Glass Box Voting

Not happy with their eSlate machines, considering paper or opscan options.

Same as it ever was:

County officials say the machines — manufactured locally by Austin-based Hart InterCivic Inc. — are superior to paper ballots because they are faster, save time and money, are easier for people with disabilities to use, and erase doubts about “voter intent” that made the hanging chads of the 2000 U.S. presidential election so infamous.

But with a high profile, high-turnout primary just a week away, critics say computer glitches on electronic systems could lead to errors in vote totals and, without paper records, make real recounts impossible.

House Bill 1502 (PDF full text, first reading, status page)

Sponsored By:

Delegate Hixson

Entitled:

Elections - Instant Runoff Method of Voting

Synopsis:

Establishing an instant runoff method for casting and tabulating votes in the State; establishing procedures, standards, and requirements relating to the implementation of an instant runoff method of voting; requiring the State Board of Elections and local boards of elections to take specified actions to accommodate and implement an instant runoff method of voting; requiring the State Board to adopt specified regulations; etc.

quoting from text after the jump (more…)

Verizon has knocked me back to 1910 by removing my network and thus VOIP, so I’m posting infrequently and hastily.

Colorado Recertifies 2 Voting Systems

The Colorado secretary of state recertified two existing voting systems, ES&S and Sequoia. Hart Intercivic optical scan machines are in limbo. One thing I want to change about the article is the assumption that new machines have to be costly.

Maryland Law on Voting Systems

Senate Bill 392 is law, but whether it takes effect is uncertain until the state budget is resolved. Here’s the text of the bill. The Glass Box Voting design complies with a great percentage of the requirements.

Here’s an NPR story on the flux in the voting system market and confusion about requirements (thanks reader for the tip)

From the Security Musings blog:

I heard Nicko Van Someren talk about finding keys in memory at RSA Europe in 2000, but when he spoke the ability steal RAM from a sleeping computer was unknown. Fascinating demonstration of the exploit. This is why cryptomodules must provide the ability to zeroize keys, regardless of where they are stored.From what I’ve heard, RAM need not be overwritten more than once like magnetic media must, but perhaps that is a commonly held mistaken impression also.

Another article about public discontent with electronic voting machines, the complaint is lack of VVPAT.  I had not heard of Danaher machines, will have to learn more about those.

From New Jersey, reports of reliability concerns with Sequoia machines.

From Ohio, news of a speaking tour by two representatives of the Secretary of State’s office to raise awareness of the results of the Everest Study (pdf link), concluded in December 2007.

On gmiller’s OSDV blog, which I found from the Facebook group .

[UPDATE]

I’m in favor of unmoderated comments.  Once I figure out how to set this site to do that, I will.

Interesting article in today’s Post.

In addition to machines, I believe most jurisdictions need more citizens to take part in election administration beyond just election day, both before the day (getting trained in election procedures generally) and after (auditing).

Glass Box Voting will not be able to supply certified machines for the 2008 elections, but we will have some ready for municipality-scale elections by election day 2009.

[UPDATE]

The administrator mentioned not knowing if the money is there. That’s a problem between him, Virginia, and the Electoral Assistance Commission, of course, but as far as I know, there are still billions in unallocated HAVA funds. What he lacks are products that are easy to administer, flexible enough to meet local requirements, and secure enough that the public will accept them. What a shame.

I just watched a very good talk on voting cryptosystems and the cryptographic principals and primitives behind them.  I found the link here.

What’s a better way to store votes?

versus

Safes

Safes are good places to keep money, because money, like mushrooms, grows in the dark. Safes are not good places to keep things that require light, like votes and houseplants.

When you put money in a safe, your concern is that the money be there when you open it. The bank that owns the safe has the same objective. You probably do not want the safe to have a readout that displays it’s contents to arbitrary members of the public.

You trust your bank, your bank trusts you (although they often should do more to be sure it is you, not someone stealing your identity) and it all works out. The bank staff knows how to operate the safe correctly, since they do it every day, and the safe is guarded at night by people who don’t know how to open it.

Ballot Boxes

As discussed before, a ballot box has different and greater security requirements.

A ballot box collects votes, not money. Votes, or more precisely ballot images, are not fungible like money - they are multidimensional, multivalued data elements rather than scalar units with a denomination. For example, my vote yesterday included the primary candidate for president (pick one), primary candidate for representative (pick one), circuit court judge (pick no more than two from a list of two), female delegates to the national convention (vote for no more than four), and male delegates to the national convention (vote for no more than four).

(How common is it to vote for delegates by gender btw?)

A ballot box must make information about it’s contents public. It should display how many votes have been cast, and should allow auditors to determine that votes have not been altered at any point.

Update

After watching the video in the post above, I realize that I overstated the case. A ballot box much make some information about it’s contents public, e.g. number of votes and the current validity status of each vote. The vote choice or ballot image must not be visible at all times, but only at the end of the process when it’s tallying time.