Former National Intelligence Director James Clapper’s new book “Facts and Fears” confirms the intelligence community’s certainty that Russia interfered with our 2016 election. But he takes it further, asserting that Moscow actively favored Donald Trump and that their efforts worked. He says it’s common sense but offers little evidence for this last claim, which I share.

Political scientists will long labor to apportion blame for the election results among Hillary’s mistakes, Trump’s demagoguery and Russia’s “active measures.” I anticipate that research will fall into three broad methodologies with many variations: (1) timeline analyses, (2) in-depth interviews and (3) dissections of district voting in the three deciding swing states.

The two main types of Russian disinformation — “bots” (mass fake digital messages) and “trolls” (reworked hacks aimed narrowly at influentials) — resemble the old Soviet “agitprop”: simple agitation for the masses but brainier propaganda for intellectuals.

Most came from the Kremlin-related Internet Research Agency in St. Petersburg — the “troll farm” — which spent millions. Facebook and Cambridge Analytica provided data to devise bots, and WikiLeaks distributed Russian-produced trolls. Useful idiots.

Some items are irrelevant to the effectiveness question, demonstrating only that the Russians were trying, not that they succeeded. Moscow celebrations of Trump’s victory show they were relieved Hillary lost. Russian Foreign Minister Lavrov seemed to gloat upon visiting Trump in the White House in May 2017, beaming at a Russian triumph.

In spring 2016, a doctored troll suggested Clinton was breaking rules. The FBI debated its authenticity, but it led to Comey’s June announcement that they investigated her misuse of a private server. Point (1), the timeline, seeks dates when survey data show the Clinton-Trump gap narrowing. Major narrowing earlier than June clouds causality. Rapid narrowing after Comey’s announcement suggests causality but does not prove it.

Then, just 11 days before the election, Comey’s announcement that the FBI was looking into the laptop of Hillary’s assistant likely accelerated voter shifts. If survey data bear out the narrowing-gap hypothesis, Trump may owe Comey the election. Fine way to say thanks.

Point (2) will be more complex, requiring respondents to recall if they saw what later proved to be Russian bots, which bots, and were they influenced by them? Did bots prompt them to vote for Trump, mistrust other Americans or muddy rational judgments?



 




A first step should collect and archive all Russian bots. Respondents could be shown bots and asked if they remember any. Trouble here is there were hundreds of bots, too many to show respondents. But a likely dozen or two important ones could be shown.

The survey must include how much respondents use social media and distinguish between heavy, medium and light addicts. Were heavier users more likely to report remembering bots and being influenced? Or less likely? Did social media encourage them to shift away from Hillary to abstaining or voting for Trump or the Green or Libertarian candidate?

Point (3) would examine this question through district voting patterns in Wisconsin, Michigan and Pennsylvania, especially in regard to supporters of Bernie Sanders. Some swore “Bernie or bust” and refused to follow Senator Sanders’ endorsement of Hillary. Some bots cleverly deepened this conviction, which could have been more than enough to tilt the narrow results in the three swing states where Trump won by just 77,000 in total. Bernieites didn’t have to vote for Trump; they just had to not vote.

How can researchers tell? First, get electoral maps of the three states, at least down to congressional-district level or even precinct level. In each, count the Sanders vote in Democratic primaries — especially in hotspots such as the University of Wisconsin in Madison. Next compare the November vote to previous presidential votes. A falloff in turnout in 2016 or a shift to the Green or Libertarian candidates suggests disgruntled Bernie supporters.

None of these approaches is air-tight, and good studies will include all three and perhaps others. We must avoid the mistake that bots turned voters into zombies who obediently followed bot instructions. The Russian effort was far more subtle, gradually adding and altering bot messages to favor Trump without alerting viewers they were being manipulated.

We should begin work now against digital election manipulation, as it’s only going to get worse. If trollsters think it works, they’ll do more of it and more cleverly. Artificial intelligence makes algorithm construction faster and more accurate, able to feed hourly updates back to St. Petersburg. A different problem: Counting and tabulating votes seem secure, but malefactors doubtless seek to penetrate election computers.

Combating information warfare will require laws and the social-media firms to train large staffs of editors. Social media are not mere bulletin boards but have the same responsibility for accuracy as news media. One technique: two-person judgments with a third in case of a tie. When two agree, delete the item and block all the sender’s posts. Let the troll farmers work for a living.