On a Thursday night in September, I raced from Midtown to Bushwick for an impromptu conference organized by Arikia Millikan in what was dubbed a mansion, but I understood to be a large house. I sat on a wooden floor as ten people talked for ten minutes each, all speaking about secrets. One such person was Wired columnist Clive Thompson, who told us how gamers had solved a decade-long scientific mystery in a single month. As a suspicious non-gamer, I was amazed to find altruism within the World of Warcraft. Weeks later, we met at a café in Park Slope to discuss how the increasing complexity of video games led to groupthink, and how groupthink has been harnessed by researchers for scientific gain.
—Erika Anderson for Guernica
Guernica: How would you describe the evolution of video games?
Clive Thompson: When games started out, they were very, very simple affairs, and that was partly just technical—you couldn’t do very much. They had like 4K of memory. And so the games started off really not needing instructions at all. The first Pong game had one instruction. It was, “Avoid missing ball for high score.” So it was literally just that: don’t fail to hit the ball. I remember when I read it, it was actually a confusing construction: avoid missing ball for high score. It’s weirdly phrased, as if it were being translated from Swedish or something, you know? But they didn’t know what they were doing.
But what started happening very early on was that if you were in the arcades as I was—I’m 44 in October, so I was right at that age when these games were coming out—the games were really quite hard in a way, and because they were taking a quarter from you, their goal was to have you stop playing quickly because they need more money. They ramped up in difficulty very quickly, like the next wave is harder, and the third wave is unbelievably harder. And so you had to learn how to play them by trial and error with yourself but you only had so much money. And so what you started doing was you started observing other people and you started talking to all the other people. What you saw when you went to a game was one person playing and a semi-circle of people around them and they were all talking about what was going on, to try to figure out how to play the game. And they would learn all sorts of interesting strategy.
But if you have tens of thousands of kids in these arcades talking and observing and sharing notes, it’s a little bit like a scientific process whereby everyone notices one little fact and you slowly compile them into a theory of gravity or a theory of how cell biology works.
So the early stuff was literally strategy, and they discovered some very clever things, like in Asteroids, there was this strategy called lurking, whereby if you parked, if you got rid of almost all the asteroids except for one little one that would be sort of soaring through the air, you could hide in the corner of the screen and it would take a long time for it to come anywhere close to hitting. You could use this strategy to hunt for hours. You would sit there for hours getting more and more points.
So there are all these little strategies like that. But there was also like weird little bugs inside the game that weren’t intentionally put there, that people would discover. Like there were certain situations when you were being chased by the red ghosts of Pac-Man, certain places on the board where you could suddenly go racing right through the ghost without being hit. That was not intentional, that’s just a bug, and there are little bugs like that. And these are incredibly difficult things to notice, I mean the designers didn’t notice them. But if you have tens of thousands of kids in these arcades talking and observing and sharing notes, it’s a little bit like a scientific process whereby everyone notices one little fact and you slowly compile them into a theory of gravity or a theory of how cell biology works.
What started happening is that the game designers began to intentionally put secrets inside the games. The first one was Adventure, it was a game on the Atari 2600, and back then you didn’t know who had made the game, there were no credits anywhere. The designers would be hired for six months to produce a game and they were dispatched and it was this freelance economy, and they didn’t like not having any credit, so this guy decided to hide the credits inside the game. And he made a secret room inside Adventure and the only way to find it if you’re like stumbling around, trying to find something and get out without getting killed by these dragons, there’s a single pixel on one room—you might not even notice it, right?—but if you picked it up, you could go through a wall—an unmarked area in a wall—and get to this room, and inside the room it would have his name spelled out. It was incredible. There were no clues at all, but somehow kids started finding this. Someone would blunder through the wall, and they’d tell someone else and it just went through the grapevine and that’s the first example of an intentional secret put inside a game that was discovered by this sort of groupthink process that kids all playing and talking in school about what’s going on.
For a little while, three to four years ago, the World of Warcraft wiki was the second largest wiki in the world after Wikipedia.
And that was sort of the beginning of what became an arms race, because essentially, as more kids started talking to each other and sharing information, any secrets inside the game would get discovered very quickly. And so the game designers started responding by putting more stuff inside the game that was harder to find, because they knew that kids were not just playing the game by themselves but playing it collectively, and so they weren’t just designing a game for one player, but for a 100 or 1,000 networked players, and a 100 or 1,000 people talking to each other is much smarter than one person individually.
And so this began this explosion between the mid-80s to the mid-90s of games intentionally including stuff that was there to be discovered. Virtua Fighter 3, a 1996 game, was typical of what would start to happen. You pick one of eight or nine characters, and each of them listed like eight moves: If you shove forward twice and punch, you get the special punch; if you forward backward up and kick, you get the special kick. But you quickly realized that there were a lot of secret moves. As it turns out, there was something on the order of hundreds of secret moves for each player.
The first piece I ever wrote about video games was in 1996 for Toronto Life magazine. I wrote about collective problem-solving of people in arcades figuring out this game, because there wasn’t really any Internet back then. There were Usenet groups where you were beginning to collectively share and make public these lists of moves, but they were pretty incomplete. You really had to go to the arcades if you wanted to participate in this puzzle solving. But very quickly that moved online. What would start happening is that someone would start writing a document as soon as the game came out. It was like, “Here’s what I found,” and they would encourage other people to email them stuff. You would get very long documents, going into the thousands and thousands of words with this list of credits at the end, of everyone that had participated sharing information for it. It really began to read like scientific collaboration, “This is what we found, and here’s a list of collaborators, thanks to everyone for doing this stuff.” And that was an informal online sharing process that got accelerated by the advent of wikis and game FAQs such that there started to become destination sites where people would go—it’s still there now—called Game FAQs, where you put your partial document up and people start immediately sending you stuff, or now the wiki where anyone can contribute. For a little while, three to four years ago, the World of Warcraft wiki was the second largest wiki in the world after Wikipedia. [Laughs] So many people had added maps, and they’d added clues, and they would find something that hadn’t been documented, and they would document it.
It’s one of the most interesting areas of collective intelligence I’ve seen, and it emerged totally organically.
In a way, it’s sort of a problem for game designers because they have to design a game now for 100,000 networked people, which is actually quite hard to do. One of the responses is that in the casual gaming world, there are no secrets anymore, right? With a really simple game that’s played on your phone, on the level of Tetris, it’s just about pattern recognition and problem solving. There aren’t any really complicated strategies, it’s just about getting better at one simple thing. And those are games that you really can play without collective intelligence. There are still a lot of games like that and a lot of game designers like making those because in a way it’s almost easier than trying to figure something out that’s going to satisfy this enormously intelligent collective mob of game players. It’s one of the most interesting areas of collective intelligence I’ve seen, and it emerged totally organically.
The funny thing is, because I was prepared by seeing that, I began to recognize when the same thing was happening over and over again, like the whole fact-checking thing that’s happening in politics. It used to be that you could say something kind of with impunity back in the 80s and 90s—who was going to check it? But now, you know, you have hundreds of people paying attention to every minor utterance and checking it and getting together online and showing their notes. And it’s the exact same process. You see it over and over and over again. Someone puts out a product, and it’s bad or it’s wrong, and people immediately start talking about it and figuring out what’s wrong and documenting it and this process happens in almost every area of life right now.
But honestly the first place I ever saw it happen was in video game players, because they were sort of confronted with these problem-rich environments, and they were very technologically savvy, and they had an offline tradition of talking to each other in arcades. That was crucial. They were already fluent in the idea that you needed to talk about these things collectively to figure them out.
Guernica: How has this been harnessed for biochemists?
Clive Thompson: The Foldit stuff—this is awesome! In some respects, Clay Shirky talks about the cognitive surplus, which is the idea that with the advent of automation in the twentieth century, we started to have a lot more leisure. Actually that began even in the nineteenth century. And what we did with that for like sixty years with all that leisure time is we watched TV. That is how we used our leisure time. In fact, we did so much of it that it was basically a part-time job. You had your day job, which was going to your job and you had your part-time job, which was watching TV. You know, that’s how many hours it was, right? That’s what Clay points out.
If you shave off the tiniest hundredth of a percent of the amount of time we spent watching TV, you get Wikipedia.
What started happening with the Internet was that people started realizing they could participate in things that were creative even if they were stupidly creative, like LOLcats or enormously collectively creative like Wikipedia or blogging. If you shave off the tiniest hundredth of a percent of the amount of time we spent watching TV, you get Wikipedia. That’s what it is. It’s like some ungodly tiny fraction of the time we used to spend watching TV. The more you shave off, the more Wikipedias you get, right? And so there’s this way of thinking of the cognitive surplus, as Clay calls it, as this intellectual resource that you can tap into. But you have to think of, you know, what’s a structure that will make people want to participate? How do you make it sort of fun and interesting and challenging? Wikipedia did that pretty well, but it’s actually rare. Most people try and create something thinking everyone’s going to show up, but usually it does not work. People do not show up. It’s not that they don’t care, it’s that you haven’t made something interesting.
In a way, science is really rife with this ability to tap into this cognitive surplus, because lot of science requires a lot of tiny bits of work and if you can collectively get them out of people, you can do a lot of work quickly. But how do you make them want to do it? So there were these guys who were working on protein folding. DNA produces proteins, our bodies are made entirely out of proteins. A protein is an incredibly long molecule, it’s like a long piece of string, it’s all wadded up into a ball. And so any protein can be folded in—I’m not even sure what it is—billions of different ways, I guess. I don’t know what the numbers are. But every different way it’s folded it behaves differently. One of the great puzzles of modern biology is figuring out how a protein folds correctly, how it got into that shape. And the more you can understand about that, the more you can design drugs that will be able to attack its particular shape and structure. Or you can predict what it’s going to do inside the body. But it’s really, really hard to do.
And so what they started doing in the early twenty-first century, was to design computer programs that would automatically fold them over and over again in semi-random ways to see if they could hit upon a new combination. But because there’s so many ways to do it, there isn’t enough computing power. You get a supercomputer doing it and you’re still not going to scratch the surface of the different ways you can fold that protein. So a bunch of scientists decided, “Why don’t we do this classic thing that people started doing, we’ll create a downloadable screensaver application, and we’ll let people download it, and it’ll start doing folding, and if we get thousands of people to do that it will be like having extra super computers. And everyone’s computer will be folding and folding.” So they created the thing and sure enough, people would download it, and you know, their computers would be folding it, and it was kind of fun, because when you let your computer go into idle, you could look over and see it actually folding over and over and over again, right?
What started happening is that they started getting emails from people saying, “the program is just folding pretty randomly, and I can see when it sort of folds and it just does something stupid and it’s just going down a bad pathway. And it would be better if I could intervene and do the folding myself. I think I actually see a way to do that.” Because it was kind of graphical. It was very beautiful. And some of these guys were actually video game players and they realized, “We could make this into a game.” And so they hired some people who are really good at design and they produced a really good-looking thing that is very video game-like. And they basically said, “OK, now, let people fold them on their own.” Thousands of people started doing it and the hilarious thing is that they did exactly what you do with any video game. They immediately set up a wiki where they started sharing all their strategy, you know, what they found worked and didn’t. They produced these long FAQs. They had discussion boards where they talked at length about different strategies they’d found and how to work with the software to get the best strategies as fast as possible.
They’d been working on trying to fold this protein for about a decade [a protein in simians called M-PMV, an enzyme that could be used to fight HIV/AIDS], and they had not cracked it, and these guys did it in about a month. [Laughs] And so there’s this paper in Nature Structural & Molecular Biology, which has the scientists and the folding teams.
In World of Warcraft, you’ll see similar types of very scientific strategies. So every time they attack a monster and they can’t kill it, they’ll observe what they did and they’ll put it into a spreadsheet, and they’ll repeat that over and over again, and they’ll slowly build a model, a scientific model of how that monster behaves, right? And they’ll use that model to predict what actually worked. This happens all the time. Fourteen-year-old kids do this. They create models. They go, “We’ve observed this, and we’ve observed this, and we’ve observed this. Now we can use that model to predict that if we do this, we’ll actually win.”
And that’s exactly what the scientific model is, right? You observe something in the world around you, and you go, I think it works this way. And you try and attack it that way, and it doesn’t work, you write down what went wrong, and you do it ten to twenty times, and eventually you get a better hypothesis and eventually that hypothesis works. That’s exactly what they do in the World of Warcraft and so it’s exactly what they did with this. Instead of thousands of computers doing it randomly, they’re thousands of very smart humans, each one with an interesting intuition, and if those people can get together and share their intuitions, you can get a lot more work done enormously more quickly. And that’s exactly what happened. So if you look at the paper it lists the Void Crushers as one of the groups. There’s another one, I don’t remember what it’s called.
Guernica: Foldit Contenders.
Clive Thompson: Foldit Contenders, that’s right. And so these two groups were exactly essentially like raiding parties in the World of Warcraft. And that’s what you do, you give your raiding party a name, you give your guild a name. That’s exactly the stuff you’d see. And so by turning it into sort of like a game-like structure, they tapped naturally into this emergent behavior that had been kicking around for like fifteen years, right? And that’s what was so brilliant, is that they worked within an existing culture.
The one thing that games teach you over and over and over again is that complex problems are interesting problems, and they only can be solved if you tinker with them and do lots of experiments, and fail over and over and over again.
And now these guys are really starting to think about whether there are other scientific problems they can begin to approach in this way, other ways to create game-like interfaces that will tap into that culture of collective problem solving. It’s not easy to do because a good game is really hard to design. And a good game-like thing is hard to design. It’s easy to make bad games. It’s hard to make good games. That’s why the vast majority of these things don’t work. It’s not like some sort of magic sauce you can just sprinkle around, you have to create an architecture for participation that makes people really really want to participate, and that’s not easy.
People are always comparing games to movies, “When are games going to be more like movies?” And this is such a disastrously horrible way to think about video games, because video games have much more in common structurally with backgammon or with football or with any game, whereby it’s a constrained environment, you’re trying to do something that’s a difficult goal. And it’s hard to do because you have all these crazy rules, like in basketball, if you stop dribbling you have to stop moving. If you let people just walk around with a ball, basketball would be easier to play but it would be a more boring game. The constraints are what make a game interesting. This has nothing to do with the way movies work. The reason why people compare video games to movies is because they have absolutely no intellectual training in thinking about ludology, which is the philosophy of play and the structure of play. The other thing I say is that if you wanted to compare video games to an older form, you would compare them to a detective novel, which is that it’s this enormously weird puzzle that you’re trying you’re trying to solve, and the fact that it’s so hard that you can really only—the complicated ones you can really only solve with can solve with the assistance of other people.
And the other thing that it actually compares to is the scientific method—here’s a big puzzle, we’re going to get together, we’re going to share our data, we’re going to try and solve it. So it’s very interesting.
Guernica: Do you think games like FoldIt can create a generation of future biochemists?
Clive Thompson: Yeah, absolutely. It’s not like you’re going to get more than one out of 10,000 players that will become a biochemist, but I am absolutely convinced that there are people whose kids will get totally into it because of that. We know from decades now of academic research that what inspires children to learn, to do something interesting, is to give them a real-world problem to work on, and not some artificial problem set that they know is not connected to reality. I’ll bet there’s fourteen, twelve-year-old boys and girls who have played the game and are super interested in biology because they actually participated in a real-world problem, and not the completely artificial, fake, weird, stunted, bad biology they’re given in school.
School gives you a succession of really uninteresting problems instead of a really big hard one that actually requires you to work with other people. They give you uninteresting problems and forbid you from working with anyone else.
In fact, one of the reasons why kids like video games, and studies have shown this over and over again, is because in a strange way they are like a pressing, interesting, real problem. And we like to solve problems. School’s very bad at giving people interesting problems. They give you a succession of really uninteresting problems instead of a really big hard one that actually requires you to work with other people. They give you uninteresting problems and forbid you from working with anyone else, right? Now wonder kids get bored. No wonder they’re more interesting, right? What would you rather do? Work on something really big and hard and interesting where you have to interact with your friends, argue about strategy and stuff like that, or just sit around with totally uninteresting stuff and being forbidden from talking to anyone?
Guernica: Do you see this as a way to revolutionize the educational system?
Clive Thompson: It’s already being done a little bit. There’s a good book out there by Kurt Squire. There’s two leading academics, one’s called Kurt Squire, one’s called Constance Steinkuehler, who’s actually currently on leave working for the Obama administration, and they’ve done this pioneering research looking at the ways in which video games can be used in classrooms. There’s been this interesting debate: Can you use games in ways that actually get kids to learn stuff? And one of the difficulties is that from my perspective, games are bad at teaching the types of things that schools traditionally want to deal with, which is content, like what happened in the Civil War. I actually think they’re terrible at that. And that’s partly because they’re so good at doing something that the schools don’t do at all, which is teaching how to deal with complexity. Any problem that’s interesting is also really complex—if you do A, later on you’re going to get bit in the ass by L, which happened because of G, which happened because of B, which happened because of A, you know? That’s an incredibly important life skill that is useful whether you are dealing with family members’ emotional stuff, running a company, you know, dealing with an illness. We deal with complexity all the time, and schools are terrible at teaching complexity.
The one thing that games teach you over and over and over again is that complex problems are interesting problems, and they only can be solved if you tinker with them and do lots of experiments, and fail over and over and over again. And they’re good at basically putting you in an environment where failure is totally okay. In fact, failure is crucial.
Guernica: Going back to FoldIt, have you ever played it?
Clive Thompson: Yes, it’s really hard. I was not that good at it. But that was largely because I hadn’t gone through the training. I think if I sat down for a day and went through a bunch of the FAQs, learned the stuff, watched the video—that, again, not the creators of the game, but the players put together—I’d probably get pretty good at it. I was reporting on it, and I was like, let me try this, and I was like, “Holy balls, this is hard.” I knew a little bit of the advice because I’d looked at some of the stuff, but I hadn’t approached it the way I approach a video game, which is I start playing it, and whenever I’m puzzled or finding something really hard, I go to the guides to sort of suss something out.
One of the things that’s really interesting about video game players, also—and this is documented in a great book Mia Consalvo wrote a book about cheating. In one way, having complete walkthroughs of games, won’t they use it to just go through the game without any challenge at all? But the vast majority of players basically treat it as assistance for when they are stuck. When they tried really, really hard, and they’re facing a problem they alone can’t solve. They want to solve the game themselves as much as they can. What the collective intelligence is for is in situations where the game designers have essentially done a bad job designing the puzzle, right? [Laughs] Or when there’s something that is needlessly difficult, or are aware that this is the type of problem that they have no interest in. But with something like FoldIt, you really want as much advice as you possibly can, because you’re trying to ramp up as hard as you can go. Before I played the game, I would really immerse myself in the training that people have put online.
It’s pretty interesting. You should try it.
Erika Anderson is an online editor for Electric Literature, teaches with the Sackett Street Writers’ Workshop, writes a book review series for Hunger Mountain, and tweets for the Franklin Park Reading Series. She has an MFA from Vermont College of Fine Arts and lives in Brooklyn.
Clive Thompson writes on science, technology, and culture. He is a contributing writer for the New York Times Magazine and a columnist for Wired magazine. He also writes for Fast Company and Wired magazine’s web site, among other places. He has a book forthcoming from Pengiun provisionally titled Outsmart: The Future of Thought in the Age of Machines.