By Wednesday morning, stories were flying all around the Internet--have you looked closely at the results of the primary? There was something strange about the votes, they said, about the difference between municipalities that hand-counted votes and those that used optical scanners. The chatter increased, and by Friday, the New Hampshire Department of State issued a press release announcing that two candidates, Democrat Dennis Kucinich and Republican Albert Howard had requested and been granted a recount, having met the following requirement:
"New Hampshire law, RSA 660:7, provides that "any person for whom a vote was cast for any nomination of any party at a state or presidential primary may apply for a recount." RSA 660:2, IV provides that if the difference between the vote cast for the applying candidate and a candidate declared elected shall be greater than 3 percent of the total votes cast in the towns which comprise the office to be recounted, the candidate shall pay the fees provided in RSA 660:2, III and shall agree in writing with the secretary of state to pay any additional costs of the recount." RSA 660:6 provides that if the person requesting the recount is declared the winner after the recount or loses by a margin of less than one percent of the total votes cast, the fees for the recount will be refunded by the State."
The recounts will begin on January 16, at a time and location to be announced after the state has completed an estimate of the cost and received payment based on that estimate.
The questions revolve around the AccuVote optical scanning system, used by 111 of New Hampshire's municipalities. The scanner, made by Premier Election Systems (formerly Diebold) was one of the machines analyzed by the Ohio Secretary of State's office in a recent review of the voting systems used in that state. The reviewers, a mix of academics and corporate computer security experts, described how the AccuVote works:
"When a voter arrives at a polling place to vote, he or she marks an optical scan ballot with a marking device, such as a pen or pencil. When finished, the voter inserts the ballot into the AV-OS optical scan machine. The voter is given the chance to reject and retrieve the ballot (such as in the case of an overvote) or to accept the ballot as voted. The ballots move from the scanner to a locked box in the base of the scanner. After the polling place closes, poll workers print an election summary off of the AV-OS. Poll workers transfer the AV-OS memory card, defined below, to the board of elections office for vote tabulating using EMPs and/or the GEMS server."
The Ohio reviewers found numerous problems with all aspects of the Premier machines, more so with the DRE model (direct-recording electronic, or touchscreen), but with the optical scanner as well. The servers are easily hackable and the memory cards used in the machines are also susceptible to tampering. Speaking specifically about the scanner, they noted that a ballot could be sent through the machine more than once without any red flags being raised. In addition to purposeful attacks, optical scanners can be affected by other ills--the type of paper or ink could cause a ballot to be scanned poorly or a machine can be miscalibrated, resulting in an incorrect reading of the ballot information it's been given.
ChecktheVotes is one of the websites that sprung up last week, parsing the primary numbers, specifically the difference between the hand-counted and machine-counted votes (the site, created by a Ron Paul supporter, was originally called www.ronrox.com, but the name was changed once it began to receive thousands of visitors). Using numbers from Politico, the site compares the percentage of the overall vote a candidate received by machine counts vs. the percentage received from hand counts. For example, according to the numbers on the site, John McCain received 36.419% of the total votes counted by machine and 39.303% of the total votes counted by hand. This results in a presumed "loss" of 2.884% of the from the machine count.
Obviously, Mr. McCain isn't complaining or asking for a recount. But the site points out other similar losses and gains: Mitt Romney "gained" votes from the machine count, as did Hillary Clinton. Barack Obama showed a "loss" from machine counted votes.
The question, though, is whether these differences mean anything. Is there any reason why the hand-count and the machine count should be exactly the same? The large towns in New Hampshire generally use the AccuVote machine, while the small towns use hand-counts. It would seem that voters in small towns and voters in large towns might have very different concerns, with candidates that appeal in large population centers failing in small towns. Look at the Romney example. Romney did better overall with the machine count vote than he did with the hand count. Was that due to someone hacking the vote in Romney's favor? Did the majority of the machines suffer malfunctions that gave him extra votes? Or was it more about the candidate and the demographics? It's not a huge leap to think that Romney's big business image just didn't play well in small towns, leading to his poor results in the hand counts.
The picture is initially less clear with Clinton and Obama. Looking at the vote town by town, Clinton decidedly won more of those that used AccuVote, winning 66 of these municipalities to Obama's 45. Obama won more of the hand-counts, taking 83 towns to Clinton's 39. In fact, their voting percentages were almost exactly reversed: Clinton won 46% of the hand-count vote to Obama's 54%. Out of the machine counts, Clinton took 53% while Obama won 48%.
Neither candidate is as easily slotted "big city" or "small town" the way we can categorize Romney. However, Ed Morrissey of the Captain's Quarters blog noted last week that this fits with pre-vote trends--Clinton was leading in big cities, which tended to use voting machines, while Obama had been polling better in small towns, where hand-counts were prevalent.
Looking at those numbers for Obama and Clinton, you could argue that the Clinton camp would have just as good a case for complaint against the hand- counting process as Obama would against the machine votes (though, remember, neither of these candidates actually filed the complaint--which, considering the Obama campaign has a lot more money and a lot more to gain than Kucinich, may say something about the whole question). After all, hand-counted votes also can be miscounted and misplaced; they can be made to "disappear."
And therein lies the whole issue. The demand for a recount isn't about the New Hampshire primary--anything short of a result showing Obama winning by more than say, 5% would still put the vote within the realm of a Clinton "comeback" from Iowa. It's about the amount of distrust that voters have in the machine voting systems--machines which studies have shown to be not just hackable, but often poorly conceived and constructed. People who conduct business over cell phones, computers, and ATMs are nervous about electronic voting because they feel that something about the whole process has gone amiss; the voting machines' source codes are proprietary to the companies so any flaws remain hidden except when extensive studies, such as those conducted by Ohio and California are done. Many states bought the systems quickly after the Help America Vote Act of 2002 made electronic voting mandatory, investing millions of dollars in e-voting and finding out about problems afterward. The EAC's oversight of HAVA has been questionable; until recently, the machines' manufacturers selected the labs that tested their voting systems.
Fraudulent voting can occur in a number of ways. There may be bad people counting; there may be bad machines counting. The New Hampshire recount probably won't reveal any widespread acts of coordinated malfeasance, but it's not wrong to double check the process. After all, as Stalin said: "The people who cast the votes don't decide an election, the people who count the votes do."
Or the machines.