April 24, 2014

avatar

Uncertified voting equipment

(Or, why doing the obvious thing to improve voter throughput in Harris County early voting would exacerbate a serious security vulnerability.)

I voted today, using one of the many early voting centers in my county. I waited roughly 35 minutes before reaching a voting machine. Roughly 1/3 of the 40 voting machines at the location were unused while I was watching, so I started thinking about how they could improve their efficiency. If you look at the attached diagram, you’ll see how the polling place is laid out. There are four strings of ten Hart InterCivic eSlate DRE (paperless electronic) voting machines, each connected to a single controller device (JBC) with a local network. These are coded green in the diagram. On the table next to it is a laptop computer, used for verifying voter registration. This has a printer which produces a 2D barcode on a sticker. This is placed on a big sheet of paper, the voter signs it, and then they use the 2D barcode scanner, attached to the JBC, to read the barcode. This is how the JBC knows what ballot style to issue to the voter, avoiding input error on the part of the poll workers. I’ve used the color blue to indicate all of the early voting machinery, not normally present on Election Day (when all of the voters arriving in a given precinct will get the same ballot style).

Poll Diagram
[Read more...]

avatar

What happens when there's no recount possible?

Greetings,

This is my first posting on Freedom To Tinker, so a brief introduction first. I’ve been involved in electronic voting technology issues for about five years, as founder of Virginia Verified Voting where I wrote the law that prohibited purchase of more DREs (i.e., paperless voting machines), as a researcher on the EAC Voting System Risk Assessment, a consultant to the Kentucky Attorney General, an advisor to the DC City Council, and (since I joined SRI International almost two years ago) as one of the investigators on the NSF ACCURATE program. It’s an honor to join this blog.

On to the real topic of this post.

I live in Fairfax County, Virginia. Virginia uses mostly paperless DREs (no VVPAT paper trail). Thanks to the law I helped write (see above), Fairfax County now has both DREs and optical scanners in polling places, and gives voters the choice which they want to use. As a pollworker on election day, my experience was that 80% pick DREs – the reason is subject for a future blog posting.

One of the closest races in the country in last week’s election was for House of Representatives in the Virginia 11th District where I live – Gerry Connolly (D) is the one-term incumbent, and Keith Fimian (R) is the primary challenger, along with five minor-party candidates. Connolly beat Fimian by 12 points in 2008, but everyone knew this race would be closer.

In fact, the unofficial totals as of this writing show Connolly ahead by about 900 votes out of 225,000, or about 0.40%. Close, but certainly not the closest I’ve seen. (The 2005 Attorney General’s race was decided by about 0.015% – yes, just over one one-hundred of a percent.) Virginia law allows for a recount when the margin is under one percent, so a recount would seem obvious.

However, The Washington Post reports that Fimian has decided not to seek a recount. Why? Well, there’s the official reason, but the real reason is two-fold.

First, a large majority of the votes in the 11th district race were cast on paperless DREs, spanning Fairfax County, Prince William County, and Fairfax City (a separate jurisdiction from Fairfax County). I don’t have the numbers yet, but my estimate is that 90% were on paperless DREs. So there’s nothing really to recount.

Second, Virginia law is very restrictive on recounts. For DREs, all you can do is look at the total tapes from election night, and add up the totals again. If the total tape is unreadable, you can print it again. But that’s it, without a judge’s order. For optical scans, you reprogram the optical scanner to count just the race in question, retest it, rerun the ballots, and use the total tape to add to the DRE total tape. If the scanner rejects any ballots (e.g., for write-ins), you can examine those by hand – but you can’t examine any that the scanner accepts, again without a judge’s order. And since the law gives the judge no guidance on why they should allow looking at anything other than the total tapes, it’s unlikely that a judge would make up his/her own rules.

So the result is that a recount is fairly meaningless – it’s really a retally of the totals, not an examination of the ballots, as a recount is meant to be.

Time for Virginia to both replace its DREs with optical scanners AND update its antiquated recount laws, as the Washington Examiner noted in their editorial yesterday.