those incidents are also listed in the link above.) In C Edge machines changed Kerry votes to Bush votes. Sequoia AVC Edge machines are used in 4 Florida counties: Palm Beach, Pinellas, Hillsborough and Indian River.

If a voting machine were programmed to "Default to Bush," he (or any other default candidate) would get the benefit of misvotes from voters who didn't notice that their intended candidate wasn't selected, along with misvotes from any voter who intentionaly did not vote in the Presidental race, any accidental non-votes AND unsuccessfully completed votes. Quite an advantage! Under normal circumstances this would result in a swing of 2% to 4%, depending on the precinct. But since it is a matter of record that in many precincts the default was hard to override and sometimes impossible, only the most watchful or purposeful voters may have been able to successfully vote for a chosen non-Republican candidate. EIRS collected reports verifying these incidents from poll workers as well as voters. Poll Workers reported that some machines in minority precincts appear to have been targeted to produce high levels of misvotes, so the swing was likely much larger in those precincts. EIRS documented a large number of incidents in minority precincts. There were also a few cases where precincts had machines set to default for a minor party candidate, but none reported to default to John Kerry. These patterns of misvotes in counties using touchscreen voting machines have been confirmed by analysts and computer experts using EIRS data and other information from Ohio, Florida, New Mexico and elsewhere. In the big South Florida counties, out of those who reported misvotes from machines that defaulted to Bush or changed votes to Bush, between 33% and 50% were unable to correct the problem. Since these statistics represent people who noticed the misvotes, among voters in general the percent of misvotes must be even higher. It was also noted in the EIRS reports that many voters did not know they had to complete their vote by pressing a button before they left the booth, and it was reported that thousands did not finalize their vote. In a "Default to Bush" situation, the touchscreen voting machine would count an incomplete vote as a vote for Bush. (Read EIRS reports here.)

Computer security experts and engineers from the International Electronic and Electrical Engineers (IEEE) found evidence that the manufacturer's technicians or representatives have remote access to some vote machines while they are in service, and can change defaults and other settings remotely. Furthermore, some of the voting machine manufacturers monitored the election results remotely on election day. (By default, the password for Elections Systems and Software (ES&ampS) machines is said to be "1111" -- not a secure password at all!.) Scientists at Johns Hopkins University found tampering with Diebold machines (which fortunately were not used in Florida during the 2004 election).

Many computer systems are designed so that service technicans are able to make changes to the software through a remote connection, but this is not an appropriate feature for a voting machine. The actual security of electronic voting machines is difficult for the general public to determine. Voting machine designs are certified by two highly-secretive consulting firms which have refused to disclose their procedures for testing machines for accuracy and security. Even when they certify a machine, the certificate applies only to the design of the hardware and software -- any individual machine may be altered before an election.

Vote Machine Fraud documented in the

Big Florida Coastal Counties Using Touchscreens

The biggest Democratic counties in Florida are Palm Beach, Broward, and Dade. Each of these counties had dozens of precincts where vote-switching was reported. Most involved switches from Kerry to Bush, but there a few reported cases of votes switching from Kerry to a minor candidate or no vote. (Recall that non-votes were likely to be switched to Bush. A smaller number of reports documented votes switched from Betty Castor to Mel Martinez in Florida's U.S. Senate race. Poll workers and poll watchers told frustrated voters who were having trouble that "this has been going on all day". EIRS found that these problems were so widespread they were reported throughout Election Day on local Florida radio and TV stations. A smaller number of vote switching cases were reported in Hillsboro, Sarasota, Pinellas, Pasco, and Lee counties. Some of these were votes switched in the U.S. Senate race. (Click here to see the EIRS report Florida's voting machines.)

If a significant number of machines were programmed for "Default to Bush," this would mean fewer than normal undervotes. (In an undervote, nobody gets the vote.) In fact, undervotes decreased by over 50% in all Florida touchscreen counites except Broward. The experience of Mahoning County in Ohio, which used ES&S Ivotronic touchscreen machines, is a good comparison. In Mahoning, EIRS reports of switching include many affidavits. The majority of machines defaulted to Bush, but some defaulted to "blank" (an undervote). Thus voters for Kerry, who chose Kerry but didn't notice that the vote later switched to Bush, would end up either voting for Bush or "blank" (an undervote), depending upon the machine default. Kerry would lose all of these votes, but the pattern of undervotes would not look so unusual. The default to blank might also be considered less suspicious than the "Default to Bush" pattern and would be less likely to get reported through the hotline system (which most voters were not familiar with -- many disasterous irregularities were never reported). A focused search of the Florida EIRS data has confirmed the same pattern also exists in Florida, especially Broward County. Several computers defaulted to blank after Kerry was initially chosen, as seen in the EIRS cases.

The EIRS cases for each county reported here are based on a limited search of the thousands of incident reports, and it is likely that these are just a small fraction of the total number of incidents.

These documented cases of vote-switching prove that the "Default to Bush" was programed into a number of the touchscreen vote machines of the big touchscreen counties. Computers don't consistently do something unless they are programed to do it. This appears to explain the big gap between the exit polls and final vote counts in the big touchscreen counties. The exit polls appear to have been correct, just as they matched the final vote counts in all the non-touchscreen counties. The only questions are, Who was responsible? and What can we do about it?

My previous study examined Presidential votes by county in 2000 and new registrations by party from 2000 to 2004. It pointed out an ususual swing to Bush inconsistent with previous voting patterns and new voter registrations in these counties.

This vote machine fraud pattern is consistent with the unusual shift to Republican candidates, which is documented by several detailed studies of voting and registration patterns, along with the exit poll data that also documented this unusual and unexplained shift in the official votes.

Widespread systematic Dirty Tricks --

Voter Supression and Malfeasance

Minority voters in at least twenty states were targeted for voter supression, including Florida, New Mexico and Ohio. This was well documented from the EIRS investigations. The following cases and analysis also have further documentation on the voting machine manipulation patterns. (Read about the dirty tricks here.)

Florida County-by-County Reports

Palm Beach County

Broward County

Dade County

Other Counties

Media Reports

Comparing Florida in 2000 with Florida in 2004

I've done an analysis of the Florida election data comparing 2004 results for President to the 2000 results and also to the increase in voter registration for Democrats and Republicans between 2000 and 2004.

As previously noted in a study by researchers at the University of California-Berkeley, there are some large and unusual patterns in the big touchscreen counties and a few of the big optical scan counties. They had a big Repub vote swing that was not consistent with the 2000 vote and the Democratic and Republican registration increases between 2000 and 2004.

The touchscreen counties with unusual Republican vote increases include: Palm Beach, Miami/Dade, Broward, Hillsboro, Duval, Indian River, Lake, Martin, Pinellas and Sarasota, Bay, Brevard, Manatee, Orange, Osceola, Polk, Citrus, Escambia, Hernando, Marion, Seminole. There were documented cases in some of these counties of minority voters targeted by dirty tricks and absentee ballot irregularities.

The only explanations for this pattern other than irregularities would be that the Republicans hugely beat the Democrats in get out the vote or a lot more Democrats voted for Bush in 2004 than in 2000. The Democrats had a record get-out-the-vote effort in Florida, and the number of Democratic voters in Florida in 2004 exceeded their turnout in 2000.

The data I used in these studies came from:

The Florida Division of Elections Voter Registration Data

The Division of Elections: "Floridians Rate the 2004 Election"

Other States

Here is a link to more FLCV data on dirty tricks around the country. The types of election fraud committed in Florida can be found in more than 20 states.


FreePress.Org has an archive of articles documenting cases from all around Ohio.

Mahoning County

Cuyahoga County

Franklin County

Hamilton County

Lucas County

Summit County


Mahoning County Official Results



New Mexico

Phantom Votes

Additonal Data from

The report on New Mexico

The report on other election problems in New Mexico

See the EIRS investigative reports that include such offenses and the cases: Ohio, New Mexico, and Florida.