The most important results of last fall’s election are just now in, and dozens of county voting registrars who invested hundreds of millions of dollars in high-tech voting machines are the losers, exposed as careless at best, irresponsible at worst.
Only a few California voters realized it during that election season, but the trustworthiness of this state’s entire voting process was at stake. The candidate who favored questioning all electronic voting machines won.
And soon after Democrat Debra Bowen took over as secretary of state last winter, she ordered a “top to bottom review” of electronic voting and vote-counting systems. The results expose the failures of the registrars: Every voting system they’ve bought since 2000, except those whose maker didn’t readily submit them to testing, proved corruptible.
No one would know this today if the former Republican Secretary of State Bruce McPherson had won last year. Despite serious questions about machines made by Diebold Election Systems, Hart InterCivic, Sequoia Voting Systems and Election Systems & Software (ES&S), McPherson never seriously examined the possibility that votes cast on them could be altered.
Voting-reliability activists claim that has happened in states like Ohio, where some precincts reported more votes cast in 2004 than voters registered in those precincts, and Florida, where a judge has ordered a new election in one congressional district because of machines that may have been hacked.
McPherson insisted all systems used here were reliable. He said the practice of manually and randomly checking one percent of votes cast was enough to tell whether votes had been manipulated. Registrars loved him.
For one thing, he accepted their common practice of sending machines home with precinct workers well before elections. McPherson contended “tamper-proof” seals placed over the slots on those machines assured they could not be corrupted while on such “sleepovers.”
Bowen questioned these assumptions. Her questions have now been answered. Neither machine makers nor the registrars who are their loyal customers like the answers.
For two University of California teams found after a full month of testing for Bowen that all systems submitted to them could be penetrated by hackers.
They reported being able to get direct access to data in Diebold machines. They showed how a wireless device could be plugged into the back of a Diebold server, with potential for manipulating numbers.
On Sequoia systems, testers were able to replace the standard boot loader with a “malicious” one.
Meanwhile, ES&S, maker of the InkaVote system used in Los Angeles County, by far the state’s largest source of votes, submitted machines and material too late for the tests, raising suspicions about whether that company sought to hide flaws.
All of which means that if McPherson were still around, registrars wouldn’t be whining now about “chaos” or the expense of carrying out Bowen’s corrective edicts. Nor would they be worrying about facing questions for months to come about how they could spend so much on machines that can be corrupted. It won’t do for them to note that paper ballot boxes can also be stuffed; touch screens and other machines were supposed to solve that kind of problem.
The machine makers, of course, never wanted any tests. They said no voter could ever get the kind of unfettered access to machines given the testers. They said no hacker ever has as much access to computer source codes.
“None of the attacks described…are capable of success,” said Steven Bennett, a Sequoia executive. “All would be prevented or detected through use of [a paper trail] and legally sufficient audit.”
By that, he meant the one percent audit McPherson used and the paper trails produced by all voting machines, something Bowen sponsored as a state senator over objections from McPherson.
But paper trails have never been used to check a disputed election result in California, primarily because they’re so expensive to count. A challenger must put up $50,000 for a paper-trail count in any state legislative or congressional district, forfeited if the original outcome stands up. So no one has demanded a full paper trail count in a state race since electronic voting began. Which makes paper trails almost useless.
The objection that voters do not get as much access as Bowen’s hacking team also doesn’t stand up. Her testers unscrewed panels to get around tamper-proof seals. They got inside virtually all machines. Any precinct worker who stores machines in closets or garages has as much opportunity as they did. So might county staffers wishing to sneak-attack machines in storage between elections.
Most suspicious election outcomes have suggested such insider interference, not tampering by voters at actual polling stations.
The UC tests were hasty and did not cover as much ground as a more leisurely effort could have. But the results and Bowen’s subsequent orders for hand counting paper trails and beefed up voting machine security are a good start toward the goal of eliminating doubt from election results, even if it all means slower counting of votes.
