December 18, 2003
Recount!

When you lose by 27 votes out of over 36,000 cast, asking for a recount should be expected.


Houston City Council candidate Jeff Daily, who narrowly lost in a runoff election to Pam Holm, on Wednesday filed an anticipated recount request.

After the Dec. 6 runoff, Daily trailed Holm by 30 votes in the race to replace Bert Keller in west Houston's District G.

The margin narrowed to 27 votes last week after ballots mailed by military personnel and civilians overseas were counted. According to final but unofficial results, Holm had 18,411 votes and Daily had 18,384.


All right, so how is this going to work?

Daily said he did not question the integrity of the county's $25 million electronic eSlate voting machines, but that the narrow margin compelled him to make the request. The election was the closest council race since 1999, when Mark Goldberg narrowly beat Maryann Young in a runoff.

David Beirne, a spokesman for County Clerk Beverly Kaufman, said the recount will be done Friday.

County officials will repeat the Election Day vote counting procedure, feeding eSlate memory cards used at District G precincts into a main computer, which then will tally the votes.

Daily said he wants ballots submitted by mail to be reviewed by election officials rather than scanned into a computer.

"I suspect the major part of the correction, if there is correction, will be from that group," said Daily, a businessman.


So we've got the eSlate memory cards, and the (presumably) paper absentee ballots. Daily's faith in the computer voting machines, coupled with his skepticism of the computer counting machines, especially given the relative maturity of the two technologies, is a little odd to me. Wouldn't you at least like for the eSlate memory cards to be inserted into a different computer than before, just to see if it gets the same result? Wouldn't you like something more than Beverly Kaufman's personal assurances that eSlate glitches Just Can't Happen? I'm sure I'd want to doublecheck the 95% or so of the votes that those memory cards contain in addition to the so-last-century paper ballots. Wouldn't you?

What bugs all of us critics of these black box voting systems is that there isn't a way to doublecheck them. With hardcopy systems, whether punchcard or optical scan or whatever, you could always perform a simple sanity check by taking a handful of ballots, totting up the votes manually, then feeding them into the counting machines to see if they give the same answer. Now you can have some confidence that a subtle error hasn't crept in somewhere. If all you're doing during a recount is exactly what was done before, without any way to verify that you're getting what you should be getting, then what's the point? You know what answer you're going to get, but you can't say with any more certainty that it's the answer you ought to get.

So go ahead and recount the absentee ballots. Who knows, with such a small margin of victory, even that paltry sample might yield a sufficiently large difference this time to change things. If that does happen, though, maybe Pam Holm will express more doubt about the eSlate totals when she demands her recount.

Posted by Charles Kuffner on December 18, 2003 to Election 2003 | TrackBack
Comments

He should also check to see that the number of votes per precinct match the number of people who signed in to vote. A difference can be explained (e.g., the "fleeing voter") but it can be looked at. The poll book is the closest thing to a paper trail with the current system.

Also, no mention in the article of the challenged/provisional voters. I wonder why not. They could at least tell us the total number of challenged voters, so we could know if it is more or less than the margin of victory.

Posted by: Rob Booth (Slightly Rough) on December 18, 2003 11:01 AM

County Clerk Beverly Kaufman appears in a testimonial... essentially an ad... for eSlate on the Hart InterCivic web site. When Kaufman subsequently praises the accuracy of eSlate systems, it is hard not to perceive a conflict of interest.

This Chronicle article mentions investigations (not, by the way, initiated by our county clerk) that indicate major security flaws in eSlate systems. David Beirne, Kaufman's spokesman, said, "We don't see reason to take a second look or question the integrity of the system." What is wrong with this picture?

I'm not saying there is necessarily malfeasance, or an actual security violation, in this instance. I am saying, however, that the appearance of a conflict of interest does not enhance my already low confidence in eSlate and in our county clerk.

Posted by: Steve Bates on December 18, 2003 3:51 PM

I've looked at the Ohio SoS report and had some comments that I posted on my blog. The Chronicle article you mention has a couple of moments wherein people demonstrate their lack of familiarity with the system.

Here:

A recent review, however, of voting machines found that the eSlate system has numerous potential security weaknesses of its own, including lack of encryption to protect election data and passwords and the ability for a hacker to shut down polls early.

A "hacker" is used in popular language to describe a person who accesses a computer remotely and bypasses/breaks through security measures. (Many people prefer the term "cracker" for these folks.) A "hacker" can't access the eSlates' Judges Booth Controller (JBC) in Harris County because the JBCs aren't connected to anything external. Not connected to a modem, no network card, just the eSlates.

Here's another interesting moment:

Wallach argues that even though researchers couldn't access the system doesn't mean it can't be done. He said someone could tamper with the system's memory card and alter or delete votes without encryption. Election officials would not be aware of the breach because the system doesn't possess the technology to detect such tampering.

The tamperer would have to physically access the JBC to accomplish this, which makes it an issue of physical security or election judge integrity. These issues were present under the punch card system and will be issues under any voting system.

Posted by: Rob Booth (Slightly Rough) on December 19, 2003 6:03 PM