How to Rig an Election

That’s the difference between you and me. You want to lose small. I want to win big. – Vladimir Putin

Алтынного вора вешают, а полтинного чествуют

The above quote isn’t actually from Putin, although it seems oddly fitting given recent international events (I’m sure Harvey Spectre  would appreciate the comparison). This post is written as a response to John Ibbitson’s recent G&M article on How to Rig An American Election. John outlines what he sees as three necessary conditions: Corrupt the pollsters, corrupt the media, and corrupt the vote itself. I don’t disagree with John, and believe that Clinton will probably win on Nov 8th, although it will be closer than many predict. But, for the sake of the argument, a take on how to rig the U.S. election. I will emphasize that this is not a how-to guide, but rather a recognition that we need to be aware of vulnerabilities. Vigilance is the price of freedom, and the electoral system is too critical to take lightly.

I’ve been struck by the overwhelming insistence that the election cannot be rigged. Pundits and public officials alike have taken every opportunity to assure American voters that the process will be fair and unbiased. Could you imagine a better narrative than Hillary Clinton attacking the legitimacy of a Trump win?  Use the media, and use the polls to sell the idea of an inevitable Clinton win. That way, if you can deliver a victory for Trump, the chaos will be all the more devastating to American public life.  

The Modern Campaign

My introduction to the theory of modern political campaigning came from a democratic campaign consultant and author named Hal Malchow. His contribution centered around the idea that we need to quantify characteristics of voters, and use this information to target campaign efforts.

Perhaps the best book currently available on the topic is Hacking the Electorate: How Campaigns Perceive Voters. This expands on Malchow’s work, and goes into the nitty gritty of how campaigns can use statistics to target voters. Would also highly recommend The Victory Lab: The Secret Science of Winning Campaigns which goes through how American campaigns have made use of these tactics in the field.

In the simplest terms, campaigns assign a series of scores to each voter. These are built using statistical models of varying complexity. They represent the probability that a voter will complete a certain action, or respond to a certain issue. For our purposes, there are three main scores we’d want to calculate. Probability that someone will vote for Hillary, Probability that someone will vote for Trump, and the probability that someone will actually go and vote.

The final of these three is the most interesting, as you see the abbreviation LV in many public polls. This often refers to ‘Likely Voters’, or those a pollster believes will vote on election day. An imperfect stand-in, but a necessary one, as no one knows the true make-up of the final electorate.

A big part of the campaign is shaping the electorate through persuasion and mobilization. Not every demographic votes with the same enthusiasm, a point made clear in the recent EU referendum in Britain. It doesn’t matter if 73% of voters under 24 support your cause if only 50% of them actually show up. Politics is a numbers game, where the fates of nations can be decided by the smallest of margins.


Rock The Vote

Last fall, the Bernie Sanders campaign was punished by the DNC for allegedly and improperly accessing Hillary Clinton campaign data. The Sanders campaign tried to pull voters data that met certain criteria – high Clinton scores, low Clinton scores, etc. By pulling lists of people that weren’t fond of Hillary Clinton, the Sanders campaign could focus their efforts on mobilizing people that would be receptive to Sanders. Much of the democratic infrastructure is built on a few national databases that have information on millions and millions of Americans and their political perspectives.  

Let’s take a look at 4 hypothetical voters, and how they could be treated by a foreign entity trying to cause mischief. 

Voter Pturnout PClinton PTrump Biography
Nathan Clerk  75% 20% 5% Nathan is a college educated banker in rural Pennsylvania. He voted for Rubio in the primary, and would rather have Hillary as president than Trump. He is very likely to vote, but could be open to a third party candidate. 
Charlie Hamilton  25% 5% 60% Charlie is from northern Florida. He works for himself, and rarely votes. He doesn’t like Clinton, but isn’t sure if he will go and vote in this election.
Paula French 100% 40% 20% Paula is a lifelong Ohio democrat. She will definitely go and vote, but is turned off by Clinton’s shift towards the center.  She backed Sanders in 04, Obama in 08, and Sanders in 16. She will definitely vote, but is considering voting for Jill Stein – just to send a message.
Elizabeth Ross 50%  80% 10% Elizabeth is a suburban mom in northern Ohio. She is busy, and while she is a big fan of Hillary, she isn’t sure if she wants to go and vote – especially if the news is true, and Trump has no chance of winning. She will go vote if she can fit it in between soccer practice, her home business.

If someone is likely to vote for Hillary, but not all that enthusiastic about going to vote (Elizabeth), make it as difficult as possible for them to get to the polling station. 

If someone is likely to vote for Trump, but not likely to go vote (Charlie), give them a push to get out and vote.

If someone doesn’t like either (Nathan/Paula), then make sure they vote for someone else, or don’t bother to vote at all. 


A Monkey Wrench into the Hillary Machine

Hillary Clinton has built an incredible ground organization. When pundits say her ground game is worth 2-3%, they mean that her campaign team can turn those 50s like Elizabeth into 60s and 70s. There is an extensive academic literature that discusses the effects of campaign activity. While 2% may not seem like a big deal, in tight races like Florida in 2000, it makes all the difference. 

Campaigns are getting better at cyber security, and track every time someone accesses a system. If the U.S. authorities are worried about the massive outages from yesterday’s cyber attacks, campaign databases may not get the same attention as financial firms, or emergency response systems. At last count, Hillary For America has 5,100 paid staff on the ground across 15 battleground states. Every local office is building mobilization plans for election day. Every one of those plans is dependent on a common database that is already shown itself to be vulnerable. Neutering this machinery could be as easy as changing a few columns in a database. 

A backdoor into the Hillary For America systems, built under the cover of a major cyber attack, could allow a motivated attacker to gain access to the information they’d need to suppress Hillary supporters. Perhaps even worse, randomizing & replacing existing voter score information could mean the campaigns focus their efforts on mobilizing the wrong voters. There would be a wonderful bit of irony to have Hillary’s ground organization mobilizing Donald Trump voters. These mistakes happen all the time – staffers are human, functioning on caffeine, alcohol and purity of mission. Sometimes the wrong lists are called with the wrong message. 

In this information would be summary statistics and dashboards that outline which geographic areas have the highest concentrations of strong, and marginal Clinton supporters. You don’t have to break into a voting machine itself if you can shut down parts of the power grid. Even intermittent power failure in heavily democratic areas would serve this purpose. Thinking back to an old Simpsons episode where the machines revolt, why not cause accidents in democratic boroughs by turning all the stoplights green. If the voting machines are malfunctioning, elections officials will revert to paper ballots. But, if these are not on site and printed, web enabled printers are vulnerable to shutdown or improper access. How long does a Clinton voter wait before going home, fed up with the whole system? Likewise, transit systems in major cities, and wastewater are tied to monitoring stations that ensure proper functioning. Every system is secure until it isn’t.  

These types of attacks are certainly not impossible, especially to a nation-state that has proven its willingness to undermine the legitimacy of an election. If Trump wins Pennsylvania amid chaos in Pittsburgh and Philadelphia, his administration begins with weaker public support than George W Bush. If Trump poll monitors are watching out of county precincts, and chaos erupts, there is real potential for violence on the ground. 

Unlikely? Yes. Impossible? No

Campaigns are a game of inches, and while the Democratic Party has an incredible breadth of technical talent, they most definitely lack the resources and motivation of the SVR. 

1) The Russians aren’t afraid to play the long game. Whether place deep cover operatives in the U.S, or building propaganda networks across the globe, they have demonstrated a willingness to plant seeds that won’t pay off until years down the line.

2) They will put everything on the table when they think it matters.  (Не давши слова — крепись, а давши — держиcь.)  If you are going to commit to something, be darn sure to carry it through. There is a willingness to double down and support friends (or at least strategic interests). Assad can attest, the silovki play for keeps.

Cheery thoughts for your Sunday morning.

2 Comments on "How to Rig an Election"

  1. The objective does not need to be a finely tuned attack to change election results. It could be bluntly focused simply to create chaos and attack the faith and confidence the American people have in their elections systems while sowing doubt about the validity of U.S. elections.

  2. One frustration with following the current election is the paucity of polling on Congressional seats. The presidency is surely all but in the bag; the Senate race is looking close but promising for a Democrat majority. But what of the House?

Leave a comment

Your email address will not be published.