Team 01: Rosa Arias

Chad Feller

Walter Smith

Project: Electronic Ballot Reader (EBR)

External Advisor: Lou Montulli – Advisor for system implementation

Technological advancements have provided a means for the implementation of direct recording electronic (DRE) voting machines, which allow voters to directly mark their votes by means of a touch screen, push buttons, or similar device. The record of the votes generated by the voter is created by the software within the voting machine, resulting in a record that is just as trustworthy as the software that produced it. The software’s ability to record and tabulate votes and maintain accurate record keeping has to be trusted. However, voters are blindly placing their votes and trusting the machine without receiving any type of external verification or recount resource. This has generated much debate and several proposed solutions in the States where DRE voting machines have been integrated. One of these States is Nevada, which has taken the step to mandate the use of DRE voting machines in conjunction with Voter Verified Recorders (VVR’s), which output a human readable record of all of the ballots cast by the voters.

State of Nevada’s Secretary of State has chosen Sequoia Voting Systems as the supplier of the VVR’s. A VVR consists of two drum rollers with a continuous feed paper roll, a printer, and a clear window on the front of the sealed device which allows the verification of the printed record by the voter before leaving the voting booth. Additionally, the records produced by the VVR offer a means to perform a recount, if necessary. Events in which a recount would be required include a close election, software failure, hardware failure, and an internal or external audit. However, such a recount would require that each roll be counted by hand, which can prove to be a time consuming, expensive, and error-prone process.

In order to improve the recount process and verify the integrity of the electronic voting system, we have been presented with the opportunity to design an electronic ballot reader, which would read the printed record and tally the votes in order to verify the accuracy of the DRE/VVR record. With such a system, users would also be able to recall and view images of any ballots and records that the machine cannot verify or read, limiting the effects of OCR errors. The execution of this will require the use of an optical scanner, Optical Character Recognition software, tabulation software, configuration software and quality control software. Our preference is to use C/C++ and Qt, in conjunction with C/C++ libraries and GNU OCR software, on a Linux box in order to implement the necessary components. Since the VVR will be scanned and interpreted by OCR software, errors are a possibility. This will require the tabulation software to perform statistical analysis on the results, which will give officials the opportunity to intervene.

The goal of this project is to provide a working prototype to demonstrate to voting officials that the only part of the voting process that is not computerized can feasibly be computerized, eliminating the human error factor inherent in counting votes by hand. In demonstrating this successfully, the system may have the opportunity of becoming an integral part of the voting system within the state of Nevada. It may also have the potential to be adopted statewide and nationwide in any county, state, or national election. Currently, there are several touch screen electronic voting systems in existence now, but there is no means to verify the system, which will make our solution the first of its kind.

In order to accomplish the successful specification and implementation of the system, Dr. Sergiu Dascalu and Mr. Lou Montulli will provide us with the necessary guidance and advisement. We shall seek advice from Dr. Dascalu with regards to the specification of the project and from Mr. Montulli with regards to the desired implementation of the system due to his active involvement in working to attain such a system.

The implementation of this project will give us the opportunity to deal with hardware outside of the PC box. The process of integrating scanning hardware and software components into a traditional software project will give us a better understanding of how real-world systems are developed and deployed. Moreover, the challenges that we will face in detecting and handling erroneous ballots appropriately will further increase our experience in providing a clean and efficient user interface. Additionally, if the system proves to be successful, our names will gain a large amount of recognition.