Tag Archives: bots

Computational Propaganda

[Sorry to clutter your inboxes; I published this in error. Consider it an “extra.”]

I am now officially befuddled. Out of my depth. And very worried.

Politico has published the results of an investigation that the magazine conducted into the popularity (in social-media jargon, the “viral-ness”) of the hashtag “release the memo.” It found that the committee vote

marked the culmination of a targeted, 11-day information operation that was amplified by computational propaganda techniques and aimed to change both public perceptions and the behavior of American lawmakers….Computational propaganda—defined as “the use of information and communication technologies to manipulate perceptions, affect cognition, and influence behavior”—has been used, successfully, to manipulate the perceptions of the American public and the actions of elected officials.

I’ve been struggling just to understand what “bots” are. The New York Times recent lengthy look at these artificial “followers”–you can evidently buy followers to pump up your perceived popularity–helped to an extent, but left me thinking that these “fake” followers were mostly a form of dishonest puffery by celebrities and would-be celebrities.

Politico disabused me.

The publication’s analysis showed how the #releasethememo campaign had been fueled by  computational propaganda. As the introduction says, ” It is critical that we understand how this was done and what it means for the future of American democracy.”

I really encourage readers to click through and read the article in its entirety. If you are like me, the technical aspects require slow and careful reading. Here, however, are a few of the findings that particularly worry me–and should worry us all.

Whether it is Republican or Russian or “Macedonian teenagers”—it doesn’t really matter. It is computational propaganda—meaning artificially amplified and targeted for a specific purpose—and it dominated political discussions in the United States for days. The #releasethememo campaign came out of nowhere. Its movement from social media to fringe/far-right media to mainstream media so swift that both the speed and the story itself became impossible to ignore. The frenzy of activity spurred lawmakers and the White House to release the Nunes memo, which critics say is a purposeful misrepresentation of classified intelligence meant to discredit the Russia probe and protect the president.

And this, ultimately, is what everyone has been missing in the past 14 months about the use of social media to spread disinformation. Information and psychological operations being conducted on social media—often mischaracterized by the dismissive label “fake news”—are not just about information, but about changing behavior. And they can be surprisingly effective.

An original tweet from a right-wing conspiracy buff with few followers was amplified by an account named KARYN.

The KARYN account is an interesting example of how bots lay a groundwork of information architecture within social media. It was registered in 2012, tweeting only a handful of times between July 2012 and November 2013 (mostly against President Barack Obama and in favor of the GOP). Then the account goes dormant until June 2016—the period that was identified by former FBI Director James Comey as the beginning of the most intense phase of Russian operations to interfere in the U.S. elections. The frequency of tweets builds from a few a week to a few a day. By October 11, there are dozens of posts a day, including YouTube videos, tweets to political officials and influencers and media personalities, and lots of replies to posts by the Trump team and related journalists. The content is almost entirely political, occasionally mentioning Florida, another battleground state, and sometimes posting what appear to be personal photos (which, if checked, come from many different phones and sources and appear “borrowed”). In October 2016, KARYN is tweeting a lot about Muslims/radical Islam attacking democracy and America; how Bill Clinton had lots of affairs; alleged financial wrongdoing on Clinton’s part; and, of course, WikiLeaks.

There’s much more evidence that KARYN is a bot—a bot that follows a random Republican guy in Michigan with 70-some followers. Why?

It would be fair to say that if you were setting up accounts to track views representative of a Trump-supporter, @underthemoraine would be a pulse to keep a finger on—the virtual Michigan “man in the diner” or “taxi driver” that journalists are forever citing as proof of conversations with real, nonpolitical humans in swing states. KARYN follows hundreds of such accounts, plus conservative media, and a lot of other bots.

KARYN triggers other bots and political operatives, and they combine to create a “tweet storm” or viral message. Many of these accounts are “organizers and amplifiers”—accounts with “human conductors” that are partly automated and linked to networks that automatically amplify content.

The article is very long, and very detailed–and I hope many of you will read it in its entirety. For now, I will leave you with the concluding paragraphs:

So what are the lessons of #releasethememo? Regardless of how much of the campaign was American and how much was Russian, it’s clear there was a massive effort to game social media and put the Nunes memo squarely on the national agenda—and it worked to an astonishing degree. The bottom line is that the goals of the two overlapped, so the origin—human, machine or otherwise—doesn’t actually matter. What matters is that someone is trying to manipulate us, tech companies are proving hopelessly unable or unwilling to police the bad actors manipulating their platforms, and politicians are either clueless about what to do about computational propaganda or—in the case of #releasethememo—are using it to achieve their goals. Americans are on their own.

And, yes, that also reinforces the narrative the Russians have been pushing since 2015: You’re on your own; be angry, and burn things down. Would that a leader would step into this breech, and challenge the advancing victory of the bots and the cynical people behind them.

 

Weaponizing Speech

A couple of weeks ago, I came across a provocative article by Tim Wu, a media historian who teaches at Columbia University, titled “Did Twitter Kill the First Amendment?” He began with the question:

You need not be a media historian to notice that we live in a golden age of press harassment, domestic propaganda and coercive efforts to control political debate. The Trump White House repeatedly seeks to discredit the press, threatens to strip broadcasters of their licenses and calls for the firing of journalists and football players for speaking their minds. A foreign government tries to hack our elections, and journalists and public speakers are regularly attacked by vicious, online troll armies whose aim is to silence opponents.

In this age of “new” censorship and blunt manipulation of political speech, where is the First Amendment?

Where, indeed? As Wu notes, the First Amendment was written for a different set of problems in a very different world, and much of the jurisprudence it has spawned deals with issues far removed from the ones that bedevil us today.

As my students are all too often surprised to learn, the Bill of Rights protects us against government misbehavior–in the case of our right to free speech, the First Amendment prohibits government censorship. For the most part, in this age of Facebook and Twitter and other social media, the censors come from the private sector–or in some cases, from governments other than our own, through various internet platforms.

The Russian government was among the first to recognize that speech itself could be used as a tool of suppression and control. The agents of its “web brigade,” often called the “troll army,” disseminate pro-government news, generate false stories and coordinate swarm attacks on critics of the government. The Chinese government has perfected “reverse censorship,” whereby disfavored speech is drowned out by “floods” of distraction or pro-government sentiment. As the journalist Peter Pomerantsev writes, these techniques employ information “in weaponized terms, as a tool to confuse, blackmail, demoralize, subvert and paralyze.”

It’s really difficult for most Americans to get our heads around this new form of warfare. We understand many of the negative effects of our fragmented and polarized media environment, the ability to live in an information bubble, to “choose our news”–and we recognize the role social media plays in constructing and reinforcing that bubble. It’s harder to visualize how Russia’s infiltration of Facebook and Twitter might have influenced our election.

Wu wants law enforcement to do more to protect journalists from cyber-bullying and threats of violence. And he wants Congress to step in to regulate social media (lots of luck with that in this anti-regulatory age.) For example, he says much too little is being done to protect American politics from foreign attack.

The Russian efforts to use Facebook, YouTube and other social media to influence American politics should compel Congress to act. Social media has as much impact as broadcasting on elections, yet unlike broadcasting it is unregulated and has proved easy to manipulate. At a minimum, new rules should bar social media companies from accepting money for political advertising by foreign governments or their agents. And more aggressive anti-bot laws are needed to fight impersonation of humans for propaganda purposes.

When Trump’s White House uses Twitter to encourage people to punish Trump’s critics — Wu cites the President’s demand that the N.F.L., on pain of tax penalties, censor players — “it is wielding state power to punish disfavored speech. There is precedent for such abuses to be challenged in court.”

It is hard to argue with Wu’s conclusion that

no defensible free-speech tradition accepts harassment and threats as speech, treats foreign propaganda campaigns as legitimate debate or thinks that social-media bots ought to enjoy constitutional protection. A robust and unfiltered debate is one thing; corruption of debate itself is another.

The challenge will be to craft legislation that addresses these unprecedented issues effectively–without inadvertently limiting the protections of the First Amendment.

We have some time to think about this, because the current occupants of both the White House and the Congress are highly unlikely to act. In the meantime, Twitter is the weapon and tweets are the “incoming.”