Saturday, February 24, 2018

voting machines and elections

Computer scientists have been running around with their hair on fire about the US voting machines ever since they were first deployed. The charitable interpretation is that the voting machine companies have a touchingly naive faith in 'computers',

A just machine to make big decisions
Programmed by fellows with compassion and vision
We'll be clean when their work is done
Donald Fagen

But the companies clearly don't understand, or do not wish to understand, the first thing about security, chains of custody, or auditability.

Time and attention conservation notice: a better summary is available at the NYRB, written by @jennycohn1.

From computer science Professor Alex Halderman testifying to the Senate intelligence committee - with footnotes, citations, eight-by-ten colour glossy photographs with circles And arrows and a paragraph on the back of each one,
Ten years ago, I was part of the first academic team to conduct a comprehensive
security analysis of a DRE voting machine. We examined what was at that time the
most widely used touch-screen DRE in the country, and spent several months probing
it for vulnerabilities. What we found was disturbing: we could reprogram the machine to
invisibly cause any candidate to win. We also created malicious software—vote-stealing
code—that could spread from machine-to-machine like a computer virus, and silently
change the election outcome.

Vulnerabilities like these are endemic throughout our election system. Cybersecurity
experts have studied a wide range of U.S. voting machines—including both DREs and
optical scanners—and in every single case, they’ve found severe vulnerabilities that
would allow attackers to sabotage machines and to alter votes. That’s why there is
overwhelming consensus in the cybersecurity and election integrity research
communities that our elections are at risk.
Princeton computer science professor Andrew Appel who can hack an election in seven minutes without breaking a sweat,
If this century has shifted our trust from away from our neighbors toward machines, it might be time to switch back again. Eight countries in Europe that once flirted with digital voting have seen six go back to paper; Britain counted its Brexit votes by hand. Even if the vote were never hacked—and it is an exceedingly implausible event—the remotest possibility is an albatross on democracy and a boon for mischief-makers, and not just the cyber attackers. Trump’s most recent jujitsu—pointing out that by virtue of the fact that the election is hackable, it could be rigged against him—illustrates this risk. Technology has amplified not only the threat of hacking, but the threat of a hack.

The Princeton alums can warn us—but they can’t protect us. “We are in a collision-course between the technology we use in election administration and the growing reality of politically motivated, statelevel cyberattacks,” Halderman tells me, arm propped on his red office chair, sunlight pouring through his westward window. “We sit around all day and write research papers. But these people are full-time exploiters. They’re the professionals. We’re the amateurs.”
recent article from the NYT covers much the same ground, though not as well.

There are Senate and House bills to fix this, with the remedies that are obvious,
- Replacing insecure paperless voting systems with new equipment that will generate a voter-verifiable paper ballot;
- Implementing post-election audits of paper ballots

However in this administration I think it unlikely either of these will pass.

Now, get your tin foil hat on. Responsible journalism would have covered these stories, but since we don't have that, they sound like conspiracy theories. The evidence is to my mind quite persuasive.

Voting machines in the 2016 election gave results that are not easily reconciled with polls and exit polls.
We use our data to explore the claim that counties with electronic voting exhibited different voting patterns than their paper peers. What we find is definitely troubling: in some of the swing states, and specifically in states that were projected to vote Democratic at the top of the ticket, those with electronic voting had a decrease in the percent of the total vote going for the Clinton-Kaine campaign, and an increase for the Trump-Pence campaign. We try to determine if this is spurious by checking for patterns in other places with electronic voting, as well as during the 2012 election. We only find this correlation for swing states during the 2016 election.
Craig Wright, a baseball statistician, found the election peculiar and came to a similar conclusion
Exit polls are no longer just polls of voters exiting polling places. Edison Research, which conducts the exit polling for the major news outlets, also polls voters by phone prior to election day, looking for feedback from the rising group of early voters to be added proportionally to their polling place data. 
The CNN exit polls — which were done by Edison Research — covered 28 states that accounted for 411 of the Electoral votes. Their exit polls correctly predicted the winner in the recorded vote in 24 of the 28 states. All four misses were swing states, and in all four cases the exit polls had indicated that [D] had won, but then the official vote count went to [R]. The four states were Florida, Pennsylvania, North Carolina, and Wisconsin, a group that accounts for 74 Electoral votes. While the official vote count gave [R] a whopping 90 Electoral votes in the swing states, if the exit polls were translated into Electoral votes, that would have completely turned that result around with 93 Electoral votes of the swing states going to [D].
Among the seventeen states using voting machines with no audit trail so an effective paper audit cannot be done, only two are swing states. Unfortunately, they were the two swing states with the most Electoral votes, Florida and Pennsylvania.
This has a precedent, too - the Presidential Election of 2000 was the first time in 112 years that the result of the popular vote and the Electoral vote disagreed. The difference here was the votes in Florida were not counted, which was oddly enough a primary motivation for the voting machine debacle and its consequences.

In 2004 something very strange happened in Ohio,
Connell ran the private IT firm GovTech that created the controversial system that transferred Ohio's vote count late on election night 2004 to a partisan Republican server site in Chattanooga, Tennessee owned by SmarTech. That is when the vote shift happened, not predicted by the exit polls, that led to Bush's unexpected victory.
The Georgia special election last year was most likely hacked as well - but we can't know since the machine was wiped clean.
Georgia’s election system was sitting insecure on the internet for months and was easily accessible by hackers. The problem was discovered ahead of time and the state was taken to court in an effort to prevent them from using the unprotected system for the special election between Karen Handel and Jon Ossoff.  But the election was held anyway.
A computer server crucial to a lawsuit against Georgia election officials was quietly wiped clean by its custodians just after the suit was filed, The Associated Press has learned.
Update November 2018: Georgia is doing it again, wiping the servers before the election results are certified. Something is rotten in the state of Georgia.  Also see @jennycohn1 on Medium, with a short summary of recent election history.

Trump can make allegations but I have not seen anywhere any evidence for his allegations. There is evidence for my theories. Ha.

When there is no audit trail the only available check is the exit polls. There's no proof but the evidence is highly suggestive. I don't know of a single instance where exit polls predicted a substantial Republican win but the election went Democratic. Examples of exit polls predicting a Democratic win for an election won by Republicans are legion. That last example is a foreshadowing of this year's Georgia state - chairman of the voting machine company (R) went into the election with polls showing a dead heat, but won by 15 percentage points. There has been considerable fine-tuning of the election machinery since then, allowing for plausibly deniable narrow wins.

More generally, the elections are no longer representative of the people who vote.
Exit polls in the same election that produced this massive GOP triumph had this finding: 55% of the very same voters had an unfavorable opinion of the Republican party compared to 49% for the Democratic Party.   Conversely, 47% viewed the Democrats favorably vs. 40% for the Republicans.

In general through the last century, the party that has controlled the House also got the most votes. In the 2012 election, something unusual happened: Democrats won 50.59 percent of the two-party vote – but just 46% of the seats in the House of Representatives. The Republican candidate for President got 47% of the vote and his party controls at least 55% of the seats. The Democratic candidate got 48% of the vote, and yet the Democrats won 44.6% of the House seats.
The state of elections is well summarized in this letter to the London Review of Books,
Half of the American voting population doesn’t vote at all – namely, the poorer half. Some countries hold elections on weekends or ad hoc holidays; US elections are on a Tuesday, following a 19th-century farm schedule. Americans with no job security working multiple jobs with no breaks often have no time to vote. If Americans have been convicted of felonies, they are in many states disenfranchised for the rest of their lives.

Most of those who are entitled to vote in the US and who have the leisure to do so will take part in a primitive winner-takes-all system of electoral districts that the Republican Party has systematically manipulated. They will vote on electronic voting machines with minimal electronic security, purchased by Republican state governments from Republican-donor equipment suppliers, machines that routinely return Republican candidates to office even when polls show a wide lead for the other party – this may inform the Democrats’ ‘loss of nine hundred seats in state legislatures’.

This was the state of the system which nevertheless, as Bromwich says, ‘voted for Obama twice’. But then things got worse. In 2013, Chief Justice John Roberts effectively terminated the Voting Rights Act of 1965 that had made it easier for black people to vote. Many states introduced racist voter-ID requirements that had been blocked for fifty years, and even in states that did not pass new laws, election officials were emboldened to invent purported requirements and refuse minority voters their rights.

Benjamin Letzler
Oberursel, Germany