Feds to Recommend Paper Trail for Electronic Votes

flanksteak writes "The National Institute of Standards and Technology is going to recommend the decertification of all electronic voting machines that don't create paper records. Although it sounds like this recommendation may have been in the works for a while, the recent issues in Sarasota, FL (18,000 missing votes) have brought the issue a higher profile. The most interesting comment in the story comes near the end, in which the author cites a study that said paper trails from electronic voting machines aren't all they're cracked up to be."

Comments

Best solution I've seen

[olivrwendl]

http://www.forbes.com/opinions/2006/11/10/voting-f raud-security-tech-security-cz_bs_1113security.htm l

This article came out in forbes a while back and the author has the best solution I've seen for verifying votes on electronic voting machines. He proposes having a touchscreen computer to make all of your ballot selections and when you are done and hit vote it prints out a piece of paper with your sslections. You then can verify your votes were recorded correctly before putting your ballot in a box so that it can be run through an optical scanner at the end of the day to count the votes.

Re:Best solution I've seen

[spisska]

The problem with this idea is the same problem with all electronic machines: It's a colossal expense to do something that would be handled just as well with $0.10 marker.

There are problems with any voting system. Here's the basic rundown:

1) Mechanical lever, paperless: a machine in which the voter sees the whole ballot, sets switches to indicate their vote, and pulls a master lever to cast the vote, which mechanically adjusts a counter mechanism that is recorded by hand after voting has closed. Problem: the gears can jam. This has happened before, and officials can generally identify jammed machines by an anomolous number of 9s -- the machine fails to advance beyond 9 in a given column because that requires the turning of two counting wheels rather than one, Trust me it's happened, and this is why mechanical levers countinue to be used in only a very few jurisdictions.

2) Punch card, paper: the voter indicates their choice by putting the ballot into a machine which makes a physical hole in the ballot that can then be read and counted by machine. Problems: There is no chance to correct a mistaken vote except by spoiling the ballot. An incomplete puch can lead to incorrect tabulation (hanging chads). Perforations that were not voted can still fall out before tabulation, meaning an overvote and an invalid ballot. Florida 2000. This is why virtually no jurisdictions still use punch cards.

3) Optical scan, paper: the voter indicates their choice by marking a paper ballot -- generally by making a mark in a given area of the ballot. Problems: Voters fail to follow directions and make marks other than where they're supposed to (eg circling names rather than checking boxes). Optical scanners identify stray marks as votes (or overvotes), or fail to identify votes. Folded or damaged ballots (particularly in the case of absentee ballots) cannot be read by machine. Optical scan ballots are still very popular, and probably the best solution now being used.

4) DRE - direct recording electronic, paperless or optionally with a voter-verified paper audit trail (VVPAT): A voter makes their selection through a touchscreen, keypad, or audio interface and the vote is recorded digitally onto a removable storage unit -- generally a PCMCIA card. Some units feature a simultaneous paper record to present the voter with a hard copy representing their vote. Problems: No transparency between the casting and counting process. No software independence -- the tabulation of votes is dependent on the same software that records the votes. No independent auditability -- there is no way of verifying that the voting and/or tabulation software has not been compromised. No physical record of the vote when there is no VVPAT, and no dependable physical record when there is a VVPAT.

Paper trails are the most failure-prone parts of the machine and offer no effective protection of the process -- printers can fail, paper loaded incorrectly, ink runs out, paper jams, paper runs out, etc, etc, etc. If it can happen it will, especially in a machine whose hardware is little tested and whose software is engineered on short notice (due to election law that often changes dangerously close to elections), and a machine designed to be used only a couple times per year.

5) All-mail: this is a system that is being pioneered by Washington State -- all voters vote by absentee ballot delivered through the mail. Eliminates the need for polling places, poll workers, etc. Problem: Opens the door widely to massive vote fraud.

6) Colored Stones Cast into an Urn, paperless: A very effective system used in ancient republics wherein voters would indicate their choices by placing stones of different colors into given vessels to indicate their vote. No question of hanging chads, hacked machines, misunderstood ballots, etc. Problems: Not machine readable, somewhat impractical for large precincts and long ballots, expensive and difficult to transport and verify stone counts. Very, very few governments or municipalities have used this me

Re:Best solution I've seen

[dasunt]

How is that better than voting by marking up a heavy card stock ballot with a marker and running it through an optical scanner? If the goal is to minimize steps, why have the touch screen mumbo jumbo at all?

Because, with a computer-generated card, the result should be more or less binary -- either Bob voted for a candidate or he didn't.

With a card filled in by a voter, there can be some debate about how complete a mark must be before it counts. Witness the hanging chad hell in Florida.

(OTOH, with computer generated cards, since they are computer generated, it should be trivial to print out fake ballots and stuff the box. But the fake ballots will lack different and unique finger prints.) :/

Nobody's decertifying anything

[Anonymous Coward]

You've got it wrong. NIST is not proposing to decertify anything. The white paper only talks about proposals for what requirements ought to be in the 2007 federal voting system standards. Read the white paper.

That will make me feel better

[notaprguy]

I'm far from a conspiracy theorist but the more I read the more convinced I am that there is something rotten in Denmark...or at least the RNC and their cronies. Check out http://www.tpmmuckraker.com/archives/002027.php for the scary details on the FL-13 race. More than 18,000 voters in primarly Democratic leaning areas showed NO vote for the Congressional candidate. An excerpt from the Orlando Sentinal:

The group of nearly 18,000 voters that registered no choice in Sarasota's disputed congressional election solidly backed Democratic candidates in all five of Florida's statewide races, an Orlando Sentinel analysis of ballot data shows. Among these voters, even the weakest Democrat -- agriculture-commissioner candidate Eric Copeland -- outpaced a much-better-known Republican incumbent by 551 votes....

Make this a priority for next year

[paulproteus]

Wow, great timing!

Democracy for America, the follow-up to Howard Dean's Dean for America organization, is running a "Put paper ballets on the agenda" drive right now. They want people to tell Nancy Pelosi, as the future Speaker of the House, to make this a priority for next year's Congress.

So if you care about this issue, make sure she hears about it!

For what it's worth, I filed testimony in the EFF lawsuit, OPG v. Diebold, where Diebold was suing kids who (like me!) posted to the Web copies of some Diebold memos in which you can read about Florida precicints with negative 16,000 votes for Al Gore and Diebold "upgrading" the software to uncertified (read: "illegal") versions in California.

Re:Paper records

[dsandler]

Doesn't matter. In the case of a recount, the paper ballots---the ones the voter verified---are used.

the Draft NIST recommendations are pretty good!!

[ukemike]

Actually this time just RTFA isn't enough. You need to read the whitepaper as well.

It appears that at least one federal agency has not turned to the dark side. The draft NIST white paper recommends a voter verifiable paper audit trail that is also the ballot of record, AND robust auditing. I was very pleased to read it. I hope the final document isn't watered down, and I hope this or something similar is implemented in time for the 2008 election.

The premise of the whitepaper is that no software dependent system for counting votes (like a touchscreen with no paper ballot) can be fully vetted, and that they should never be used without a software independent record for use in mandatory statistically robust audits.

In other election reform news... There is an organization that has been a key mover in the election reform movement called electionarchive.org. They did a lot of very interesting statistical analysis of the 2004 elections and found some startling results. They have made a very solid list of 15 legislative recommendations. They can be found here:

http://electionarchive.org/ucvInfo/US/EI-FederalLe gislationProposal.pdf

here is a list of the electionarchive.org recommendations
1.Manual Audits
2.Voter Service Reports
3.Auditable Voting Systems
4.Fund Manual Audits and Voter Service Reports
5.Teeth (enforcement)
6.Public Election Records
7.Election Monitoring Website
8.Submission of Reports
9.Public Disclosure of Voting System Software
10.Prohibit Certain Network Connections
11.Qualifications for Technical Guidelines Development Committee
12.Public Right to Observe
13.Vote Count Audit and Recount Committee
14.Repository for Voting System Disclosure
15.Prohibit Practices that Disenfranchise Voters