Anyone who follows elections knows that non-precinct place balloting modes are growing rapidly in popularity. Many states have relaxed their absentee voting provisions in the last few years, and more are likely to do so prior to the 2008 election.

This makes the issue of signature verification a new and interesting area for election administration. If you have to process tens of thousands of absentee ballots, how do you maintain the integrity of the ballot? How, in short, do you make sure that the right person actually filled out the ballot?

Most jurisdictions with which I am familiar rely on a ballot design that works this way:

1. Part one is the ballot itself, which typically consists of ovals that are filled in and read by an optical scanner.
2. The ballot is placed within a security envelope. In some states, the voter signs the security envelope and this signature is viewable through a windows in the …
3. Outside envelope. This is the actual mailing envelope. In Oregon, you sign this outside envelope, while in Washington State, you sign an inner envelope, and the outer envelope has a special flap that is removed at the elections office, or a special “window” (designed by Pitney Bowes) which is opaque until viewed under a special light.

Ok, now you have the envelopes–how do you check the signature?

Today’s talk by Northwest Forensics was a brief version of the longer training they provide to Oregon poll workers. I’m going to skip the details–what struck me most was that the clear gap between what is considered “forensic” signature checks and what has to happen in an elections office, under the crunch of both time and workload.

Some vendors are marketing technological solutions, however. I looked at three at this conference (I hope to post pictures of these machines a bit later).

Cowart Gagnon showed a ballot sorting machine which takes the envelope, captures the signature, and then sorts the envelope into various bins. They told me they have a signature checking facility also, but I couldn’t see it at the conference. (The machine actually was manufactured by Tritek Technologies–Cowart Gagnon is the reseller for them).

The second machine is the Vote Remote suite, marketed by Diebold. All we are being shown here is the system for signature checks. The system consists of Windows program that allows the election official to set a verification level (margin of error?) ranging from 0 to 100–the higher you set the level, the closer the match much be or the signature gets flagged (presumably to then be checked by a human). I wonder if jurisdictions will tell us the levels that they set and the ‘error’ level of the rejected signatures (it is saved somewhere).

Finally, Pitney Bowes has a demo/powerpoint on their ballot tracking system, which has been adopted by three Washington counties. The sales staff told me they are also developing a signature checking system (we heard about such a system at the Caltech VTP Vendor’s Conference in March), but it sounds like it’s not yet being sold.