False stories, currently referred to as fake news, have been a part of many US elections, as far back as the Adams-Jefferson race for the presidency when Adams was accused of being almost a hermaphrodite.
2016 was no better. The specter of fake news lifted its head in the last election and appeared on the web and in social media in near epidemic proportions. But was the mass distribution of fake news actually a substantial influence on the election outcome? Trump backers point to a 2016 post-election Stanford/NYU study which found that fake news most likely had no effect on the election. According to the study, of the total respondents to the survey, only 8% actually read the fake news they came across and even less believed what they were reading.
Although the Stanford/NYU study would seem to back up Trump supporters’ claims that fake news had no influence on the last election, I feel confident that certain players, primarily the Russians, used fake news to target rust belt, swing state voters with near surgical precision and persuade them not to vote for Hillary Clinton.
Study’s authors (Hunt Allcott, Economics, New York University and Matthew Gentzkow, Economics, Stanford University) started their work by producing a database of 156 unique fake news articles gleaned from snopes.com, politifact.com, and BuzzFeed.com. The “news” stories consisted of 41 anti-Trump/pro-Clinton and 115 anti-Clinton pieces, reflecting real-world anti-candidate item proportions. They then created a SurveyMonkey.com survey to gather participants’ recognition and rating of 15 fake news headlines randomly selected from the database on an individual by individual basis.
Authors went to great pains to assure reliable results. They developed accuracy ratios to offset participants’ ability to recognize actual fake news shared from the placebo fake news (fake “fake news”) they created. They performed re-weighting operations to assure responses represented a broad, national cross section of Americans who voted in the last election. Of 30 million “opt-in” SurveryMonkey.com members, 1,208 contributed. Demographics were asked of the contributors, categories included political affiliation, income, age, education, web, and social media use.
Impressive and credible to say the least, and quite frankly, I agree with their conclusions. That on a broad national basis, fake news had very little influence, if any, on voters. But no one needed a survey to see that, simple common sense would tell you otherwise. Anyone who sought to use fake news as a means to sway voters against Clinton failed miserably as well indicated by her 3 million popular vote lead over Trump. Failed, of course, if influencing the electorate against Clinton on a national scale were the intent. But what if that weren’t the intent?
What if the intent was simply to convince a minimum number of voters in key swing states with relatively high numbers of electoral votes not to go for Clinton? What if instead of millions of voters on a national level, you only had to sway tens of thousands of voters in three or so key states? States where the two top candidates were running a close race with Clinton just a few points ahead of Trump. All the most opportunistic circumstances brewing up a “perfect storm” for electoral manipulation. The Russians saw it. They spotted all the conditions needed to produce the best atmosphere for swaying the most effective number of voters within a focused region of the electorate and they took advantage of it.
A little over a month ago, Chairman of the Senate Intelligence Committee, Senator Richard Burr (R) and committee ranking Democrat, Mark Warner, joined together to inform the press that the Kremlin had hired over a thousand workers to produce anti- Clinton fake news for use in the swing states. They further claimed that the Russian also had a near thousand trolls working out of a location controlling botnets to disseminate the anti-Clinton falsehoods to specific areas. These commandeered computers of unsuspecting voters used their Facebook and Twitter accounts to spread the falsehoods to followers under the guise of a familiar user.
Just a few days later, Clint Watts, an ex-FBI agent, testified to that same Senate Intelligence committee that Russian operatives had used “armies of Twitter bots to spread fake news using accounts that seem to be Midwestern swing-voter Republicans.”
To explain how the Russians recognized the elements of the optimal situation in which to influence the right amount of voters away from Clinton and take advantage of it, I’ll use an emulsion as an analogy. An emulsion is a substance that doesn’t “exist” until all its components are at the right state at the right time.
An easy emulsion for my analogy is French buttercream frosting. Buttercream is made of four simple ingredients, water, sugar, eggs, and butter. However, you can’t just add all the ingredients into a bowl and whip it up into frosting.
To make buttercream, you first beat the eggs to a certain thickness, then boil the sugar and water to certain temperature, slowly add them to the eggs while mixing it continuously until room temperature when you will add the butter, softened and in one tablespoon chunks.
At first, the mixture looks like a messy bowl of butter egg soup, but you keep beating it until it reaches the right temperature and conditions and suddenly you have buttercream frosting in a matter of seconds. Remove any component for the process and you get nothing as each part is essential for the success of the others and the whole.
So the same for the Russians and a perfect set of conditions to use fake news and sway voters against Clinton. You have your necessary components: a close race in a swing state, an over-confident Clinton taking her lead in the polls for granted and the Russians recognizing the situation and precisely targeting the localized voting block with anti-Clinton fake news. Take any one component out of the combination and it fails and the Russians lose the advantage and the opportunity for success.
Reviewing the election results of three key swing states, Wisconsin, Michigan and Pennsylvania, we see how just a relatively small amount of the voters of each turned the win over to Trump. In Wisconsin, Trumps beats Hillary by 27,257, a mere 0.9% of votes cast. Out in Pennsylvania, Trump wins 68,236 more votes or a meager 0.6% of total vote. And in Michigan, Clinton loses by 11,612, a scant 0.2%.
Remember earlier in this post where I mention Stanford study’s conclusion that fake news was read and believed by less than 8% of the national electorate? All these winning percentages fall far below the 8% benchmark, coincidentally. And those are the percentages of votes Trump won over Clinton votes, the actual percentage required to win those elections is slightly more than half the candidate percentage difference.
The Russians only required a change in the hearts of a small percentage of probable Clinton voters to sway the election away from her. Inundating the social media accounts of targeted areas proved to be an effective means of achieving this goal and with a modicum of effort. And while Trump supporters continue to point to the Stanford/NYU study as irrefutable proof that fake news, no matter what it’s origin, had no effect on the national electorate, they miss the same exceptionally important demographic that the study’s authors did, regional data. And without that regional data, they cannot say, with great certainty, that fake news wasn’t responsible for swaying a small number of voters in some strategically important swing states. It was those states, their surprising election results, and their electoral votes Trump the election. He says so himself.