More on eVoting (DRE Machines)

I just finished reading a document written by one Daniel Castro at ITIF (http://www.innovationpolicy.org/), a think-tank. When you hear the phrase "think-tank" two question should spring in your mind before you can trust whatever the organization has put out:
a. Who's behind it?
b. What's their agenda?

In this case, ITIF seems to be very industry-friendly sometimes (http://www.itif.org/index.php?id=76) and somewhat neutral or even unfavorable at other times (http://www.itif.org/index.php?id=56)

Read through the website to get a flavor of what they do and what they're about. It does appear they are non-partisan to some extent, but it's going to be hard to guess without knowing who funds them.

This particular post deals with a 'paper' written by Daniel Castro about the use of paper audits during voting (on DRE machines, or Direct Recording Electronic machines). Curiously, he seems very much against it.

A good section of the document deals with the problems of paper ballot (when he should be discussing problems with paper audits). The document lists a couple of DRE concepts that could be applied for audit purposes, but somehow seems dead set against a paper printout.

One argument is that it'd be less secure. In my opinion, NOT having a receipt would be totally insecure. Would you like to do without your bank statement? Would you like to blindly deposit and spend money not knowing what's going on? I thought so.

Much the same way, a voter MUST know if his vote was recorded (he may not know if it was tallied, but the document has a section that deals with it quite well) and he must be able to store that receipt for reference.

The author also seems against disclosure of source code, arguing (disingenuously) that not only the system's code but also any third-party software code as well as OS code would have to be distributed. I've read many issues relating to Diebold's source code for one of their systems (http://avirubin.com/vote/analysis/index.html) so the author's argument is dangerous in that sense.

Some redeeming features of the article:
a. It does seek some sort of auditing
b. It proposes interesting new concepts to deal with eVoting issues (mainly tallying and verification)
c. It advocates a favorable outlook to companies that disclose their source code

However, the way the paper tackles (condescendingly) those that are FOR a paper audit is childish, amateurish, and completely runs against what the tone of such a paper should be (at least a little scholarly, IMHO).

Overall, I'd simply ignore this document (See this: http://home.businesswire.com/portal/site/google/index.jsp?ndmViewId=news_view&newsId=20070920005900&newsLang=en) -- especially because of this (extract from the above link):

About the Election Technology Council (www.electiontech.org)
The Election Technology Council (ETC) consists of companies that offer voting system technology hardware products, software and services to support the electoral process. The ETC represents manufacturers of the voting equipment used by over 90% of the population in the United States. These companies have organized as an industry trade association to work together to address common issues facing the industry. Membership in the ETC is open to any company in the election systems marketplace. Current members of the Election Technology Council include Election Systems & Software, Hart InterCivic, Premier Election Solutions and Sequoia Voting Systems.

Also, see my post here:
http://threeheadeddog.blogspot.com/2007/07/review-of-electronic-voting-systems.html

My view is this: People DESERVE to

• have a record of whom they voted for (verification/validation of vote)
• know their vote counted (verification/validation of local tally)
• feel satisfied that the total vote count reflects their decision/will with 100% accuracy (verification/validation of overall tally)

You can ensure the first by giving a paper copy of their vote (can be/should be anonymized -should have no tracking or identifiable information).

You can ensure the second by having a release of the total count from each machine for each county (along with audit verification by a third party - not the govt, not the officers of the electoral system, not a private company. How about the UN? :-)

You can ensure the third by repeating the above for the entire system (here the author introduces the use(fulness) of homomorphic cyrptography - a good idea, I think, with the use of, ironically enough, paper)

In conclusion, I wouldn't listen to any such so-called "think-tanks" or "policy centers" -- especially those that start with "Americans for" or have the words "Insititute" or "Center" in them.
I'd listen to the people.

Start by asking sane questions, and you will get surprisingly clear, smart, and highly implementable suggestions.

• What are you comfortable with using?
• Do you trust this voting system? What if we can show independent proof of how this works?
• Do you have any better ideas?
• How would YOU do it?
• What problems do you face with DRE machines or mechanical systems?
• What can we do to make sure we earn your trust in accepting the outcome of an election?

Grassroots organizations can help, too.

• Initiate focus groups to determine how to make the system simple, easy, and trustworthy
• Get communities involved (especially in depressed localities) in advocating the need for participation in the effort -- and to enhance turnout for the big event
• Initiate training and education throughout the nation (will help in removing doubts, suspicions and misconceptions; with the added bonus that you may get some excellent user-centered design ideas)

And a non-partisan election commission should:

• Make sure all machines pass a complete software (source code) test by a third-party validation agency (staffed by non-partisan scientists and researchers)
• Perform all manner of intrusion and hacking (physical, electronic, electrical, remote) to gauge the security of the product
• Assess all existing vulnerabilities and assign strict deadlines to fix the issues, then perform 100% regression testing
• MAKE ALL RESULTS PUBLIC - the process should be totally transparent and auditable/verifiable by any interested parties

Be safe!