A Long Time Coming
Many of the touch screen voting machines that we have were quite obviously designed rapidly to meet the emerging marketplace. Frankly, I don't know how so many of them were certified in the first place.California Sec State Debra Bowen opened her eyes to all the criticism. She commissioned an academic study to determine the viability of the machines from a security standpoint. Security teams were able to undermine the technology in a very short period of time.
A real solution
One thing that vendors have done properly is that they have provided a model for the future. Touch screen voting has been widely accepted by voters, and the implementation of voting machines throughout the country has brought up the things that are important to voters.Now that we know what we need to do, why not implement a real open source solution? A software based open source solution could be implemented on commodity level hardware - perhaps even the existing voting machines - and would achieve true security through open research as opposed to security through obscurity and haphazard deployment of updates.
Being that in all the years I've heard complaints about electronic voting I don't see anybody tackling the issue. Perhaps in my spare time I can start an open source project and pick up support for it down the road as I achieve real progress.
The Requirements
I've thought about this for some time, but I've rarely tried to put anything down. There are a lot of better ways to do it, but my blog is right in front of me, so I offer a few notes as to design requirements.This isn't a complete set of requirements, but a few notes that should be considered. The first place to start is vote fraud. The best way to detect and deter fraud is through public scrutiny. At the end of each election cycle, a voter should be able to verify that their vote was cast correctly. Further, an independent media organization should be able to verify groups of votes on a larger scale. One thing to keep in mind is that many people prefer their politics remain private - so along with verification, real voter privacy must exist.
Another issue is printing. Many people consider a printed record of every vote to be of vast importance. I don't think like that. Recording every vote in every election on paper creates an immense amount of waste, much of it unnecessary because the records do not get reviewed and voters just throw their receipts away. Instead, I would elect for a system that printed ballots on demand and stored no data in regards to which ballots were printed. The verification system remains intact and waste is minimized.
Verification should be performed in stages - a local verification at the polling center is necessary, and there should be verification capability at each level of government from the municipality all the way up the ladder to the federal government. This enables verification for smaller local elections and geographically centered run-offs and re-votes. It also creates traceability so that mystery votes cannot be generated outside of polling centers.
Each ballot in the entirety that is appropriate for the given level of government should be stored together. In other words, when examining votes for verification, you should be able to see that a particular voter voted for x Senator, y Representative, and z Presidential Candidate. The purpose of this is true transparency. Many of the voting fraud cases I've read about consisted of padding the numbers for an individual candidate. With record based transactions stored, votes for individual candidates cannot be padded, and any record based padding can be analyzed for patterns - helpful for identifying other questionable data.
Additionally in regards to verification, the verification must be fast and highly scalable. It should support every voter in the United States verifying their data within a 24 hour period.
What I envision is a system that would enable voters to verify their own individual votes. It would enable polling center volunteers to verify the vote counts produced by their polling centers. It would enable voters to pass verification information off for external unscrutinizing voluntarily. It would enable printed output of any individual record at any time. The media would have better and more reliable access to exit poll information. And votes could be verified at any given point and time from when they are recorded through and past the official closure of the election. The more open the system is, the better assured the voters are that the election is valid.
On other notes, the system needs to support a touch screen interface, but not require one. The system should have an easy way to provide security updates, but that method needs it's own security measures to be in place. Voting needs to be simple. Verification needs to be simple. Setting up an election needs to be simple. Updating voting machines needs to be simple. Deploying voting machines needs to be simple. All of these things need to be so simple that a volunteer not educated with computers or electronics who has a character trait of not following instructions carefully can accomplish the task without a problem 100% of the time. Everything should be as simple as dropping bread into a toaster.
Well, there you have it... a few notes to get started. Open Source project coming soon to a theater near you.
California Voting Machines Decertified Commentary
