20 January 2017

Friday fooled

Posted by Callan Bentley

I’ve been thinking about information flow lately, in light of the events that lead up to today’s inauguration of Donald Trump for the most powerful job in the solar system.

One of the key factors in Trump’s ascendancy is the increasing saturation of our news sources with falsehoods and propaganda. The phenomenon was dubbed Fake News in the aftermath of the election, though that term seems to have gone sour on the tongues of the media cognoscenti soon thereafter. I’ll use it in this blog post regardless, conscious that it imprecisely covers a range of phenomena, since the various flavors all have the general effect of misinforming people and muddying the waters of honest conversation. In a “post-truth” world, it’s as good a term as any to start our discussion.

On the most recent episode of his lively podcast We The People Live, Josh Zepps made an excellent point about the rhetorical maneuver that Trump used following the back to back publication (a) by CNN that a dossier on Russian kompromat of Trump exists, and both Obama and Trump had been briefed on it, and (b) by Buzzfeed of the alleged details of that dossier, an unsubstantiated and wildy salacious set of specific accusations. Guess which of those stories got more attention? The “golden shower” jokes ran wild on Twitter for hours afterward, but the incredible implication that Trump may be a sort of “Siberian Candidate” acting in clandestine collusion with a foreign government has largely set aside. The notion that it’s not even worth an official investigation has been voiced by some of the same national politicians that relentlessly investigated far less treasonous allegations of other public figures. Trump was discomfited enough to call a press conference in response, and there he infamously shut down CNN White House correspondent Jim Acosta’s question on the dossier, sneering “You are fake news.” This was an act that intentionally conflated the two separate stories and the two separate media outlets. Trump used the weaker story and less-reputable venue to silence the stronger, more disturbing story and the more trustworthy outlet. That’s a clever slight of hand; a process that Zepps reasonably dubs “Trumpfusion:”

It’s a bummer no quick-minded journalist on the scene recognized what was happening and called the President-Elect on it. Anyhow, listen to the first five minutes of the podcast for the relevant discussion. I think it’s worth it to wrap your brain around the way this trick works, because I bet it won’t be the last time it gets thrown at you over the next four years. The rest of the episode is also very much worth listening to: In it, Josh interviews Alexey Kovalev, who wrote a powerful piece about American media from the perspective of Russian media after a decade under Putin.

For a thoughtful, entertaining exploration of the Fake News issue that explores historical analogues, I recommend the “Extra! Extra!” episode of the podcast Flash Forward, by Rose Eveleth. In exploring the conundrum of Fake News, the realization dawns that the future where fake news is indistinguishable from actual factual news seems to be… now. Examples include Pizzagate and the Rwandan genocide, William Randolph Hearst and the heyday of Yellow Journalism, and you either getting this blog post in your Facebook feed, or the algorithm intentionally shutting it out.

An additional angle (not discussed in either of the podcasts linked to above) looms in my mind as problematic. That is the intentional manipulation of audio and video imagery to create misleading “footage” that never actually existed. A malicious person with access to this technology can create utterly convincing audio and video of anyone saying anything. This takes Fake News to a whole new level. Consider this: these faces in the image below weren’t really smiling when they were photographed:

Tom White made these images via a Twitter bot he created called smile vector.  But the technology isn’t limited to grinning GIFs. Consider this demonstration of Face2Face, a Stanford project of Matthias Nießner and colleagues which accomplishes real-time facial modification of video of another person:



The most compelling audio example I’m aware of is Adobe’s “Project VoCo,” nicknamed “Photoshop for audio,” in which you can type what you want a person to say, and the audio is instantly generated in their voice. Watch this demonstration, and consider the implications for our ability to distinguish reality from deceitful lies:



A good summary of the surprisingly advanced state of the technology can be found in this article on The Verge by James Vincent. Spend 3 minutes reading that piece and see if you don’t feel seriously unnerved.

Just as anecdotes are more powerful than empirical evidence or logical coherence to the non-scientist, and knowing more about a topic may not matter and in fact may even calcify an identity-based position, I am concerned about easy-to-generate fictional anecdotes that involve real people. How much more violent would the Pizzagate gunman have been if he had not merely read a fake news article about Hillary Clinton and a local DC pizzeria, but in fact had seen a video of the purported acts? It’s a chilling thought that (a) reputations could so easily be drenched in mud, and (b) people’s lives can be recklessly endangered when these tools are paired up with a credulous, armed audience.

So what’s the solution? It’s not clear.

We can’t trust figures in authority. We can’t trust the news in our social media feeds. Soon, we won’t be able to trust audio and video we hear and see. It’s an astonishing situation.

I’m an avid skeptic, and I value ‘thinking for yourself’ as highly as anything else in life, but the pollution of discourse and sensory evidence is unprecedented. The solution is likely to be unprecedented as well.

I think there’s a particularly good discussion of this in the Flash Forward podcast I link to above. The problem may be intractable in a society that values free speech. It wouldn’t be an issue if every citizen valued skepticism and independent verification of news stories, but there’s so damn much of it all that no one reasonably has the time to investigate it all for themselves to the extent it deserves. Independent fact-checking organizations like Snopes certainly have a role to play going forward, and perhaps they might even take on a role of awarding a provisional “verified” status to individuals or organizations that demonstrate they aren’t lying to their audiences. Another, immediate and more manageable step is for each of us to make a commitment to having our echo chamber dismantled. For any given issue we encounter, we need to put some work into thinking about it, rather that simply running it through the internal “Does This Agree With What I Already Think?” filter, and then hitting the digital button for “retweet” or “share.” A question I ask myself which you may find useful is “What would it take to convince me otherwise?” Running the thought experiment in our minds about information that would change our minds is a way of clarifying the useful role of evidence in sorting out what is real. Additionally, I think it would be beneficial if there were a social consequence for an individual who adds to the mess of the public conversation, and a reward to those who clarify things. Somehow, we need to make it socially repugnant to live in a media bubble, and socially laudable to speak honestly about facts.

Today, we enter a new era in the political history of the United States. The issue of distinguishing reality from destructive fiction is a skill we will need in the next four years.