by Julie Bryce
We got the chance to chat a bit with Alan Dechert during the 2007 Red Hat Summit. We asked him some questions; he gave us some answers.
Red Hat Magazine: To help readers understand the terminology, can you explain “disclosed source” and “open source” software as it relates to the voting process?
Alan Dechert: Mostly, we explain these terms informally. Where a more precise definition of open source is required, we point to the Free Software Foundation definition (four freedoms) and/or the Open Source Initiative definition (more nuanced, ten criteria). We describe disclosed source as source code that is published — freely available for public perusal. Open source is disclosed source that is not only freely available for examination, but can be used and/or modified by the recipient in accordance to a free software license.
Intellectual Property (IP) rights of vendors is an important issue, and we’ve had to make the case that the public’s right to know how their votes are counted overrides the vendors’ right to keep these processes secret. Still, there is an “IP taking” issue about which we have to be sensitive.
RHM: What function does the software perform in the voting process?
AD: There are quite a few pieces of software used in the administration of elections. The software may look quite different depending on the system used. There are three main areas of concern:
- Ballot layout (could be on screen, paper, or audio presentation
- Vote capture
- Counting the votes.
Mistakes in any of these three functions could lead to incorrect election results.
Capturing the voter’s input should be unambiguous on a computerized system. With hand-marked paper ballots, there will be ambiguous marks or marks that cause mis-reads. Problems sometimes appear when a voter tries to change their selection, or when they write in a candidate. No matter how good the scanning software, there will always be some voter intent issues with hand marked ballots. For example, what if a voter writes in the name of their candidate but forgets to fill in the bubble next to the “write-in” label? Voters will sometimes write in the name of a candidate even when their candidate is on the ballot. If the bubble is not filled in, the vote might not be counted even though the voter’s intent is clear.
Perhaps surprisingly, the greatest challenge in presenting choices to voters has to do with access for people with disabilities. Federal law requires that each precinct have a machine available that will accommodate such voters. Computers make it feasible for people with disabilities to vote privately and unassisted, but the task is complicated by several factors. For one thing, the population to be served by these machines is not well defined. Accommodating reading-impaired voters is difficult enough, even if their hearing is okay, and they have good manual dexterity, and they have good cognitive ability. A voter with any combination of disabilities, may be very difficult to accommodate.
Also there are two databases required for the voting process. There must be a database that lists all the precincts in a county along with all the contests in the county for a particular election. This database needs to keep track of which precincts are involved in each contest. In addition, all the candidates for each office need to be kept along with party affiliation (if any). Initiatives need to be there too along with the explanatory text for each one. You also need an indication for how each contest will be scored (for example, vote for one, vote for three, yes/no, ranked choice, etc.). This database must be complete, accurate, and ready to use in order to layout the ballots or to aggregate the votes.
The other important database would be the voter registration database. Keeping this up to date and accurate is a great challenge. Voter disenfranchisement due to errors in this database is common From this, there must be a way to produce a poll book that lists all the eligible voters in a precinct. Some jurisdictions are moving to electronic poll books, which are a necessity to enable a voter to vote with the correct ballot even if they are unable to vote at their normal pollsite. For manually marked paper pollbook, the voter’s name can simply be checked off to indicate they have voted. If an electronic pollbook is used, this needs to be indicated electronically so the voter cannot vote more than once.
There is also software for testing and auditing purposes. Better audit tools are needed, and this is an area ripe for open source software development.
RHM: How corrupt is the current system of voting?
The short answer is that I think the system is quite corrupt, and problems with the voting system are generally underreported in the media.
The slightly better news is that the general trend over the years and decades, is improvement — slow and sometimes unsteady improvement. This doesn’t give much comfort because there is a long history funny business in the United States voting system. We have a long way to go before we should be happy with the system.
In order for democracy to work, voters must have confidence that their votes will be counted correctly. If they don’t have this confidence, why vote? And, why pay any attention to the issues if you’re not planning to vote? Election officials know this and the media knows this too. That’s why the “every vote counted” myth has been heavily sold even though they have long known that it is not exactly true. If you are making too much of these problems, you are undermining voter confidence, thus undermining democracy.
Voter turnout in the U.S. is near the low for established democracies around the world. Turnout has declined significantly since the 1960′s. Nearly half of our citizens eligible to vote do not show up for presidential year elections. It’s much worse for off-year elections.
How confident are we that the correct winner was declared when the margin of victory was one-tenth of one percent? My educated guess is that, depending somewhat on what jurisdictions are involved in the vote, it’s about 50-50. In other words, the system still isn’t good enough to count votes with this level of precision. We may as well flip a coin instead of letting the courts decide. We should not have to pray for landslides in order to be confident in the outcome. of an election.
RHM: Are technically modern and transparent polls more secure than traditional voting methods?
AD: Computers do offer gains in efficiency and security. Not to the extent that we can go with paperless voting machines, however. We (OVC) vigorously oppose paperless voting.
The main objection to computerized voting systems is that we have taken a process that was once easy for everyone to understand, and put the process in the hands of experts. We don’t know if we can trust the experts and important details about how the systems work and how they are tested are simply unavailable to the public. Before computers, the system could be hacked, but at least everyone could understand how it was being done.
Open Voting seeks to employ the advantages of computers, while making everything as open to public scrutiny as possible.
RHM: Back to terminology, can you give a brief explanation of the difference between traditional voting methods and the more technically modern?
AD: In 2000, about half the votes were cast on two types of very old systems: mechanical lever machines and punch cards. Due to reliability issues with these systems, they have been mostly replaced at this point. The more I look at the Diebold TS voting machine, the better those old systems look. However, we are not going back to punch cards and levers.
The most modern design that has substantial market share is the DRE (Direct Record Electronic) machine. With this design, you make selections on a screen (usually a computer screen, but not necessarily, and usually a touchscreen, but not necessarily). When you get to the end, you select a button that says, “cast vote,” and you are done. There have been a lot of concerns about the security of votes that are purely electronic and difficult to audit. Papertrail printing mechanisms have been added to many (perhaps most) of these machines, but this makes for an expensive kludge. One of the main motives to going with DREs was to have a voting machine that could be used by people with disabilities (reading impaired, especially) as well as everyone else. In practice, this has not worked out well because many jurisdictions have opted to buy only one per precinct (due to high cost) and use optical scan. So, officials have to administer two different voting methods.
The Florida Governor recently announced his intent to dump DREs in favor of an optical scan system. With optical scan systems, the voter manually marks a paper ballot, which is tallied using a scanner. This system has some advantages and may soon be the most popular method. At least if there is a problem, we can go back and look at the ballots.
Optical scan has two main disadvantages: Since the ballots are hand-marked, there will be a few that are problematic. This is not normally a big problem, but when a contest is decided by fraction of one percent, these few questionable ballots can mean the courts decide the outcome. The other problem is that accessibility requirements mean that two different systems must be available at the pollsite.
We believe that another design may be best, but needs further development. This is called an Electronic Ballot Printer (EBP). It would be something like a DRE, but when you get to the end, you select “Print Ballot,” instead of “Cast Ballot.” The voter casts the ballot by taking the printout and placing in the ballot box. The fundamental representation of the vote is on a human verifiable and hand re-countable piece of paper. We have shown that a reading-impaired person can use this system (with headphones attached) making it possible for everyone to use the same machine. If it can be made inexpensive (we think it can), it could also become the most cost-effective way because it would eliminate the need for preprinted ballots at the pollsite.
Hand-counted, hand-marked paper ballots are an attractive option in some ways, but it is very labor intensive and requires very active public oversight to ensure no manipulation of the ballots. This method is used for less than one-half of one percent of the votes cast in the U.S. It tends to work in jurisdictions that have relatively simple ballots (no more than about a dozen contests). This method does not seem feasible for urban areas with large complex ballots. Accessibility requirements also limit this method.
RHM: Okay, here’s the million-dollar question everyone’s probably thinking themselves right now. Do you think we’ll ever vote via the internet?
AD: I think the Internet will continue to play an increasing role in the administration of public elections. However, if you mean are you going to be able to log on to you home computer in your pajamas and cast a ballot in a public election similar to casting a vote in some online poll, I would say this is not likely any time soon, if ever.
Consider my own home environment. I am the default IT department here. I set up all the computers and provide tech support for my wife and two children, and I know all the passwords. If we had a “father knows best” (or “mother knows best”) situation in a household, how can we be sure one parent did not cast all the votes for the entire family?
It’s too hard to identify the voter while ensuring a secret ballot in this scenario. There are also difficult security issues with the Internet. See www.servesecurityreport.org for details.
It may become practical to use the Internet for voting at a remote location if there is an official available to identify the voter.
RHM: The voting machines code base (source here: http://sourceforge.net/projects/evm2003) is comprised of open source applications on top of a Linux operating system. How has the success of Linux in the mainstream technology industry helped?
AD: The success of Linux has been a great help to the Open Voting project. We often cite Linux, along with Apache, OpenOffice, and Firefox as examples of how open source is taking over. When people ask how a company is supposed to make money selling free software, sometimes we point to Red Hat as an example.
Microsoft has a lobbyist in Sacramento actively working against our bill. I don’t know of any other large companies lobbying against us. There are a few consortia, like the ITAA and the AeA, that have written opposition letters and testified at hearings. These seem to be led by Microsoft as well.
If Open Voting becomes a complete success, rather than run for president, I would probably start the Renewable Energy Consortium and Renewable Energy Foundation like I should have done 20 years ago, but lacked the knowledge and resources to do so. The foundation would focus on education and R&D, while the consortium would promote legislation to help bring about a transition to renewable energy. I’ve already outlined what our state-level legislation would do (search on DECHERT GETLOCAL if you want to find out about that).
RHM: What do you know about the Bilderberg Group?
AD: This reminds me of a bumper sticker I saw some time ago. It said something like, “The world is run by a very small group of very wealthy and very powerful people, none of whom you or I know or have any influence over.”
We have no basis for denying the right of any group of people from holding meetings that are private and confidential. This lack of transparency is unacceptable for governments, however. I don’t care if Henry Kissinger is in some secret meeting with industrialists. I do care if he’s using his connections to broker some secret deal with government officials. The public should be able to get records of any such contacts via public records requests.
We have to insist that, as much as possible, governments conduct business in public. Again, processes in the voting system must all be public.
RHM: Any parting thoughts?
AD: We all have some power, and everything we do makes a difference. If enough ordinary people get behind some idea, it could overwhelm any political-economic force on the planet.
Part of my job is reminding people of the power they already have.