Silicon Valley Tech Pioneer Jaron Lanier Says the Internet Took a Wrong Turn, and You Can Help Save It

First, cats or dogs — which do you choose?

Jaron Lanier believes saving the internet has to start with you. Step one: delete all your social media accounts, or at least deactivate them for six months.

The computer scientist and old-guard Silicon Valley star entrepreneur uses cats and dogs as an analogy in Ten Arguments for Deleting Your Social Media Accounts Right Now, his new book, published late last month: get off social media and you can live more like a cat.

Cats live by their own ideas. They don't take orders.

Dogs can be trained to respond differently to different commands. That, he says, is what social media users are like — except the one doing the training isn't a specific person but the AI algorithms controlled by big companies.

These companies watch everyone's every move in the digital world in real time. Once they have the data, their algorithms show different people different information and different ads.

When an algorithm finds that fake news gets a stronger reaction out of you than real news, it recommends the fake news. When it guesses that a piece of content might make someone angry, it sends that too — it doesn't actually know what anger is, but it cares about the comments and the engagement anger produces.

What you see is shaped by them.

Algorithms decide who sees the photos and words you post, and how many people comment and like. And the dopamine those outside comments and likes release guides what you write, what photos you post, which filter you use... When was the last time you felt anxious because too few people liked your post?

What you do is shaped by them too.

If you've heard this argument before, it may be because Lanier has been making it for eight years. His doubts about social media turned into You Are Not a Gadget in 2010 (《你不是一件电子设备》, Chinese edition) — if social media is one big machine, then don't turn yourself into one of its parts, don't shrink your own capacities to fit the machine's rules, like writing only 140 characters at a time, or posting only the kind of photos that do well on social media.

Critics raved about that book, and Time named Lanier one of its 100 most influential people of 2010 as a result. But Silicon Valley was split on it, and it wasn't especially popular with general readers either — back then, attacking social media and Web 2.0 meant going against the current.

Lanier is one of Silicon Valley's technology pioneers. In 1984 he founded the virtual reality company VPL and coined the term "virtual reality." Eyematic, where he was chief scientist, was sold to Google in 2006 and became the starting point for Google's image-recognition technology. But by his own account, the book still cost him some old friends.

Eight years on, Lanier's views no longer sound so strange. Facebook's founding president Sean Parker said late last year that social media was a business about addiction from the very start:

"We need to give you a little dopamine hit every once in a while — someone likes or comments on your photo or your update... It's a social-validation feedback loop... The inventors, the creators — me, Mark [Zuckerberg], Kevin Systrom at Instagram — we all understood this, and we did it anyway... This mechanism is changing your relationship with society, changing your relationship with other people."

Chamath Palihapitiya, who ran user growth at Facebook until 2011, went on to say that this dopamine-based feedback loop is a global problem destroying how society works, and urged the Stanford MBA graduates in front of him to stay away from social media as he does. Facebook replied that he had left years ago: "It was a very different company, and we have come to realize that our responsibilities grow with our size."

Twitter co-founder Evan Williams put it this way: "I thought once everybody could speak freely and exchange information and ideas, the world is automatically going to be a better place. I was wrong."

That social media uses information and product design to shape behavior is no secret. Facebook ran a large-scale human experiment back in 2014 — pushing particular content to 690,000 users who had no idea, to see how their moods would change. Chinese researchers have likewise found that Weibo encourages the spread of negative emotion.

It isn't only social media executives doing the soul-searching. So are the people who helped build computers and the internet.

Tim Berners-Lee, who invented the World Wide Web, has warned once, twice, and again. Google product designer Tristan Harris quit the company and started the Time Well Spent movement, calling the addictive product design of today's internet companies unethical. Richard Stallman, once Lanier's roommate and the spiritual leader of the free-software movement, was blunter: "Fuck it — if the price [of a free service] is that these companies get to know everything about us, they shouldn't exist."

After a string of scandals, the way social media operates has had its legitimacy questioned too. Zuckerberg faced questioning from the U.S. Congress and the European Parliament, and spent at least US$30 million on apology ads.

The movement to delete Facebook heated up for a while, then faded.

Lanier doesn't think we can stop now. At this April's TED conference he argued that these companies are no longer social media platforms and advertisers; they are Behavior-Modification Empires, shaping the free will of hundreds of millions of people.

But Facebook, Twitter, WeChat Moments, Weibo — these platforms have already worked their way into the lives of half of humanity, along with similarly designed products like Toutiao and YouTube. How do you get away from them?

"Well, this is a very, very, very hard technical problem," Lanier told a QDaily reporter after his TED talk, straight-faced. "I can share some tools I've found. It took quite a bit of effort. It's genuinely a bit technical, so it may be hard to explain."

"One is email," Lanier said, then paused and could not help laughing out loud for several seconds. He went on: he also calls friends on the phone, has people over, gives talks — and partway through, burst out laughing again. The high-pitched laugh made him sound at least thirty years younger.

The 58-year-old hacker still passes up no chance to make a joke, and he is always the first to laugh at it.

Lanier believes leaving social media won't make it impossible to live, and that he himself is proof you can stay off it and still have a public life — though "social media" here does not include point-to-point messaging tools, which means WeChat is fine and Moments is not.

He has never had a Twitter or Facebook account of his own, only a few fake ones for trying out various services.

By his account, of course, those accounts belong to a cat. He now worries about that cat's mental state; he suspects she has become a Trump supporter, since she is always hostile toward visiting cats.

The future he fears most is turning into reality, and yet Lanier manages a kind of optimism.

He believes every technical development in the history of computing was the result of human choice, and that nothing had to happen. So everything can still be saved. And so everything can still get worse.

Having felt the cruelty of an older era, Lanier once saw virtual reality as a chance to restart the media

Unlike entrepreneurs raised in the home of a prominent lawyer or a dentist, Lanier learned very early what an older era was capable of, and what people are capable of. His mother had been in a Nazi concentration camp; most of his father's relatives died in anti-Jewish pogroms in Ukraine.

Being born in America was no guarantee of safety. When Lanier was seven, a distant relative was killed on the street by an antisemite, and his parents took him from New York to a small Texas town near the Mexican border. Even the surname Lanier was a safety measure — the earlier name, Zepel, sounded Jewish the moment you heard it.

His mother died in a car crash when he was ten. America then had none of the political correctness so many people complain about today: a teacher "comforted" him by saying the crash was no tragic accident, only payback for the Jews murdering Jesus.

That isn't to say his childhood was nothing but misfortune. His parents, who moved in artistic circles, encouraged him early on to take up instruments, painting, and building things with his hands. At 11 he designed a hemispherical dome, built it with his father, and moved in.

At 14 Lanier began studying at New Mexico State University, paying tuition by keeping camels at home and selling their milk to hippie restaurants. That was his first encounter with computers.

Even as he marveled at what computers could do, Lanier also came across, for the first time, the warnings of the computer scientist Norbert Wiener.

In the early 20th century a run of behavioral-psychology experiments became a hot topic. Pavlov's dog, the Skinner box (rats), Little Albert (an infant)... experiments run in the name of science made their way to human subjects.

Wiener argued in a book in the 1950s that if there were a device people carried with them that could collect everyone's data in real time and feed it back to them, these behavior-modification experiments could be extended to all of humanity. He thought the result would be catastrophic, but too far off to worry about — it was hard, then, to picture everyone holding a smartphone 60 years later.

That came later. Lanier first caught Silicon Valley at its most alive.

In 1979, the 19-year-old Lanier had already dropped out of two universities and found his way to Silicon Valley to make video games.

There he got to know some brilliant minds: Richard Feynman, still teaching at Caltech then; Kevin Kelly, editor of the Whole Earth Catalog; John Perry Barlow, the music-loving co-founder of the digital-rights group EFF; and Timothy Leary, the godfather of LSD (in his last years Leary became a preacher for all sorts of computing technology, virtual reality included).

In 1984, Lanier and Tom Zimmerman, a colleague from Atari, founded the virtual reality company VPL with others.

What VPL made worked on roughly the same principle as Oculus, HTC Vive, and PSVR, which had their moment a couple of years ago: put on a head-mounted display and you are fully immersed in a virtual world. The virtual reality gear sold today still looks a lot like what VPL invented.

Lanier demonstrating the EyePhone and DataGlove in his early years

Unlike today's virtual reality companies, which concentrate on games, what Lanier wanted was social connection.

The product's name says as much: VPL's headset was called the EyePhone — let your eyes connect to the world.

He saw virtual reality as a way for people to talk to one another. In a virtual space, people could drop their real appearance, their race, their gender, even the limits of species, and meet each other as whatever they wanted to be.

In an ABC television program about virtual reality from the early 1990s, a much younger, much more optimistic Lanier said to the camera:

I think we have already brought about more technology than people actually need. The main problems left in modern society are disease and natural disasters. Every other material problem, technology has solved or is about to solve.

The main problems left are the ones human behavior creates. When I think about what technology the future needs, what comes to mind is new forms of people meeting each other, and virtual reality is the form best suited to that.

...The best thing about virtual reality is that it is new. It gives us a chance to start over, to take responsibility for a new technology and for the next generation of media (and so to fix the fact that television is too shallow a medium).

But VPL was too early. Computers then had no 3D graphics processors, and the World Wide Web would not leave CERN until 1991. VPL was bound to be a frontier project with no chance of commercial success.

With no consumer market available, VPL sold gear to companies and to NASA (for controlling robotic arms), built the first virtual reality surgical simulator, and held on until it went bankrupt in 1992.

What actually changed how humans socialize was not VPL or its competitors. It was thefacebook.com, the website Zuckerberg wrote in his dorm room in 2004.

Declaring war on digital collectivism, Lanier became the least polite critic of Silicon Valley inside Silicon Valley

In the dozen-plus years after he left VPL, Lanier took part in dozens of research projects and consulting jobs, large and small, and composed music for films and radio ads.

The more important ones included the image-recognition company Eyematic and Internet2, an early consortium working on the internet's underlying technology. He was also one of the technical consultants on Minority Report (both Lanier and Spielberg were regulars at the early TED conferences).

Some of that consulting work put him back on alert about technology. Lanier recalls an insurance executive in the 1990s who, on learning about machine learning, excitedly sketched a business opportunity: computers could use data to predict who was more likely to get cancer, so the insurer could steer clear of them in advance.

In the early 2000s, personal computers spread, Google connected every website in the world, Wikipedia proved that mobilizing user-generated content (UGC) at scale worked, and social media moved out of the college dorm.

Not even the dot-com crash could stop the boom that followed.

Silicon Valley still carried its early utopianism, convinced everything on the internet should be free. People built one free website after another, contributing information for free and taking information for free.

Free did not mean anti-capitalist; Silicon Valley never rejected capitalism. By the 2000s, entrepreneurs like Steve Jobs had replaced the technology pioneers as Silicon Valley's biggest idols.

Lanier believes every problem that followed comes out of that conflict, and out of the path Larry Page, Sergey Brin, and Mark Zuckerberg then found.

Google's two founders found a way to have both free and capitalism: advertising. Google builds free web services for you to use, takes your data and sells targeted ads, builds more and better free services, takes more data and sells more targeted ads. Nearly every free internet service since has copied the model.

Silicon Valley decided Google had found the way out for a free internet. That is when Lanier wrote his own "minority report."

In 2006, Lanier published an essay in the elite online magazine Edge, Digital Maoism, charging that an internet that had once encouraged independent websites and individual thought was being swept up by collectivism.

He used Wikipedia as his example: "Reading a Wikipedia entry is like studying the Bible closely; the authors and editors leave only the faintest trace."

Aggregation sites like Wikipedia, Facebook (social media), and Digg (news aggregation) began multiplying in 2006 — so-called Web 2.0. Information certainly started moving faster.

But on those sites the author was deliberately played down. What each person wrote, the photos and links they posted, became so-called user-generated content (UGC), there for visitors to consume. Once authors stopped mattering, one more or one fewer made little difference, and the aggregation sites naturally gave them none of the return they were owed.

This was truer still after Facebook invented the News Feed in 2006. On social media, people increasingly scrolled the feed the site handed them, looking at the photos and words posted by all sorts of people, and no longer needed to open each person's page to see their updates. The News Feed has been standard on every social media and news-aggregation site since.

Search engines put these sites ahead of independent small and mid-sized ones and sent them more users. The internet went from a huge decentralized community made of countless sites, big and small, to a handful of large social media platforms pulling away most of the users.

On those platforms, users supply information for free and add to the platform's traffic, but get little back. You may have thousands of followers, but what all of those followers are really following is the feed the platform itself supplies.

• In the beginning. Information was scattered across different sites and linked by hyperlinks; users went looking for information through search engines; authors were prominent.
• Web 2.0. User-generated content; a few big sites aggregated the content, but information was still ordered by hand or by time; users began receiving information passively; authors mattered less.
• Now. Most content sits on a small number of sites; the biggest social media platforms have started blocking search engines from indexing their information; social media and news-aggregation sites use algorithms to recommend different information and different ads to different people; users passively receive information tailored to them; authors are replaceable.

"The most beautiful thing about the internet is that it connects everyone. What has value in this is other people. If we start believing the internet itself is an entity with a voice of its own, then we are devaluing other people, and we are idiots," Lanier wrote.

A few months after Digital Maoism appeared, Eyematic, the image-recognition company Lanier was involved with, was sold to Google and became Google's image-recognition division.

As Lanier tells it, Google co-founder Sergey Brin told him company employees could not write pieces that controversial. So he cashed out and left.

Lanier then joined Microsoft, where his day job got simpler: first the development of the Xbox Kinect gesture-control device, and since then experimental virtual and augmented reality projects at Microsoft Research, which he is still doing.

More and more of his energy has gone into preaching — on the harm done by the internet, social networks, and artificial intelligence. From 2010 to now he has published four books and given hundreds of talks and conversations.

Lanier believes the free-internet-plus-advertising model is the root of the problem

Meanwhile the internet and the world changed enormously. The revolution the iPhone set off kept everyone connected to the internet 24 hours a day; Google became a company worth more than Microsoft; and Facebook and Tencent, keeping up with the mobile era, cornered social media worldwide.

Smartphones and AI algorithms magnified social media's problems and made the Behavior-Modification Empires Lanier feared genuinely possible.

He sums it up as the result of the free economy.

All social media is free and earns its money from advertising. The number these companies chase is getting people to see more ads, which means getting users onto their services more often and for longer.

And what is easier to stir in people is negative emotion — whether that is agreement with a piece of hate speech or a counterattack on speech you disagree with. Once a platform is big enough and holds enough user data, its algorithm can always find negative emotion an audience.

This is natural soil for fake news. Fake news is built to stir emotion, which suits it to spreading on social media. Media outlets used to be the gatekeepers — imperfect ones, but gatekeepers — while on social media there were, for a time, no limits at all.

By the end of 2016, Zuckerberg posted on Facebook to apologize for the spread of false information. More than one person saw fake news to the right of the apology.

Lanier believes the way information travels on social media has directly caused the backlash that various civil-rights movements have met in recent years.

Take the Arab Spring: young people living under authoritarian rule used social media to gather and topple the dictatorships in Egypt, Syria, and elsewhere. A few years on, those places had not improved — extremism spread faster with social media's help.

Likewise, the movement for women's equality in gaming met a counterattack that became #GamerGate, a mass harassment campaign full of rape and murder threats. After the Black Lives Matter movement against police violence toward Black people gathered force in the U.S., far-right groups like the KKK and neo-Nazis expanded fast enough to hold mass rallies and set off violent clashes. In Sri Lanka and other places already prone to conflict, the hatred fake news carries has directly touched off violence.

Historically, a backlash this fast and this fierce rarely followed a civil-rights movement's victory. Lanier believes these backlashes keep coming because group emotion is being shaped by algorithms. The algorithm can always find an audience that hates these movements, even though that was never the engineers' intent.

The world is answering the problems the internet has caused, but that guarantees no good outcome

Lanier is still not part of Silicon Valley's mainstream, but he is no longer alone.

In April 2018, TED invited Lanier on stage to lay out his warning about Behavior-Modification Empires — his first talk in nearly 20 years at a conference built on preaching innovation and optimism.

"We used to believe that once the internet connected everyone, the world would get better on its own. It didn't turn out that way," TED curator Chris Anderson said as he brought Lanier on stage. The conference then released Lanier's talk as its first video of the year.

While Lanier gave his talk, Facebook CEO Mark Zuckerberg was being questioned by the Senate in Washington.

The world is answering the problems the internet has caused.

Google and Apple both put addiction limits into their new operating systems this year, and began telling users which services they used each day and for how long. You can choose to cap your own time.

Apple, which has no advertising revenue to protect, went further, announcing in June an upgrade to its browser's anti-tracking features, cutting the user information internet services can track.

Social media is answering too. Zuckerberg said Facebook's human review team will grow to 20,000 people this year, with artificial intelligence helping to filter out hate speech.

Lanier does not believe any of this ends well. He sees it as one more case of the tech world's religious worship of AI.

"We still don't have a scientific definition of how the brain works. For example, we don't know how information is encoded in the brain. We don't know exactly how conversation is produced," Lanier told QDaily. "That doesn't mean we will never know. But right now we have to be honest and say we really don't know."

"That's not to say AI algorithms are fake, only that we don't understand how they work," he added. Feed in data and a machine-learning algorithm gives you a result. Behind that is still a black box; even the engineers don't know exactly how the result was reached, and there is no way to guarantee it will keep getting better.

Because machine-learning algorithms are black boxes, tuning them means reasoning backward from the results engineers can see — so an algorithm's ability to hold off extreme content cannot improve reliably.

ISIS's new-media team got around social networks' defensive algorithms more than once. They built a Twitter app that, once a user downloaded and registered, automatically posted the group's propaganda to that user's account. When ISIS announced its march on Baghdad in 2014, two of the first four photos that came up in a Twitter search for "Baghdad" were ISIS propaganda.

Organizations that spread hatred are using manpower and human wits to attack social media's filtering algorithms.

Facebook admitted last month that its algorithms filter out violent and pornographic images effectively, but that human reviewers plus AI algorithms catch only 38% of hate speech. The hate speech that slips through is pushed to social media users as usual.

Social media companies policing information themselves brings an obvious new problem.

Among the ideas Zuckerberg has raised is grading media outlets, with Facebook deciding which outlets are more credible. Put into practice, that would be censorship of speech, plainly.

Zuckerberg has said outright that running this company is more like running a government. Unlike a government, though, today's internet companies use their share structures to hold voting power in the hands of a few managers (most of Facebook's voting power belongs to Zuckerberg alone) — which makes them more centralized than most authoritarian states.

Going from a growth-chasing algorithm deciding what 2 billion people see, to a small group of people deciding what 2 billion people see, does not sound better.

In some other countries that shift is already under way — China, Russia, Egypt and others. Governments are already taking over the internet. Every company's algorithm serves a single voice that cannot be argued with.

Lanier believes the final solution still lies with each person, not with sitting and waiting for big companies to search their souls or for governments to step in.

Every user who cuts the connection to social media puts a little more pressure on the tech companies. That would be a start.

It would force social media to change how it works, from free to some form of paid — subscription, or some automated micropayment system that pays back creators at scale. The point is that time and user attention cannot be the only number tech companies care about.

Lanier has no concrete answer for how to design a paid system. He believes no alternative has appeared because the whole internet, back then, followed Google's and Facebook's business model and never explored other paths far enough.

Bloomberg reported last month that Facebook is studying internally whether an ad-free paid version is feasible.

"Nothing in the development of computing technology is fated." Lanier still believes that.

Let's hope he's right.

Lead image: a still from The Matrix

Banner image: Lanier's award speech at the Frankfurt Book Fair, which opened with him playing the khaen, a Laotian folk instrument

Lanier's four books:

2010 You Are Not a Gadget | 《你不是个玩意儿》 (a Chinese title hard to make sense of)

2013 Who Owns the Future | 《互联网冲击：互联网思维与我们的未来》 (another Chinese title hard to make sense of)

2017 Dawn of the New Everything | 《虚拟现实：万象的新开端》 (memoir)

2018 Ten Arguments for Deleting Your Social Media Accounts Right Now | 《立刻删掉所有社交媒体账号的十个理由》 (no Chinese edition yet)