Controversy has surrounded the use of “voter-verifiable paper audit trail” records (VVPAT) on Direct Record Electronic (DRE – often called “touch screen”) voting machines. The VVPAT is a printed record of an individual’s ballot, which the voter is supposed to review and is required to approve before the ballot is cast and recorded on the machine.

Advocates say the VVPAT is an essential source of voter intent and must be available in case of a recount or audit. Opponents say the VVPAT printouts may not match the electronic tally, and the electronic tally is normally the source of the official results.

A local election on May 20, 2008 proved that both sides are right — thanks to the ES&S iVotronic touch screen voting machines that produced incorrect electronic results in Faulkner County, Arkansas.

Here’s what happened.

There was an error in the ballot programming of the two iVotronic machines in the East Cadron B precinct. The District 45 State House contest simply wasn’t there. The county clerk noticed the error before the election and provided paper ballots for that contest in that precinct. (Here's a pdf of what the paper ballot for the precinct looked like.)

East Cadron B voters voted for District 45 candidates on the paper ballots and then voted for the other contests on the touch screen. If any of them had looked at the VVPAT, they would have seen votes for all the contests except the District 45 contest — naturally, since that race wasn’t on the ballot.

But the machine took the votes for candidates in the Constable race and handed them to the corresponding candidates in the District 45 House race. The electronic tally for East Cadron B precinct showed 57 votes in a race that wasn’t on the ballot.

"We assumed, erroneously, that it would not record that race since it was not on the ballot," [Faulkner County Election Commissioner Bruce] Haggard said, adding that the votes for the constable race were later found to have recorded accurately on the voter-verifiable paper trail and therefore would not have appeared erroneous to voters either. ("County to investigate latest election snafu")

After the error was discovered, the election commission used the VVPAT — which showed votes for Constable, but not votes for the non-existent race — to determine where the votes actually belonged. So, assuming the VVPAT correctly showed the voters’ votes, the results were adjusted correctly and all was well, except that Terry Fiddler who thought he had won the District 45 House seat actually ended up losing. (For more inforamation, read "Arkansas Election Officials Baffled by Machines that Flipped Race.")

Clearly, it is essential for e-voting machines to produce a VVPAT.

What is not clear is the value of electronic totals. As Faulkner County proved, the electronic totals can’t be trusted to be correct, or even to match the VVPAT. So why have the electronic totals at all?

The only logical conclusion from the false results reported by the iVotronics in Faulkner County is that wherever e-voting machines are used, they must produce a VVPAT. Furthermore, the electronic totals must be ignored, and the VVPAT must always be used for the official tally.

In fact, since the iVotronic has proven that it far exceeds the maximum error rate allowed by the Help America Vote Act, using the electronic tally of the iVotronics to produce official results for a federal election is a violation of federal law.

In short, iVotronics cannot be used in November UNLESS they produce a VVPAT and the VVPAT is used to produce the official tally. Any counties doing otherwise are knowingly violating federal law.

Intellectuals solve problems,
geniuses prevent them.
~ Albert Einstein