ITTLE ROCK, Ark.— The Russians Are Coming! The Russians Are Coming! is a great Alan Arkin comedy from the sixties, but the panic on a New England island inspired by Theodore Bikel’s stranded Soviet submarine isn’t even half as hysterical as the current Potomac Panic over deepfakes coming out of Moscow.
Suddenly there’s a full-bore freakout going on among poor deluded souls, many of them with Ivy League degrees, who think that fake audio, fake video, fake websites, fake news articles, and fake fakers are going to ruin our election process.
By this time next year the world will be so full of apps that allow you to fake audio, fake video, and generate fake-news bots that the planet will be swamped with multitudes of lying liars attempting to manipulate our sacred ballot box.
These Bunched Panties People seem excessively focused on The Election. The presidential election. They apparently don’t care about any election before or after that one. Who cares if a few senators, governors, mayors, or (it goes without saying) mere state legislators stay up til 6 a.m. recording fake white nationalist screeds to pin on their opponents? They also don’t seem to care about deepfakes used in any kind of random garden-variety way, like producing tapes that make your ex-husband seem more evil in the eyes of the child-custody judge, catfishing widows and orphans on dating sites, or—my prediction for the most popular future use of deepfakes—impersonating a CEO saying something nasty so that the stock price will drop dramatically and you can short-sell it before the deepfake is discovered.
The implication here is that only two types of people are likely to use deepfakes to accomplish their diabolical goals—Russian spies and Trump campaign operatives.
Their evidence? “Look at what they did to Nancy Pelosi!”
And yes, it’s true, someone with a slightly off-kilter sense of humor doctored some Nancy Pelosi video to make it look like she was drunk, but that wasn’t even a deepfake. You or I could have done that. They just slowed down the frame-rate. It’s the equivalent of putting your thumb on the edge of a vinyl record as it rotates, thereby making the singer sound ten octaves lower and deranged, or, in the case of Ross Bagdasarian, speeding up the record to create 60 years of Alvin and the Chipmunks residuals.
Are we really this naïve at the highest levels of our government?
Let’s start with the obvious:
Yes, they’re gonna mess with our election. But here’s the part no one mentions:
We’re also gonna mess with their election.
We’ve messed with every Russian election since 1917. There’s probably a special desk in Langley, Virginia, with a sign over it that says “Messing With Russia Department.” It’s what intelligence services do. It won’t change. Some of these congressmen should admit that the only reason they’re incensed is that Russia did an especially good job of messing with us in 2016. (I was especially impressed by the two hot babes they sent to the United States to attend Trump rallies. We’ve done some pretty remarkable work in Moscow, but I don’t think we’ve ever had actual CIA operatives on campaign committees, handing out flyers and flirting with undecided voters.)
I’ll just use one example that I’ve used before.
In 1995 the Communist Party achieved a working majority in the Duma, or Russian parliament, and their candidate, Yennadi Zyuganov, was poised to win the June 1996 presidential election.
After four years of praising Russia for becoming a democratic state with free and fair elections, Bill Clinton ordered the CIA to interfere with their free and fair elections.
The Russian people were sick of the government run by Boris Yeltsin. They had just gone through four years of gangsterism, with their country’s infrastructure being carved up into fiefdoms by violent corrupt oligarchs. The Caucasus was in a full-scale revolt, and it was well known that Yeltsin could do very little about it, partly because he was an alcoholic in declining health.
Yeltsin’s approval rating in January of 1996 was 8 percent, and it was clear that a majority of the common people in Russia wanted to return to the security of Communism.
We decided that was not going to happen.
Our allies: the oligarchs.
By law, Russian presidential candidates were limited to $3 million in campaign spending. Yeltsin somehow came up with $2 billion.
Then, suddenly, the International Monetary Fund gave Russia a $10.2 billion loan, which Yeltsin used to pay everyone’s back wages and pensions.
Then, suddenly, a well-funded disinformation campaign got started. If Zyuganov wins, there will be cruel purges. People will lose their jobs and be sent to Siberian prison camps. There will probably be a civil war. This was pre-Internet, at least in Russia, so we did it the old-fashioned way. Pamphlets started showing up on the streets, outlining the Communist platform—only it wasn’t the Communist platform, it was a fake platform promising a return to Stalinism.
Everything worked beautifully. In the first round of voting, Yeltsin got 35 percent of the vote, Zyuganov 32 percent, and the rest was spread among nine candidates.
Inconveniently, Yeltsin had a heart attack before the runoff election and was still in the hospital on the day of voting. But media control is a wonderful thing, and nobody found out.
Final vote: 40 million for Yeltsin, 30 million for Zyuganov. The lopsided victory was helped greatly by vote counts in Tatarstan, Dagestan, and Bashkortostan that resembled the 1946 gubernatorial election in Georgia, the one where several hundred deceased residents rose from the Telfair County cemetery in alphabetical order and cast their write-in votes for Herman Talmadge.
If you compare Russian meddling in our 2016 election to American meddling in their 1996 election, I think we win.
So when we talk about deepfakes affecting the 2020 election, let’s start by admitting that life has always had deepfakes. Even the most casual moviegoer understands the use of blue screen, by which you can make a person appear to be somewhere he’s not, or CGI, which is used to do things like photograph 10 actors and turn them into an army of half a million. Carrie Fisher was de-aged 40 years in Rogue One and brought back to life in The Rise of Skywalker. (I’ve gotta think our Congress, with an average age of 93.7, would be in favor of this particular type of deepfake.) Somebody did a great job of inserting Harrison Ford into Solo to make it better, thereby proving that deepfakes can be used for good as well as evil.
But apparently people started getting nervous this year when media labs unveiled stuff like making Mona Lisa smile and talk, making Albert Einstein speak his own words from an old transcript, and, of course, the Jordan Peele video in which Barack Obama calls Trump “a total and complete dipshit.” Then there’s the thispersondoesnotexist.com website, which uses trial-and-error algorithms to categorize and connect billions of bits of visual data to create believable avatars that look like actual human beings, even though they’re composites. All you have to do is use a similar algorithm to synthesize a voice for the avatar and—voila!—you have an untraceable being that can be used to post millions of inflammatory Instagram photos, paste heads into porn scenes like the Redditor did to Gal Gadot a year ago, and, if necessary, commit suicide and write your own obituary.
Yvette Clark, a Congresswoman from Brooklyn, is so alarmed by these developments that she introduced a bill called the Defending Each and Every Person from False Appearances by Keeping Exploitation Subject to Accountability Act. (Get it? The acronym is DEEPFAKES with “accountability” added. It’s actually a pretty amazing acronym since it does mean basically what it represents.) But here’s what the bill calls for:
If you alter media, you have to state in writing exactly how you altered it and then watermark it.
Let me repeat:
What part of Brooklyn does Yvette come from? We need to check her street cred. Because this would be like saying, “Anyone planning to burglarize a residence is required to report to the nearest police precinct and register the address of the target house as well as your own address and contact information.”
Not to mention, you could simply describe your deepfake, watermark it, then email it to somebody in Ukraine who will immediately remove the watermark and circulate it. You are then protected from prosecution because of the DEEPFAKES Act. It wasn’t you, it was a third party.
Result of this law, if it passes: any fifth grader planning a parody of his teacher now has a lot of paperwork to fill out. Not to mention any short-film director on Saturday Night Live. Every supermodel will need at least three additional lawyers to chronicle all the changes made to every published or broadcast image. For that matter, almost every meme and GIF would have to be reported to the government. It’s like licensing entertainment and I doubt it’s constitutional.
And all of this hysteria arises from a single false assumption—that a recorded image is an accurate depiction of real life.
As any person who works in the film industry can tell you, it never is and it never has been. And whenever it’s used in a courtroom for the purpose of proving something, it can always be challenged. The jury saw the entire Rodney King video—and so did the rioters. The National Football League thinks that a video image can be more accurate than the human eye—not true—and so they’re constantly using instant replays to alter the judgments of officials on the field. Even then, they sometimes have to study multiple images for six or seven minutes to get to an educated guess about what actually happened in real life. If you haven’t yet seen the video of Rasputin singing Beyoncé, I would recommend it, just to indicate how sophisticated we can get. Show it to a person who doesn’t know who Rasputin or Beyoncé is and you’ve got a whole new audio/visual experience. DAMN that guy with the beard can lay down a lyric.
The illusion everyone is laboring under is the idea that audio and video of the past has been accurate but now it’s going to be manipulated in order to fool the Stupid People. It won’t fool you or me, of course, and even if it does fool us temporarily, it will be corrected in the New York Times the next morning and so we will be intelligent about it. But all these Stupid People out there will vote based on the illusion.
It’s an Us vs. Them argument. There are these yahoos who won’t be able to recognize a deepfake or won’t believe the debunking of the fake. Remember the DeepNude site? (It was shut down due to threats of lawsuits—which demonstrates the system in place already works.) DeepNude could take any picture and make it nekkid, but it was attacked by feminist groups and run out of business—although I don’t know why it was a feminist issue, since they were doing nekkid males as well as females. At any rate, the idea is that, if you can alter the image and voice of any person and distribute it widely on the web, you can capture enough yahoo votes to win a close election.
But this has always been the case. At the height of Mafia election corruption in New Jersey, for example, the going rate for yahoo manipulation was a dollar a vote. They would stand outside the voting booth and, in some cases, mark the ballot in advance so that nobody could cheat the mob out of that dollar. It wasn’t as sophisticated as a deepfake but it had the same basic motivation: “Let’s find people dumb enough to sell their vote for a buck.” (In the mob’s defense, I think the going rate for Jersey vote-buying had gone up to $5 by the turn of this century.)
Any radio deejay, if he’s honest, will tell you that his show is a deepfake.
Any television director would be astonished to find out you believe the show is anything but a deepfake.
So-called “reality tv,” which uses the very term “reality” to announce “we’re not doing any deepfakes,” has more deepfakes than scripted fictional shows.
The very first movie ever made, Sallie Gardner at a Gallop, was a deepfake. Supposedly it was a film of a galloping horse. In fact it was multiple camera images strung together to make you think it was a film. That was in 1878. A year earlier, Thomas Edison made the first audio recording by reciting “Mary Had a Little Lamb” on his Phonograph.
Why would he use prepared material instead of just saying “Hi there, I’m Tom Edison and we’re making history here today”?
Because a year before that Alexander Graham Bell had screwed up the first phone call when he said, “Mr. Watson, come here, I want you.”
When he wrote his memoirs, Bell decided that sentence was inelegant and abrupt and so he changed it to “Mr. Watson, come here, I want to see you.” So the first phone call, as recorded by history, is a deepfake.
In fact, the world loves deepfakes. When Neil Armstrong stepped onto the surface of the moon in 1969, he said, “That’s one small step for a man, one giant leap for mankind,” but apparently his Motorola communications system hiccupped in the middle of the sentence and omitted the word “a.” They could have cleaned it up and put the “a” back in, but then they thought, “Naw, it sounded good that way.” And so nobody ever heard “a man” at all. Armstrong turned himself into all men instead of a mere man, and we were okay with that. Sometimes the deepfake is better.