[For comment on our weak Yankee defense against information warfare, click here. For some popular recent posts, click on the links below:
In his classic novel 1984
, George Orwell imagined a dictator called “Big Brother” who controlled all means of mass communication. By means of “newspeak” (which today we call “spin”) and occasional outright lies, Big Brother was able to slant people’s views of reality and current events. Thus he manipulated their understanding of social reality, controlled their political views, and maintained a totalitarian state. He even could get people to like the result.
As terrifying as it was, and still is, Orwell’s dark vision had a logical flaw. It was based on the media of mass communication of Orwell’s time—the “one-to-many” technologies of radio and television.
But we are all different. No only do we have variant genetic makeup, which presupposes different levels of intelligence, perception, understanding and doubt. Even more important, we all lead different lives and have different backgrounds, parents, experiences and education. What you say or show on video to push another person’s emotional buttons may not push mine.
So a single message won’t do for all. There will always be doubters, renegades, rebels and misfits who think differently from the mass. Without being too explicit about it, Orwell apparently presupposed controlling these recalcitrant folk by pedestrian means— the military and the police—as in all oppressive societies throughout history.
But Orwell never imagined the Internet. As we are all slowly coming to understand, it differs from radio and television even more than those media differ from the telephone, a mostly one-to-one means of communication (except for rare conference calls). Not only does the Internet allow one-to-many communication, like radio and TV. It also permits many-to-many communication (as in Facebook), and many-to-one communication, as in e-mail.
These Internet capabilities are well understood and much remarked. But another, much more powerful aspect of the Internet is only now beginning to be recognized. Its consequences are barely discernible now. But as investigations soon may reveal, they could include the election of Donald Trump and both the Trump campaign’s and the Russians’ roles in it.
The as-yet unappreciated aspect of the Internet is the impact of artificial intelligence, or “AI.”
When combined with AI, the Internet permits one-to-many communication and many-to-many communication in which a message is precisely tailored to each individual recipient. In other words, AI now makes it possible to “weaponize” political propaganda and “fake news” in order to push individuals’ emotional buttons individually. It makes possible individually tailored messages and videos, designed precisely for each individual receiving or viewing them.
The speed and massive informational capacity of modern computers not only allows today’s Internet media to do this. It allows them to send such messages and videos to millions of people at about the same time. From an individual, human point of view, recipients appear to receive separate “blasts” of propaganda virtually simultaneously.
But things can get even worse. With pedestrian encryption and erasure techniques, dark media can send these individually tailored messages secretly. Then they can erase them after they are read or viewed, or after they have been “posted” for a specified period of time. In this manner they can cover their tracks and avoid discovery by political opponents or the authorities.
With all this capability, it’s easy for dark media to assuage recipients’ doubt and raise new doubts. They can, for example, bombard each individual recipient with multiple variants of the same basic message, ostensibly all from different sources and different sites, but all ultimately motivated and prepared by a single, small human group. They can also overwhelm opposing messages with both variety and volume.
They key thing that most pols and policymakers fail to understand is what AI now makes possible. Today’s propagandist no longer needs an army of paid trolls, like that which Russia apparently uses in its now-routine attacks on Western democracies.
With clever enough AI programming, it is now possible to automate the entire process of preparing and sending the propaganda. Dark media can automate the production and transmission of “fake news,” even fake videos. Then they can respond automatically to opposing fake news and propaganda counterattacks. The entire process is analogous to so-called “high-frequency trading” in the world of finance.
For short messages, such as Tweets, the process can be almost completely automated, without significant human intervention. For longer messages, some online human composition and editing may improve the propaganda’s content and impact. It’s even possible to automate the process of producing short videos, of the type commonly posted on YouTube, with minimal human intervention, using available video clips and computerized voice synthesis.
How does AI let propagandists figure out what emotional buttons to press for each recipient? By gathering data from each recipient’s use of the Internet.
Here, for example, is a recent report
on what information AI can
gather simply by scooping up a prospective recipient’s “likes” on Facebook:
[W]ith a mere ten ‘likes’ as input [a particular computer] model could appraise a person’s character better than an average coworker. With seventy, it could ‘know’ a subject better than a friend; with 150 likes, better than their parents. With 300 likes, [this] machine could predict a subject’s behavior better than their partner. With even more likes it could exceed what a person thinks they know about themselves.”
With the aid of such intimate knowledge of a recipient, it’s possible to tailor propaganda precisely to his or her “likes” and dislikes on many social issues. In this way, it’s possible to maximize psychologically the chance of a positive response.
The quotation above may seem hyperbole, and it may in fact be a reporter’s personal take. But its general thrust is well within the capability of current technology. Even without AI, it’s a simple feat today for Web “crawlers” to collect massive amounts of data from publicly open Websites, such as those of most for-profit corporations and those of many public figures, political candidates and political parties. These people and institutions have every incentive to keep their Websites and Facebook pages free from privacy restrictions and therefore open to unrestrained data mining.
Against this background, a recent bill passed by both the Senate
and the House
, which President Trump is likely to sign, assumes sinister importance. By this bill, Congress would overturn an Obama-era FCC rule prohibiting Internet service providers from collecting and selling customers’ Web-browsing histories to third-party marketers without the customers’ consent. In other words, when/if this bill becomes law, it will create open season on use of voters’ entire Web-browsing histories.
Of course browsing histories are not the same as Facebook “likes.” But they are part of the complete online experiences of American consumers and voters. Indeed, checking a “like” box might be deemed a part of Web browsing and therefore fair game for rampant data mining under this bill.
The important point is that extensive Web-browsing histories, as much as “likes,” can be used to create an automated, online dossier on every American consumer. Pols and policymakers simply don’t understand how easy, quick and cheap it is today—with computer reaction times measured in nanoseconds and two-terabyte disk drives
available to consumers at retail
for under $70—to “mine” and store a single voter’s entire Web browsing history for future use.
Of course those data are valuable to commercial firms, which can use them to target ads more precisely and efficiently at consumers’ needs and desires. But the data are even more
valuable to pols and propagandists, who can use voters’ Web-browsing histories to record what they read, what sources they return to over and over again, what issues matter to them, and to which causes and candidates they give real money. Then they can use these data to target “fake news,” “spin” and other propaganda precisely to each voter’s online personality. They can even automate the process of “weaponizing” their precisely targeted propaganda.
Thus does casual FCC-rule overruling, intended to facilitate commerce, threaten to undermine democracy and make all of pols’ stump speeches and in-person campaigning practically irrelevant.
Technology matters. It can shape a society, and it can undermine social order.
Twice on this blog (see 1
) I have pointed out how practically absurd is trying to interpret our Second Amendment as our Founders did. In 1791, when they drafted and adopted it, breech-loading a single shot into a musket or pistol took about a minute—the powder first, the shot second, and a cloth tampon to hold both in place third. That minute allowed ample time for a crowd or even a couple of people to subdue a rogue shooter. Today’s technology, with which small arms with big magazines can discharge twenty or so rounds in a single second, creates far greater danger to life, limb and social order.
So it is with our First Amendment
. The traditional theory for having virtually no effective control over speech is the “free marketplace of ideas.” If everyone can talk and write, the theory goes, readers and voters will have access to all competing ideas. They may stumble a bit in reaching the “right” conclusion, but the truth eventually will out.
That theory may have been viable at our Founding, when rival paper pamphlets were ubiquitous in our (then) relatively small cities, when each pamphlet typically ran just a few pages, and when opposing views were equally available. The theory was starting to crumble in the sixties, when the FCC adopted the “fairness doctrine” (which Ronald Reagan later abolished), requiring all broadcast attacks on political candidates to admit responses from those attacked on the very same broadcast channel. Today, the theory is an absurdly broken assumption about the Internet, in which there are millions of “channels,” each can contain book-length writings, video and audio, most users live in their own informational “bubble,” and clever media can propagandize them individually and even automate the process.
This is one of many reasons why it is so important to reject Neil Gorsuch as a candidate for the Supreme Court
. His “originalist” view of the Constitution would allow him to ignore—selectively or totally, at his whim—the technological and practical realities of our time. Human history is littered with the bodies of nations and societies that did that, including a key model for our own government, ancient Rome
I have worked in and around the computer industry for decades as a scientist, lawyer, law professor, and occasional, incidental computer programmer. As a lawyer, I drafted some of the first-ever commercial licensing agreements for AI programs. That was back in the 80s.
Although I cannot claim to have stayed abreast with AI over the years, I do know one thing. Even the best of our pols has absolutely no conception of how much and how quickly AI, applied to Internet-based propaganda individualized by data-mining, can change our society and the practice of politics and government. The level of understanding of the threat that I have seen in our pols’ public statements and hearings so far is pathetic.
Our pols are just starting to come to grips with the army of paid trolls and reams of fake news that Russia apparently mobilized to sway our recent election. What they don’t realize is that Russia’s attack on the information front, as effective as it may have been, was absolutely primitive compared to what AI now permits. Today, a small organization of less than 100 individuals can duplicate Russia’s army of trolls with automated equipment and software costing a few million dollars.
Moreover, the threat of weaponized AI propaganda can come as much from within as from without. According to two sources
, Breitbart.com—Steve Bannon’s erstwhile avocation—may have used AI data-mining software offered by a firm called Cambridge Analytica, which has reported ties to the wealthy Mercer family that backed Breitbart.com early on and continues to finance an array of Republican and right-wing causes.
To put it simply, the information war against our democracy and our Democrats may have been a two-front war, waged both from Russia and internally from Breitbart.com and later the Trump campaign itself. Indeed Breitbart.com’s understanding of and access to modern AI capability may explain much of the inexplicable: Trump’s late-campaign appointment of Bannon as his chief campaign advisor and (after Trump’s election) chief political strategy advisor.
At this point, most of our pols are much like Gabby Giffords. Back in the 1790s, she might have retained all her faculties after the armed assault on her. Unless her assailant had hit her with his first musket shot, her sympathetic crowd would undoubtedly have subdued him while he reloaded and so avoided further (or any) casualties.
Similarly, our pols today are trying to fight Internet-era automated AI propaganda with an image of radio and TV in their minds. They seem blissfully unaware that they are facing a technological threat as different from radio and TV as today’s rapid-fire, big-magazine assault rifles are from eighteenth-century breech-loading muskets. In just a few years, that threat can make their floor speeches, their stump-speeches, and even their websites pathetically obsolete. If they can’t get their minds around the threat and meet it for the sake of our democracy, maybe they can for the sake of their own careers and their jobs, which AI can take away as easily as modern small arms maimed Giffords.
The first step is to educate themselves on AI and on the threat of weaponized, individualized AI propaganda and fake news. Then they must ask the right questions of the producers and users of these technological threats to democracy, our intelligence agencies, those who dealt with the Russians before and during our last election, the Trump campaign, Bannon, and Breitbart.com and its backers.
Adapt or die. That is the universal rule of biology. It applies to societies, too. This three-front investigation—of the Russians, the Trump campaign, and the AI weaponized-propaganda threat to democracy—may take some time. But without it, the loss of our Republic to scoundrels with no scruples and AI capability is virtually assured.
As I did, the reader should take the principal source for this post
with a grain of salt. It comes from an obscure Website “scout.ai,” which describes itself as “accept[ing] submissions of original science and technology reporting, speculative and science fiction
on a rolling basis.” (emphasis added) This particular post did not state whether it fell into the italicized categories, but its organization, writing and use of quotations followed the format for factual reporting.
More important, the facts that I sought to check from other sources did in fact check out. Michal Kosinski, the reported author of the Facebook “like” data-mining program, is indeed a graduate of the University of Cambridge (UK), now an Assistant Professor of Organizational Behavior at Stanford Business School
. This article
from the New York Times
corroborates the Trump Campaign’s use of Cambridge Analytica and their involvement with Bannon and the Mercer family.
The quotation on the value of Facebook “likes” comes not from Kosinski himself, but from a Swiss magazine’s report on his path-breaking study of assessing personal traits from Facebook “likes.” In 2013 this study was published online, apparently after peer review, in the Proceedings of the National Academy of Sciences of the United States of America. An online version of the paper is available here
After doing as much research as I could do online in a reasonable amount of time, I concluded that the story in scout.ai, notwithstanding the website’s ambiguous self-description, both purported to be factual reporting and generally conformed with both the facts I could check and with my own understanding of AI based on a forty-year career working in and around the computer industry.
The threat is real. The only question today is whether our pols will rise to meet it or will remain sitting ducks for a disruptive and likely destructive technology.
The Death of Truth in America?
“Those whom the gods wish to destroy they first make mad.” — attributed to Euripides
“Divide and conquer.” — Julius Caesar
I worked hard to publish the foregoing essay on March 31, the day before April Fool’s Day. I didn’t want anyone to think it was or is a joke. But in fact the whole phenomenon of targeted misinformation has aspects of a gigantic April Fool’s prank. The difference is that this
prank will last far beyond April; it may never stop.
The joke is not on Hillary, on the Dems, or on the voters deluded by Internet lies to believe that Hillary was running a child-abuse ring
. It’s not even on the Russians, whose recent spate of youth street protests against Putin may have been partly motivated by an American cyber-counterattack
, or by secret, untraceable Yankee cyber-assistance to rebellious Russian youth groups.
The joke is on our entire species. It could, in the long run, kill us all, either by provoking a real
war that goes nuclear, or by distracting us from our species’ key current task: slowing the acceleration of global warming
. Or it could kill us by distracting us from the ever-present threat of a global pandemic, which President Trump has made more dangerous by defunding medical research.
In this blog, I have outlined the insanity of “total war
”. That notion was a German invention from the First
World War. It brought innocent civilians—for the first time in modern, supposedly “civilized” warfare—within the bull’s eyes of history’s most terrible and destructive weapons. In the Second
World War, it motivated random V-2 attacks on London, the fire-bombing of Dresden and Tokyo, and the nuclear incineration of Hiroshima and Nagasaki. In the Cuban Missile Crisis of October 1962, it brought a large fraction of humanity—perhaps our entire species—within hours, if not minutes, of nuclear and/or radioactive annihilation
Fortunately for our species’ uncertain future, the advent of more accurate weapons
has reduced the likelihood of species seppuku in warfare. Newer weapons like drones, ninjas, snipers, and (soon) accurately targetable “dial-a-yield” nukes make it possible to kill the bad guys without wiping out whole cities of innocent civilians.
But now a similar process of social evolution appears to be starting in the less immediately lethal field of deliberate disinformation and delusion. Un
fortunately, in that field the process of evolution is now going backwards
, from limited combat among professional warriors to attacks on entire populations.
In the old days, “disinformation” was a prank that military intelligence agencies played on each other. It was a limited kind of information warfare waged by one group of expert, well-trained, cynical and alert professionals against another.
Now what we have, at least in incipient form, is a type of information warfare designed to distract and delude entire populations. This new cyber-ability, made practical by artificial intelligence (“AI”), brings that warfare against innocent (and naive!) civilians to a new level. It’s a modern, information-only counterpart to the sui-genocidal notion of “total war.”
The most notable instance is probably Russia’s attempt, with its army of online trolls, to influence our Yankee 2016 presidential election. That attempt appears to have been successful, at least in part. Nobody doubts the intention
of the attempt, nor the outcome; only doubt as to the extent of the influence remains. Given the difficulty of precisely assessing the effect of any
information initiative in the blood sport that passes for politics in America, we will probably never know for sure.
But Trump’s victory is not the only possible result of this “total cyber warfare.” Brexit may be another. In the unlikely event that far-right-winger Marine Le Pen wins the upcoming French elections, that might be a third. And, if the truth be told, the toppling of pro-Russian tyrant Viktor Yanukovych in 2014, in street protests and elections that followed in Ukraine, may have been a third—this time, a rare victory in Yankee cyber-warfare over Russia. At least Putin and his spooks think the whole event was designed and engineered by our CIA.
For a moment, let’s suppose the truth of these more-than-plausible speculations. If we do, a sad conclusion rises unbidden. In this sort of “total” cyber-warfare, we Yanks (and the West generally) are far more vulnerable than authoritarian societies like Russia and China.
There are three reasons for this conclusion. First, Western populations, as distinguished from those of Russia and China, must fight a three-front war. Not only must they fend off disinformation from foreign powers like Russia and China. They must also ward off disinformation coming from within their own nations, specifically their own political parties and factions and their own self-interested, private corporations.
With the 2016 election so close behind us, we are all well aware of the massive disinformation campaigns mounted not only by political parties, but by factions within them and various PACs as well. There is plenty of material there for multiple doctoral theses on cyber-information war, including the fake news that brought an armed nutcase into a fast-food restaurant for the purpose of stopping Hillary from abusing children
. So I won’t dwell on this obvious point.
What’s less well known, and less well associated with cyber-information warfare, are the corporate instances of this phenomenon. The two most notable instances are already well-known to both pols and most informed citizens.
For over four decades, our Yankee tobacco industry concealed and distorted information and scientific research demonstrating the devastating effects of smoking tobacco on individual human health. At one point the CEOs of five major tobacco companies all stood up in testimony before Congress and swore that the nicotine in tobacco is not addictive. This corporate campaign of disinformation delayed effective control of the health scourge of tobacco for nearly half a century; today it still precludes direct regulation of tobacco as a dangerous substance.
A second corporate disinformation war is still going on today: global-warming denial. World-spanning corporations like Exxon Mobil spent millions and decades denying (1) that our planet is heating up and (2) our burning of fossil fuels is responsible. Yet these truths were and are accepted and indeed promoted by the overwhelming majority of qualified scientists and peer-reviewed scientific publications. The recent climate epiphany of the penultimate CEO of Exxon Mobil, Rex Tillerson, who is now our Secretary of State, only highlights how long, strong and politically effective corporate climate-change denial has been. And of course it’s still going on.
These corporate programs of disinformation kept Americans and our government from responding effectively to important scientific truths for decades. They pulled the wool over the eyes of the freest, most open great empire the world has ever seen. And they are still doing so, albeit perhaps in more limited ways.
In all this highly successful effort, our big corporations used only old fashioned, wholesale
disinformation techniques. They used one-to-many advertising on radio and TV, “public relations,” and false statements that managed to skirt the letter of the law of libel, defamation, perjury and false advertising. Imagine what they could do with today’s retail
technology of many-to-many lies and “fake news,” individually tailored to each recipient.
The second reason why we Yanks (and also Europe, Japan and South Korea) are more vulnerable to unrestricted information warfare than Russia and China is more fundamental. We are open societies; they are not. They have laws and large institutions designed specifically to control and censor both disinformation and real
information, especially that coming over the Internet. China, for example, reportedly has
[search for “50-cent”] 30,000 - 50,000 full-time Internet censors, as well as an estimated 250,000 - 300,000 party members paid to act as Internet trolls. And its censors are pretty effective: just try to find, online and inside China, a history of what happened at Tiananmen Square in June 1989.
We have no such protective laws, institutions or censors. Not only that, our First Amendment precludes our doing so.
Which brings me to the third and final point. Our First Amendment is our prime directive for speech, writing, video and news, including “fake news.” With few and very weak exceptions, its basic rule is that, in free expression, “anything goes.” So we have no real disincentives for disinformation, lies and fake news.
Sure, we have laws against defamation, false advertising, business libel, and child pornography. But those laws are narrow and weak. Defamation of a public figure (celebrity or pol, for example) must be done with “malice,” i.e., more than mere neglience, in order to justify a civil suit. An innocent or even stupid mistake (gross negligence) is not enough. That kind of malice is not easy to prove.
Laws regarding criminal
defamation do exist, but they are hardly ever used. Our society simply doesn’t tolerate censorship, no matter how inaccurate and damaging the “speech” involved.
This basic feature of our society leaves us unavoidably weak on defense against lies and disinformation, no matter how strong and sophisticated our offensive capabilities may be. It may be, for example, that our CIA and other spooks ousted Yanukovych in part with a sophisticated, retail-level cyber-warfare campaign. But if Russia retaliated by putting Trump in the White House despite his obvious lack of experience and qualifications, who won that exchange?
The hard fact is that “truth” is now entirely up for grabs in our open and relatively defenseless society. It’s up for grabs because our First Amendment gives us no effective defense against lies—even tailored lies foisted on our voters individually, one by one, by a foreign power using AI. It’s up for grabs because we have no effective laws against lies and disinformation generally and no institutions to enforce them even if we did. It’s up for grabs because our big corporations are getting into the disinformation game, in their own self-interest. Our political parties and PACs are already there.
Most of all, truth is up for grabs because we no longer have any national institutions of truth, whether formal or informal, like the Big Three television news networks of my youth: ABC, CBS and NBC. These networks still exist, but they are shadows of their former selves, with a fraction of their former audience, a sliver of their former prestige, and only a remnant of their reputations for quality and accuracy. The closest thing we Yanks have to a reliable national truth-teller, like the BBC in Britain, is PBS. But it has very narrow viewership and is under assault by the GOP and Team Trump.
So we Yanks are as defenseless against concerted cyber-information warfare from Russia and China (or from our own corporations, rich folk and PACs) as we were against the military tyrannies of Germany and Japan during our isolationist phase before World War II.
To put it simply, our civilian population is a sitting duck. We are trending toward a state in which the wisdom of both Euripides and Caesar at the head of this essay can destroy us. A deliberately engineered inability to know truth from lies can make us mad. And both foreign enemies and our own self-serving corporate and political empires will divide us, confuse us and ripen us for conquest. Truth in America may truly be on its deathbed.
So what can we do? In the long run, we must strengthen our laws against lies and disinformation. But that’s a long-term project. Near-absolute freedom of speech is so engrained in our laws and culture
that censorship of any kind is anathema. We Yanks will never have an “official” version of “truth,” let alone a single, dominant government news medium. And in any event, we are now so polarized politically that any attempt to cure our defenselessness by law or regulation will be viewed as an attack by one side on the other, and so will go nowhere.
We can, I suppose, begin to explore the possibility of some control over international
lies and disinformation by treaty and international agreement. That, too, is a long-term project, but worth pursuing. Just as treaties on nuclear disarmament have reduced the risk of species self-extinction by accident or miscalculation, so international agreements might set some basic, salubrious ground rules for information warfare. Doing so would be in all parties’ long-term interest: Russia’s leaders no more desire another spontaneous youth rebellion or bloodless coup in a vassal state (like that in Ukraine) than we want another Trump, let alone a smarter, more diabolical one.
But the most promising path out of madness and division lies in our private sector—our news media and our individual reporters. At the moment, they are the sole credible repositories of truth in our society. They must work hard to increase their credibility and respectability by avoiding sensationalism, double-checking their facts, and spreading their influence into retail information media, including Facebook and Twitter.
If the New York Times
and Washington Post
, for example, can leverage their independence from Murdoch and other sources of crass commercialism and obvious bias, they might have a chance of becoming analogues to ABC, CBS and NBC in the old days. They might become cultural icons, sources to which everyone turns to resolve doubts arising from the blizzard of retail-level propaganda and “fake news” roiling through the Internet. That is a consummation devoutly to be wished.
We Yanks are a creative and innovative people. Until recently, we have had a marvelously open society—open to new people, new ideas, new technologies, and new possibilities.
But as an essentially optimistic and forward-looking people, we have often been blind to serious threats. We were blind to the massive military buildups in Nazi Germany and Imperial Japan until history’s most terrible war became inevitable. We were blind to the threat of world-destroying stockpiles of nuclear weapons until we and the Soviets almost extinguished ourselves and the rest of our species in October 1962.
Today, we are blind to the grave threat of misuse of the Internet that we Yanks invented and gave to the world. It’s the most powerful system of communication ever devised. But its power can be used both for good and for evil, and we are blind to its dark side.
China and Russia have taken steps to protect themselves from its dark side and to use its power against us. So far, we have focused on offensive capability, but our defense is pathetic.
If we are to keep our open society from foundering in a sea of lies and delusion, that blindness must end soon. The place to begin is with a full, fair, bi-partisan investigation of Russian influence in our 2016 election, the collusion (if any) of Team Trump in it, and the ongoing vulnerability of our open society to lies and disinformation, spread over the Internet that we invented, and magnified a million-fold by “weaponized” individually-tailored delivery with the aid of AI.
This quotation has a long and tortured history, beautifully outlined with quotations from original Roman and Greek sources here
. This history suggests that the quotation is a gem of human wisdom refined through the ages before reaching its modern English form.
Before leaving office, President Obama announced that a cyber-counterattack against Russia would come at a time of our own choosing. He implied it might well be and stay secret. If such a counterattack had been in the works during the transition, the Trump Team would have been crazy to abort it or even to slow it down. For with all the evidence of possible collusion between Team Trump and the Russians, any evidence of such a move would be the kiss of death.
So the most likely causes of the brief Russian youth revolt are an American cyber-espionage counterattack or a spontaneous Russian youth movement so clever and advanced as to defeat the mammoth Russian cyber-warfare machine that may have put Trump in the White House. Either view is little more than informed speculation, but I find the former more probable.