by Adam Alter
Thirty-seven years after Zeiler published his results, a team of Facebook web developers prepared to unleash a similar feedback experiment on hundreds of millions of humans. Facebook has the power to run human experiments on an unprecedented scale. The site already had two hundred million users at the time—a number that would triple over the next three years. The experiment took the form of a deceptively simple new feature called a “like” button. Anyone who has used Facebook knows how the button works: instead of wondering what other people think of your photos and status updates, you get real-time feedback as they click (or don’t click) a little blue-and-white thumbs-up button beneath whatever you post. (Facebook has since introduced other feedback buttons, so you’re able to communicate more complex emotions than simple liking.)
It’s hard to exaggerate how much the “like” button changed the psychology of Facebook use. What had begun as a passive way to track your friends’ lives was now deeply interactive, and with exactly the sort of unpredictable feedback that motivated Zeiler’s pigeons. Users were gambling every time they shared a photo, web link, or status update. A post with zero likes wasn’t just privately painful, but also a kind of public condemnation: either you didn’t have enough online friends, or, worse still, your online friends weren’t impressed. Like pigeons, we’re more driven to seek feedback when it isn’t guaranteed. Facebook was the first major social networking force to introduce the like button, but others now have similar functions. You can like and repost tweets on Twitter, pictures on Instagram, posts on Google+, columns on LinkedIn, and videos on YouTube.
The act of liking subsequently became the subject of etiquette debates. What did it mean to refrain from liking a friend’s post? If you liked every third post, was that an implicit condemnation of the other posts? Liking became a form of basic social support—the online equivalent of laughing at a friend’s joke in public. Likes became so valuable that they spawned a start-up called Lovematically. The app’s founder, Rameet Chawla, posted this introduction on its homepage:
It’s our generation’s crack cocaine. People are addicted. We experience withdrawals. We are so driven by this drug, getting just one hit elicits truly peculiar reactions.
I’m talking about Likes.
They’ve inconspicuously emerged as the first digital drug to dominate our culture.
Lovematically was designed to automatically like every picture that rolled through its users’ newsfeeds. If likes were digital crack, Lovematically’s users were pushing the drug at the heavily discounted rate of free. It wasn’t even necessary to impress them anymore; any old post was good enough to inspire a like. At first, for three experimental months, Chawla was the app’s only user. During that time, he automatically liked every post in his feed, and, apart from enjoying the warm glow that comes from spreading good cheer, he also found that people reciprocated. They liked more of his photos, and he attracted an average of thirty new followers a day, a total of almost three thousand followers during the trial period. On Valentine’s Day 2014, Chawla allowed five thousand Instagram users to download a beta version of the app. After only two hours, Instagram shut down Lovematically for violating the social network’s Terms of Use.
“I knew way before launching it that it would get shut down by Instagram,” Chawla said. “Using drug terminology, you know, Instagram is the dealer and I’m the new guy in the market giving away the drug for free.” Chawla was surprised, though, that it happened so quickly. He’d hoped for at least a week of use, but Instagram pounced immediately.
—
When I moved to the United States for grad school in 2004, online entertainment was limited. These were the days before Instagram, Twitter, and YouTube, and Facebook was limited to students at Harvard. I had a cheap Nokia phone that was indestructible but primitive, so the web was tethered to my dorm room. One evening, after work, I stumbled on a game called Sign of the Zodiac (Zodiac for short) that demanded very little mental energy. Zodiac was a simple online slot machine much like the actual slot machines in casinos: you decided how much to wager, and then you lazily clicked a button over and over again and watched as the machine spat out wins and losses. At first I played to relieve the stress of long days filled with too much thinking, but the brief ding that followed each small win, and the longer melody that followed each major win, hooked me fast. Eventually screenshots of the game would intrude on my day. I’d picture five pink scorpions lining up for the game’s highest jackpot, followed by the jackpot melody that I can still conjure today. I had a minor behavioral addiction, and these were the sensory hangovers of the random, unpredictable feedback that followed each win.
My Zodiac addiction wasn’t unusual. For thirteen years Natasha Dow Schüll, a cultural anthropologist, studied gamblers and the machines that hook them. The following descriptions of slot machines come from gambling experts and current and former addicts:
Slots are the crack cocaine of gambling.
They’re electronic morphine.
They’re the most virulent strain of gambling in the history of man.
Slots are the premier addiction delivery device.
These are sensationalized descriptions, but they capture how easily people become hooked on slot machine gambling. I can relate, because I became addicted to a slots game that wasn’t even doling out real money. The reinforcing sound of a win after the silence of several losses was enough for me.
In the United States, banks aren’t allowed to handle online gambling winnings, which makes online gambling practically illegal. Very few companies are willing to fight the system, and the ones that do are quickly defeated. That sounds like a good thing, but free and legal games like Sign of the Zodiac are also dangerous. At casinos, the deck is stacked heavily against the player; on average the house has to win. But the house doesn’t have to win in a game without money. As David Goldhill, the C.E.O. of the Game Show Network, which also produces many online games, told me, “Because we’re not restricted by having to pay real winnings, we can pay out one hundred and twenty dollars for every hundred dollars played. No land-based casino could do that for more than a week without going out of business.” As a result, the game can continue forever because the player never runs out of chips. I played Sign of the Zodiac for four years and rarely had to start a new game. I won roughly 95 percent of the time. The game only ended when I had to eat or sleep or attend class in the morning. And sometimes it didn’t even end then.
In contrast to free games, casinos win most of the time—but they have a clever way of convincing gamblers that the outcomes are reversed. Early slot machines were incredibly simple devices: the player pulled the machine’s arm (hence the term “one-armed bandit”) to spin its three mechanical reels. If the center of the reels displayed two or more of the same symbol when they stopped spinning, the player won a certain number of coins or credits. Today, slot machines allow gamblers to play multiple lines, in some cases as many as several hundred at once. The machine below, for example, allows you to play fifteen lines:
Say the machine charges ten cents per spin. If you decide to play all fifteen lines, each spin will cost you $1.50. Basically, you’re playing fifteen spins at once, instead of drawing the experience out by playing a single spin fifteen times. Casinos are very happy for you to play this way: if they’re going to beat you, they’ll do it fifteen times more quickly. But every time you play, you’re fifteen times more likely to win on at least one line, and the machine will celebrate with you by flashing the same bright lights and playing the same catchy tunes. Now imagine you play all fifteen lines, costing you $1.50, and one of your lines spins two bombs in a row, as line four does, above. If two bombs are worth a payout of ten credits, you get a payout of $1. Not bad—until you realize the net effect of that spin is a loss of fifty cents (your $1 payout minus the cost of the spin at $1.50). And yet you enjoy the positive feedback that follows a win—a type of win that Schüll and other gambling experts call
a “loss disguised as a win.”
Mike Dixon, a psychologist, has analyzed these disguised losses. With several colleagues, he focused on a game called Lucky Larry’s Lobstermania (which I found online and played for three hours while I was supposed to be writing this book—I was lucky that U.S. laws forced me to play the free version). Lobstermania allows players to spin up to fifteen lines simultaneously. The game features five reels with three visible symbols per reel, for a total of more than 259 million possible outcomes. Dixon and his team calculated that gamblers are more likely to strike a loss disguised as a win than a genuine win any time they play six or more lines per spin.
Losses disguised as wins only matter because players don’t classify them as losses—they classify them as wins. Dixon and his team hooked up a group of novice gamblers to electrodes while they played Lobstermania. He gave them ten dollars each, and told them they could win up to an additional twenty dollars. They played for half an hour and spun, on average, 138 times. After each spin, a machine registered minute changes in how much the students were sweating—a sign that the event was emotionally meaningful. Lobstermania, like many modern video slots, is full of reinforcing feedback. In the background, the bouncy B-52s song “Rock Lobster” plays over and over whenever you spin. It’s replaced by silence after losing spins and by louder, bouncier versions of the song after wins. Lights flash and bells ding just the same whether the spin represents a true win or a loss disguised as a win. The students sweated more when they won than when they lost—but they sweated just as much when their losses were disguised as wins as when those wins were genuine. This is what makes modern slot machines—and modern casinos—so dangerous. Like the little boy who hit every button in my elevator, adults never really grow out of the thrill of attractive lights and sounds. If our brains convince us that we’re winning even when we’re actually losing, how are we supposed to muster the self-control to stop playing?
After a string of losses, even die-hard gamblers begin to lose interest, some faster than others. This is a big problem for casinos, which aim to keep the gambler in front of the machine for as long as possible. It would be easy to change the odds of winning so that players become more and more likely to win after a series of losses, but, unfortunately for casinos, this is illegal in the U.S. The odds need to stay consistent across every spin, regardless of the previous run of outcomes. Natasha Dow Schüll told me that casinos have come up with some creative solutions. “Many casinos use ‘luck ambassadors.’ They sense that you’re reaching your pain point—the moment when you’re about to leave the casino—and they dispatch someone to give you a bonus.” These bonuses were either meal vouchers or a free drink or even cash or gambling credits. Bonuses are classified as “marketing” rather than a way of changing the odds of winning, so regulators turned a blind eye. With a new dose of positive reinforcement, gamblers tended to continue playing anew, until they reached another pain point after a series of losses.
It’s expensive, however, to keep dozens of luck ambassadors on the floor, not to mention paying a team of data analysts to identify frustrated gamblers. One man, a casino consultant named John Acres, proposed a creative solution that skirted the relevant laws. Schüll explained Acres’ technique. “As you play, a tiny portion of what you lose goes into a pot which counts as the marketing bonus pot. An algorithm within the machine senses your pain points, and knows ahead of time what the next outcome will be.” Normally the algorithm sits by and lets the machine dish out a randomly drawn outcome. When the player reaches a pain point, though, it intervenes. “If the machine sees that, oh, that outcome sucks,” Schüll said, “instead of BAR, BAR, CHERRY, it goes ‘chink’ and nudges the third reel so that it displays BAR—a jackpot outcome of three BARs.” Those winnings are taken from the “marketing bonus pot” that grew in size while the player continued to lose. Instead of relying on a human luck ambassador, the machine plays that role itself. Schüll has seen many dastardly tactics in her time investigating casinos, but this one she calls “shocking.” When she asked Acres how this wasn’t “a complete violation of laws in place to protect people from precisely this,” he replied, “Well, laws are made to be broken.”
—
The success of slot machines is measured by “time on device.” The longer the average player stays seated at the machine, the better the machine. Since most players lose more money the longer they play, time on device is a useful proxy for profitability. Video game designers use a similar measure, which captures how engaging and enjoyable their games are. The difference between casinos and video games is that many designers are more concerned with making their games fun than with making buckets of money. Bennett Foddy, who teaches game design at New York University’s Game Center, has created a string of successful free-to-play games, but each was a labor of love rather than a moneymaking vehicle. They’re all available on his website, foddy.net, and apart from attracting limited advertising revenue, they aren’t a significant source of income, despite some having achieved cult status.
“Video games are governed by microscopic rules,” Foddy says. “When your mouse cursor moves over a particular box, text will pop up, or a sound will play. Designers use this sort of micro-feedback to keep players more engaged and more hooked in.” A game must obey these microscopic rules, because gamers are likely to stop playing a game that doesn’t deliver a steady dose of small rewards that make sense given the game’s rules. Those rewards can be as subtle as a “ding” sound or a white flash whenever a character moves over a particular square. “Those bits of micro-feedback need to follow the act almost immediately, because if there’s a tight pairing in time between when I act and when something happens, then I’ll think I was causing it.” Like kids who push elevator buttons to see them light up, gamers are motivated by the sense that they’re having an effect on the world. Remove that and you’ll lose them.
The game Candy Crush Saga is a prime example. At its peak in 2013, the game generated more than $600,000 in revenue per day. To date, its developer, King, has earned around $2.5 billion from the game. Somewhere between half a billion and a billion people have downloaded Candy Crush Saga on their smartphones or through Facebook. Most of those players are women, which is unusual for a blockbuster. It’s hard to understand the game’s colossal success when you see how straightforward it is. Players aim to create lines of three or more of the same candy by swiping candies left, right, up, and down. Candies are “crushed”—they disappear—when you form these matching lines, and the candies above them drop down to take their place. The game ends when the screen fills with candies that can’t be matched. Foddy told me that it wasn’t the rules that made the game a success—it was juice.
Juice refers to the layer of surface feedback that sits above the game’s rules. It isn’t essential to the game, but it’s essential to the game’s success. Without juice, the same game loses its charm. Think of candies replaced by gray bricks and none of the reinforcing sights and sounds that make the game fun. “Novice game designers often forget to add juice,” Foddy said. “If a character in your game runs through the grass, the grass should bend as he runs through it. It tells you that the grass is real and that the character and grass are in the same world.” When you form a line in Candy Crush Saga, a reinforcing sound plays, the score associated with that line flashes brightly, and sometimes you hear words of praise intoned by a hidden, deep-voiced Wizard of Oz narrator.
Juice is effective in part because it triggers very primitive parts of the brain. To show this, Michael Barrus and Catharine Winstanley, psychologists at the University of British Columbia, created a “rat casino.” The rats in the experiment gambled for delicious sugar pellets by pushing their noses through one of four small holes. Some of the holes were low-risk options with small rewards. One, for example, produced one sugar pellet 90 percent of the time, but punished the rat 10 percent of the time by forcing him to wait five seconds before the casino would respond to his next nose
poke. (Rats are impatient, so even small waits register as punishments.) Other holes were high-risk options with larger rewards. The riskiest hole produced four pellets, but only 40 percent of the time—on 60 percent of trials, the rat was forced to wait in time-out for forty seconds, a relative eternity.
Most of the time, rats tend to be risk-averse, preferring the low-risk options with small payouts. But that approach changed completely for rats who played in a casino with rewarding tones and flashing lights. Those rats were far more risk-seeking, spurred on by the double-promise of sugar pellets and reinforcing signals. Like human gamblers, they were sucked in by juice. “I was surprised, not that it worked, but how well it worked,” Barrus said. “We expected that adding these stimulating cues would have an effect. But we didn’t realize that it would shift decision making so much.”
Juice amplifies feedback, but it’s also designed to unite the real world and the gaming world. One of Foddy’s most successful games is called Little Master Cricket, which does this very well. In the game, a cricket player hits one shot after another, scoring runs (or points) according to where those shots go. When he misses the ball or hits it in the wrong spot, he’s “out” and the game begins again at zero runs. “When I released Little Master, my wife was working at the head offices of Prada in New York,” Foddy said. “Much of the finance department consisted of cricket fans from India—and they were hooked.” When they discovered that their colleague was married to the game’s creator, they were starstruck. It’s very difficult to simulate the game of cricket in an engaging way, but Foddy somehow managed to keep the game simple but true to life. Players move the mouse back and forth in a way that mirrors the swing of a real cricket batsman. Just as in real life, the highest scoring shots in Little Master travel far through the air while avoiding the clutches of fielders who might catch the ball before it falls to the ground. (As in baseball, this renders the batsman “out.”) This sort of feedback, which ties the game to the real world, is called mapping. “Mapping is sort of visceral,” says Foddy. “For example, you should always use the space bar sparingly. It’s a loud, clattery key on the computer, so it shouldn’t be used for something mundane, like walking. It’s better saved for declarative actions that aren’t quite as common, like jumping. Your aim is to match sensations in the physical realm to those in the digital realm.”
