Slashdot Mirror
Will the Next Election Be Hacked?
plasmacutter writes to let us know about the new article by Robert F. Kennedy Jr. in Rolling Stone, following up on his "Was the 2004 Election Stolen?" (slashdotted here). Kennedy recounts the sorry history of electronic voting so far in this country — and some of the incidents will be new even to this clued-in crowd. (Had you heard about the CERT advisory on an undocumented backdoor account in a Diebold vote-tabulating database — crediting Black Box Voting?) Kennedy's reporting is bolstered by the accounts of a Diebold insider who has gone on record with his concerns. From the article: 'Chris Hood remembers the day in August 2002 that he began to question what was really going on in Georgia... "It was an unauthorized patch, and they were trying to keep it secret from the state," Hood told me. "We were told not to talk to county personnel about it. I received instructions directly from [president of Diebold election unit Bob] Urosevich...' According to Hood, Diebold employees altered software in some 5,000 machines in DeKalb and Fulton counties, the state's largest Democratic strongholds. The tally in Georgia that November surprised even the most seasoned political observers. (Hint: Republicans won.)
Maryland Governor Bob Ehrlich, a Republican, ran into stiff opposition after (Diebold?) voting machines caused major problems in the state's primary elections this year. Ehrlich wanted to switch to paper-based methods that were known to be reliable. The opposition was NOT from his own party, but from the state's Democratic majority and career bureaucrats.
It's because typically they don't perform exit polls at all precincts, just those that are seen as particularly close. Thus they may make an election-night projection based upon pre-election polls for the majority of precincts, and exit polls at a select portion. If the results in the precincts that weren't specifically exit-polled turn out differently than expected, then the overall election results will differ from predictions made by those exit polls.
Stop by my site where I write about ERP systems & more
I just wanted to point out an interesting method of creating a secure paper trail that came out recently (September 28th 2006) by Ronald L. Rivest of M.I.T's Computer Science and Artificial Intelligence Laboratory. It's called the ThreeBallot Voting System (.pdf format).
The interesting thing about it is that it handles both voter privacy and verifiability without requiring encryption of the ballot. Rather than give a poor explanation because of lack of space (the paper itself is 13 pages long), I encourage interested people to read it.
This tagline is copyrighted material. Please send $10 for an affordable replacement.
You need to do a bit more research before making your claim.
http://www.washingtonpost.com/ac2/wp-dyn/A36425-20 00Nov16?language=printer
It is true that you can certainly tamper an election that's based upon paper ballots. Heck, in 2001 San Franciscans suddenly found ballot box lids mysteriously floating ashore after the November election.
That said, the amount of shady crap surrounding Diebold voting machines is fairly ridiculous. Lets ignore the fact that you have a former CEO, who resigned for allegations of corruption, and who was committed to "delivering" an election to one party. As well as drastically skewed exit polling. All in all, you have a slew of voting machines models that lack the most basic security procedures... such as proper, or any, locks. You also have a fairly complicated voting solution that presents a number of opportunities for a compromise.
"Things are more moderner than before- bigger, and yet smaller- it's computers-- San Dimas High School football RULES!"
I have... and our results were off by a quite a bit.
Why? I can think of a few reasons:
1. It takes time. It usually takes about 10-15 minutes to fill out a good exit poll form. People with less time on their hands - people with steady jobs, people with kids, people who vote in the morning on the way to work, etc. - are much less likely to accept the polling sheet. On the other hand, people with lots of time on their hands - the retired, the unemployed, often younger voters, etc. - are much more likely to fill out exit poll forms. Given that the unemployed are more likely to vote a certain way (generally for the opposition party, whoever that may be), this can lead to skewed data, not to mention other groups.
2. People fill out polls to make a statement. Again, this tends to favor opposition parties, or parties that are less likely to be represented in a region. People like the idea of voting twice.
3. The organization you poll for could determine who answers your questions. Example - "Hi, I'm performing a poll for University X! Could I take ten minutes of your time?" If the person you are trying to poll doesn't like your university's football team, they may not participate. Or, if a poller represents a news organization the person dislikes, a potential pollee (?) may opt out as well.
4. People honestly forget. This doesn't happen so much in presidential elections, to be sure, but on many exit polls people mark their own votes wrong because they forget what proposition x was or who the candidates for a seat on whatever were.
As someone who has worked exit polls before, let me assure you that they're not always accurate and there are a LOT of things that can throw them off.
In any case, though, the CNN exit poll data from 2004 should make the case for a Bush win, if you go by exit poll data alone.
It would seem that seem that RFK Jr and many in the public have a rather myopic memory when it comes to allegations of vote fraud. One would expect that Mr Kennedy would certainly be aware of the controversy surrounding the outcome of the 1960 Presdential election especially since his uncle John F. Kennedy was elected.
Or was he? Rather than Ohio and Florida, that election came down to narrow wins in Illinois and Texas. Both states were Democrat-controlled and rife with allegations of fraud. Did Mayor Daley of Chicago arrange for the dead to vote? Did Johnson's own political machine throw Texas? Like 2004, the answers depend on who you ask.
In his words:
Here in the UK we use old-fashioned paper ballots, hand counted. No tabulating machines, no hanging chads, no technology at all. In a General Election, the polls close at 10PM and the earliest constituencies usually declare their results around 1AM. By 8AM the next morning there are only a few left to declare and the result is known. This is in a country of some 60 million people - there is no reason why it couldn't scale up to the US population. Why complicate things and introduce more potential for fraud?
Exactly. In the 2004 Presidential election, exit poll numbers in key battleground states varied drastically from the actual results. That's extremely suspicious because people often have no reason to lie to unbiased pollsters about who they voted for. According to several statistical experts (who are far more knowledgeable about stats than the average Slashdot poster, and 7 out of 10 people would agree...) the discrepancies were statistically impossible. It was a big, red, flag that something was amiss in Ohio, Pennsylvania, and several other states. Read this statement from Kennedy's first article:
If you have a rational, scientific mind, that's about as conclusive as it gets. It's certainly enough to pique the interest of people like Robert F. Kennedy, who then ask questions like "Was the last election fixed?" and "Will the next one be?" So Kennedy digs around and does some good old-fashioned investigative reporting, he follows the money trail, and lo and behold it leads to disgraced Republican influence peddler Jack Abramoff and several sleazy advocacy groups!
Honestly, it doesn't matter. He could have replaced that claim with "X-type of people don't respond to polls as often as Y-type" (which is almost always true) and his point remains.
Statistical extrapolation can be a wonderful, scientific tool when a couple basic requirements are met: representative, objective datasets and truly random methods. When pollsters and polls fail, it's typically because the analysis lacked these requirements. In many cases, adequately meeting the requirements is impossible. For instance, how do you objectively define people's views on a controversial matter? In other cases, pollsters just get sloppy. For instance, often during elections, pollsters are asked--typically by the ignorant media--to return results before the voting is finished (translation: non-representative dataset). Pollsters who aren't trained properly might also be inclined to interview some types of people more often than others. (non-random methods). And even if everyone involved does everything perfectly, (which itself is nigh on impossible for an operation as large as a national poll) something that everyone seems to forget is that there is still a chance of random error. Even if the p-value is .01 (it is usually .05), that still means that there's a 1:100 chance that the result is wrong due to random variation alone.
The bottom-line is this: the results from exit polling are never more valid than the ballots in the box. Because of the strict requirements proper polling requires, the problem is more likely to be found with the polls rather than the votes--simply based upon the difference in complexity of the math between the two methods alone. (This is one of the few times in history in which Occam's razor legitimately would apply.) Furthermore, if one is willing to accept the possibility of a rigged election (on the basis of the discrepancy, alone), then he or she must also be willing to accept the possibility of rigged polling, which--strangely--is something that nobody ever does.
-Grym
Holy cow. You made some huge assumptions about what you think was assumed when the exit poll analysis was conducted. Members of The National Election Archive Project, the non-partisan watchdog group to whom the the one in 3 billion figure can be attributed, will be the first to tell you that every single factor you just described made its way into their analysis of the polling data. Yes, seriously.
If I were a ruder Slashdot poster, I would have responded with something like "Who the fuck are you? The woman who wrote the white paper on this graduated Cum Laude from the University of Utah with a master's in mathematics and has been analyzing poll data for 10 years..." and so on, but rather than resort to an ad hominem attack, I'll just assume that you replied without taking the time to check the sources that describe in vivid detail how the analysis was performed. Here is a link to the pdf that describes the process that was used. I know reading an 18 page document is not half as easy as just writing a paragraph where you just make random, uneducated guesses about what it contains, but you might want to give it a shot.
I'm astounded that given the accusations and evidence that this may have actually happened, you refuse to believe that it's possible! Point by point:
1) I can't speak to Diebold's willingness to fix an election, but you are vastly overestimating the risk involved. The whole company is not at stake here - more likely, if some vote tampering were discovered, it would "come to light" that a single rogue coder inserted the offending code into a routine security patch. This guy alone would take the fall for the bulk of it, that is, assuming rock-solid evidence (it would probably take a copy of the actual offending source code, since all other evidence of foul play evaporates in to the papertrail-less void) ever came to light. I believe people have already shown how easy it is to write a self-deleting virus that would remove all evidence of itself as soon as it did its work. (it's easy enough to make something get rid of every trace of itself even in a bloated mess like Windows; it's child's play when your company controls the design, security, and handling of the operating system, hardware, and software at every step along the way). If I had to guess, I would say that we're talking well under a 1% chance of discovery if most of the knowledge of details was confined to the top tier of the company. Even supposing this was discovered, after offering up a patsy, the company would probably just lose its voting machine business and continue as usual with its other stuff. Diebold was getting along fine before getting into the voting biz, they'll do fine if they're kicked out of it, too. Depending on the price, it could well be a very profitable (risk vs. rewards-wise) decision to throw an election to the highest bidder.
2) I'll agree, I don't know how easy it would be to get coders inside the company to knowingly agree to this level of risk. More than that, I would worry about the possibility of the involved coders leaking the fact that they did this so as to push responsibility up the chain if it looked like evidence was mounting against them. If I was to run such a scheme, I'd make sure to go outside the company for this bit, possibly by going to whoever wanted to buy the election to find someone that they trusted - anyone scummy enough to buy an election knows where to go for something like this. It would be easy enough for a high level Diebold exec to obtain API details to hand over, and any knowledgeable programmer could figure out what to do with them in a little time for the right price. Also, it's possible that the specs for the vote-shifting code could be phrased in such a way that the programmer didn't even realize what they were doing - I don't know how Diebold's software works, so I can't really comment any more on that. One easy way would be to ask a programmer to do a security audit, and prove any flaws by writing exploits for them. They hand over the info happily, assuming the flaws will be fixed, only to have them abused instead.
3) From the article, it appears that the people who installed most of the Diebold patches had no idea what was on them, so probably wouldn't face much exposure. And they did agree to install them, despite being suspicious about what was on them, so I think that point is proven - if given a malicious patch, they would install it.
4) This has been thoroughly addressed before - in some cases, all it takes is altering a single machine a local election result. Even for a national election, the margins are so small these days that a few thousand votes here and there really can shift