Posted on 09/14/2006 1:47:32 PM PDT by WmShirerAdmirer
TRENTON, N.J. - A Princeton University computer science professor added new fuel Wednesday to claims that electronic voting machines used across much of the country are vulnerable to hacking that could alter vote totals or disable machines.
In a paper posted on the university's Web site, Edward Felten and two graduate students described how they had tested a Diebold AccuVote-TS machine they obtained, found ways to quickly upload malicious programs and even developed a computer virus able to spread such programs between machines.
The marketing director for the machine's maker — Diebold Inc.'s Diebold Election Systems of Allen, Texas — blasted the report, saying Felten ignored newer software and security measures that prevent such hacking.
"I'm concerned by the fact we weren't contacted to educate these people on where our current technology stands," Mark Radke said.
Radke also question why Felten hadn't submitted his paper for peer review, as is commonly done before publishing scientific research.
Felten said he and his colleagues felt it necessary to publish the paper as quickly as possible because of the possible implications for the November midterm elections.
About 80 percent of American voters are expected to use some form of electronic voting in the upcoming election, in which the makeup of the U.S. House will be decided, as well as 33 Senate seats and 36 governorships.
The AccuVote-TS is commonly used across the country, along with a newer model, the AccuVote-TSx. While Felten wasn't able to test the new machine, he said he thought much of what he found would still apply.
The machine Felten tested, obtained in May from an undisclosed source, was the same type used across Maryland in its primary election Tuesday, according to Ross Goldstein, a deputy administrator with the state's Board of Elections. Goldstein said he couldn't comment on the report until he read it.
Diebold and other machine manufacturers, including California-based Sequoia Voting Systems Inc. and Nebraska-based Election Systems & Software Inc., have been the subject of lawsuits, claiming the machines are vulnerable to hacking and breakdowns that can assign votes to the wrong candidate.
Election officials in some states have also complained.
Previous studies have claimed hacking vulnerabilities with the machines. But Felten claims his study is the first time that an independent research group has obtained an actual machine and tested it extensively.
Felten and graduate students Ariel Feldman and Alex Halderman found that malicious programs could be placed on the Diebold by accessing the memory card slot and power button, both behind a locked door on the side of the machine. One member of the group was able to pick the lock in 10 seconds, and software could be installed in less than a minute, according to the report.
The researchers say they designed software capable of modifying all records, audit logs and counters kept by the voting machine, ensuring that a careful forensic examination would find nothing wrong.
The programs were able to modify vote totals or cause machines to break down, something that could alter the course of an election if machines were located in crucial polling stations.
It was also possible to design a computer virus to spread malicious programs to multiple machines by piggybacking on a new software download or an election information file being transferred from machine to machine, Felten said.
"I think there are many people out there who have the type of technical ability to carry out the sort of attacks we describe here," he said.
Felten said hacking dangers could be mitigated with better software, more restrictions on access to machines and memory cards, and paper receipts verified by the voter.
Radke said Diebold already has implemented many of those things.
I'm not swayed by your anecdote. I described the system I saw and wanted to know what's wrong with it and how could it be hacked. The paper ballot system has been rife with fraud for as long as I know. Are you saying the Riverside County method is more corrupt than the paper system?
Electronic voting could be useful if it were done with the current technology for security. Diebold didn't even do something as simple as what every TiVO does, which is use cryptographic keys to verify the validity of the software being loaded onto it. That would have stopped Felten cold.
Just think of how rediculous it is. TiVO, a friggin' entertainment-related company, thinks more about security than Diebold, which is involved in who will run the country.
Remember, when they "count" the machines kick out certain cards to the reject bin. That's where you find the incomplete or "hanging" chads ~ so you take those cards, face and stack them, then run a straightened coat-hanger down through the whole depth at the location you want.
That way your guy gets all the loose votes. If there was actually a valid vote for someone else, that vote will be spoiled.
Anyone trying the equivalent with an electronic voting machine would need to run a coathanger through the innards and risk getting turned into a crispy critter at the buss.
Here's the problem if you were using the Diebold machines. One staffer at any one location managed to get a minute alone with a machine and infected it. The votes on that machine are now modified. The card from that machine is then collected with the others, and fed into the main machine. Whenever that first machine's card is inserted, the main machine is infected, and all of the votes from the district can be modified without a trace.
Government is.
Dibold sold them what they asked for.
It seems you have a different machine than this flawed Diebold one that will be used in thousands of polling stations across the country.
thanx
Apparently not. Depending on the voting procedures in a precinct, one lone Dim operative with one minute alone with a machine can turn a Republican district Democrat. That's a lot easier than rigging thousands of paper ballots, and only requires one person.
That's the problem. "All you need is" many things that Diebold doesn't do.
Of course, I presume Diebold simply fixes the software holes to prevent this stuff.
Nope. Some internal Diebold memos were leaked a while back, and Diebold tried to have them suppressed. The memos showed that there are flaws in the system, and Diebold just doesn't care.
Your systems sounds like ours. I really like it. There are three verifications, the number of ballots issued, the paper tape generated when the ballot is input to the machine, and finally the ballot itself. This system is very simple, very quick and voter friendly. Why anyone would go to an unverifiable system is simply stupid and makes me wonder about their intent.
That's obvious. Me buying a book at Amazon has more inherent security, yet that's not really an important thing on a grand scale compared to, say, who will be the next leader of the free world. It's kind of obvious to this programer that you might just want to up your security on the system when programming for something so sensitive.
Dibold sold them what they asked for.
Yes, sad. OTOH, in some cases governments asked for accountability and code visibility, and Diebold sued, and won, to avoid the laws. Diebold also violated other laws relating to certification of the machines, like installing software after the certification.
How long do you think it would take to find and alter 5,000 votes? It takes about a minute with these machines using a lock pick and a flash card.
And you honestly think that something like this would go by untraceable and unnoticed? Um, ok.
The modification can easily be done unnoticed. It is traceable only if the software on each machine is later audited, but then the actual vote count could never be known.
Given 10 minutes I could come up with the basic concept for a hardware/software/procedure system that is basically far superior to what Diebold has spent years developing.
Well yeah, given a specific set of circumstances, having a high degree of technical knowledge AND outdated software, then yeah, it's possible. It's much easier for some dem hack to falsify thousands of paper ballots. They've been doing it in Wayne County (Detroit) Michigan for years and it's pretty much down to a sloppy science.
I think they are all pretty much alike. I really dont know. What i do know is that they are each a separate computer, not plugged into any system that sends or receives figures. The only connection with Marylands machines is the electric extension cord that supplies the power.
Read earlier posts about the sneakernet virus vector.
Why doesn't someone build a machine that uses non-rewritable storage devices for the election parameters and the vote records? Consumables costs should still be less than when using paper ballots even if the parties went through the expense of using counterfeit-resistant seals to mark the legitimate media. Indeed, the ability to use higher quality seal markings than would be affordable for authenticating individual ballots would suggest that a properly designed device could be even more secure than using paper ballots (though obviously if dogs, dead people, etc. are allowed to vote an improved voting machine won't fix that).
OTPROM is cheap, readily available, and once a bit is written it generally cannot be altered without visibly physically damaging the chip. Some devices may be erased by high levels of radiation, but tamper-resistance seals could be constructed of a material that would by visibly changed by such an attack.
Any good reason not to do things that way?
Good question. But the article isn't about all voting machines, just Diebold's extremely lax security in theirs.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.