Carlos Guerra: Congressional recount? Hope there are no criminal minds

Web Posted: 03/28/2004 12:00 AM CST

San Antonio Express-News

It has been years since a major local political race ended as closely as Henry Cuellar's challenge to U.S. Rep. Ciro Rodríguez, which the incumbent won by 145 votes.

With the recount, attention has focused on Bexar County, the district's largest concentration of voters — but whose turnout was abysmally low.

But the thin margin also is due to the Republican-enhancing congressional remapping, said St. Mary's political scientist Henry Flores, who over three decades has testified for both Republicans and Democrats in redistricting challenges.

"The Republicans went after the congressmen that gave them a hard time (and) they weakened all the Latino districts except (U.S. Rep. Henry) Bonilla's, whose district was strengthened," he said, by replacing blocs of Mexican American voters with Anglos.

"I saw (the new map) and figured that several Latino congressmen could lose in the general election," Flores recalled, "but when Cuellar filed, I saw that Ciro might lose in the primary.

"Historically in recounts, the winner usually gets a bigger lead, but I said this election was too close to call, and now the recount is also too close to call."

Asked Thursday how the recount would be conducted, Bexar County Elections Administrator Clifford Borofsky said "very carefully." He explained that the Democrats will decide which "mail ballots, provisional ballots and election-day electronic ballots" will be reviewed. The latter, he said, are "ballot images, which are essentially a computer listing" recorded by each machine.

I told Borofsky about my 2002 interview of William Sterner, a Tampa, Fla.-based embedded programmer for 30 years who is very familiar with the technology employed by electronic voting machines.

Citing many cases in which bad programming produced massive miscounts, Sterner concluded that electronic machines are only as infallible as their human programmers. He said that without a voter-verified paper trail, errors can't be known, much less corrected.

"We test (all voting machines) beforehand and we test them afterward," Borofsky said. I then related that Sterner had said that someone like him could easily program voting machines to, for example, make every 10th ballot cast for one candidate count for an opponent, while making the computer screen show the vote was tallied correctly.

What's more, if a miscount was programmed to begin 30 minutes after the polls open and stop 30 minutes before they close, errant machines would go undetected, Sterner said.

"This election, we had 191 ballot styles," Borofsky said, "so you would have to know every one, steal every machine and return them without getting noticed."

But ES&S — the firm that sells these machines — does all of the programming, and when we spoke, Borofsky wasn't sure who would pay for an ES&S expert if one is needed for this recount.

"The theory of what (Sterner) is suggesting is probably not incorrect, but it wouldn't be easy," Borofsky said. "But maybe I'm just not using my criminal mind."

Let's hope everyone is like Borofsky. But just in case, why not join the states that require electronic voting machines to generate a voter-verifiable paper trail.