From: Adrienne Vayssières Kandel [adrienne@dcn.davis.ca.us] Sent: Tuesday, September 09, 2003 12:47 AM To: Mariko Yamada; MarchandBA@aol.com Subject: Citizen Input on Voting Machines I have some comments for your committee choosing voting machines. First a brief list, then the 1st three are developed in detail. In brief: 1. Make sure the machines support instant runoff voting. 2. Make sure the machines have their programming built into the hardware, not into PC software that can be hacked. 3. You absolutely need a realtime paper trail (a paper record of each vote that the voter can verify). 4. To prevent fraud, make sure the vendor gives you the computer code (used to program and build the hardware). 5. Beware in checking machines by random small tests - a dishonest vendor could program the machine to start changing votes only after a large number of votes have been recorded. 6. Test any proposed voting system on a cross-section of the public. The people who designed the infamous Florida butterfly ballots thought they were doing right, but they never tested their system's accuracy. 7. Vendors should expect that you wil check for fraud by having exit polls at random polling places, and doing recounts (using the paper trails verified by the voters) where the vote outcome is beyond reasonable sampling error from the exit poll outcome. They should expect to go to jail for a long time if they try to steal the election. Now for the details on the 1st three: 1. INSTANT RUNOFF VOTING allows the voter to right his or her 1st choice, 2nd choice, 3rd choice, and so on for each office. If there's a majority winner among 1st choices, s/he wins the election. If there isn't, there's the equivalent of a runoff election: people voting for the low-ranking candidate as 1st choice have their 2nd choice counted. Because all choices are listed, the computer program can hold as many runoffs as necessary to get a clear majority winner. These runoffs cost nothing to administer, and don't invite additional campaign spending. Ireland elects its president by instant runoff voting (IRV), and has done so for decades. Australia elects its House of Representatives that way, and has done so for nearly a century (of course these places originally did it without computer help). Since 2000, London elects its mayor that way, and Bosnia-Herzegovina its president. It's a growing trend in the U.S., too. Cambridge, Massachusetts and San Francisco recently adopted if for city elections. It has been discussed in the California State Legislature, and will likely be discussed more in the future. Other localities include San Leandro, Santa Clara County, and Vancouver, Washington. UC campus student associations have been adopting it. IRV is the one of the two most Democratic ways to hold an election. The other is non-instant runoff voting - holding costly elections again and again until a candidate has a clear majority. It completely eliminates the "spoiler" effect, an effect I have been ruing since 2000. To understand its workings, I recommend the following advocates' website, which I found today using Google, and which lets you do some IRV yourself so you see how it works: http://www.calirv.org/ - it's fun. My first choice of ice cream won, even though it did not have the initial plurality of votes. 2. MAKE SURE THE MACHINES HAVE THEIR VOTING CODE built into the hardware, not programmed into the software. Otherwise a hacker can change election results. (Hackers can reprogram code, but not hardware, with its hardwired programs. Here I reproduce an article my husband found in PC magazine, and which you can also read at http://www.pcmag.com/article2/0,4149,1232598,00.asp while it lasts: "HACK THE VOTE An interesting battle is taking shape between Diebold Election Systems and researchers at Johns Hopkins University and Rice University. The Diebold AccuVote-TS electronic voting machine, used in 37 states, apparently has PC-type innards, a touch screen, and a smart card to forestall potential fiascoes such as the 2000 presidential election in Florida. Researchers have been investigating some source code, ostensibly that of the AccuVote-TS, and have found it wanting. They discovered ways to hack into the system, monitor the progress of an election, vote multiple times, and do all the things that highly motivated, morally deficient people with a political bent have been doing since time immemorial. You can find the charges and analysis at www.blackboxvoting.org/access-diebold.htm and http://avirubin.com/vote.pdf. You can read Diebold's rebuttal at www.dieboldes.com. As you look at both sides, you may find some of the attacks unrealistic or at least unlikely. You may also find Diebold's defenses somewhat idealistic, and the technical rebuttal is anything but technical. I won't claim that Diebold's voting machines are deficient, but I think that building a voting machine (or a medical machine or a space probe) on PC hardware and the Windows operating system is a terrible idea. Give me a microcontroller and burned-in code that can't do anything but what I program it to do, not a general-purpose environment that is universally and routinely hacked. Enough thinking for one vacation—surf's up!" 3. A PAPER TRAIL is the only way to hold a valid recount and fully ensure the honesty of election machines. (Exit polls should also be used to alert the county to potential fraud.) This means every ballot either has to be printed out, or it has to be a paper ballot in the first place, that was then read by a machine. (Some people have suggested scantrons, with other arrangements available for the handicapped. I think this would work - and lessen the number of working printers required at each election site.) The authors of the report to Kevin Shelley discuss an electronic trail, but I don't see how you can make that foolproof, since the voters cannot check the electrons and see that they are sending the proper message to wherever the info is stored. They can read a paper and verify it truly represents their vote. But a screen can lie to them. I noticed that in Kevin Shelley's panel's report of voting machine possiblities, the 2 computer scientists said you needed a paper trail right now, to avoid fraud, while some of the less computer-educated members voted against it. I trust the computer scientists on this. Thank you. Adrienne Kandel 1420 Chestnut Place Davis, CA 95616 adrienne@dcn.org 530-758-3969 I'm sending these to the people I know are in YTAC: Betsy Marchand, Mariko Yamada, and the county clerk at her general public email address (cntyclrk@yolorecorder.org). I request they be forwarded to the rest of the committee. Thanks.