Beating Nintendo's classic videogame The Legend of Zelda is hard enough on your own. So it's kind of amazing that hundreds of players—each tugging our elfin hero Link in different directions—were able to beat the game three times over in less than a week.
This experiment played out on Twitch Plays Zelda, an online emulation of the 1986 single-player fantasy game. Anyone could influence the game by typing instructions into a chat window. If you entered "up," Link would move up a pixel. But if someone else typed "down," your command would be overridden.
Instead of making Link walk forever in tight circles or constantly bump into walls, players managed to defeat the final boss so many times they eventually got bored. The Twitch Plays Zelda channel is now streaming Pokemon, the game that launched the Twitch Plays trend in mid-February.
I wanted to know how exactly this crowd-sourced confusion led to success. So I met up with Zach Gerlock, the La Jolla-based software developer who programmed the game. Here's an edited version of our conversation.
Why did you set up Twitch Plays Zelda?
Well, I obviously got the idea from Twitch Plays Pokemon. It was one of those ideas that seemed so obvious, I thought, "why didn't I do that?" So I figured I would do that.
With so many people all trying to steer the ship, how come this game wasn't a complete disaster?
Basically, you need to eliminate the possibility of failure. Hypothetically, given enough time, a random number generator could beat it, but it would take 100 years or something.
And the Zelda players took a lot less time?
The first round took like 62 hours. I was surprised that it went that fast! The second quest took like 50 hours. The third time through, there was only like 10 people actively playing. They managed to beat the game in 16 hours.
Could you explain the dynamics of the game?
It's a single-player game not designed to be played by more than one person. You're all one character. So if you have any ownership, it's shared equally with everyone. And anything you do affects everyone in the stream. So in a way, you're more connected to the other players than you would be in a traditional multiplayer game. But you have no individual stakes.
How is any progress ever made?
For Twitch Plays Pokemon, there's two different modes: anarchy and democracy. In anarchy mode, every command anyone puts in gets executed in the order it was taken. With democracy mode, the game waits for a few seconds of commands to accumulate, and then does the single most voted-for command.
And there's some disagreement about which mode is better.
It's gotten really weird in a lot of ways. These tongue-in-cheek "religions" have popped up in Twitch Plays Pokemon. They're related to anarchy and democracy. In the game, there's a choice between a helix fossil and a dome fossil. The helix came to represent anarchy and the dome came to represent democracy. You have different factions lobbying for different things.
So you're not fighting against your opponents. You're fighting with them?
Some people might be trying to sabotage the game. Or they might not be trying to sabotage the game. Some people just think the concept of democracy is against the spirit of the game. They just don't believe in democracy in this context.
Why don't they believe in democracy?
[Laughs] I think they probably believe in it in general. But in the beginning, there was only anarchy. Then, when it exploded too much, democracy was implemented. It wasn't an option. It was actually a change to the system, and a lot of people didn't like that.
Which system was in effect for Twitch Plays Zelda?
Every entry got picked up in Twitch Plays Zelda. I might have had to consider democracy if I had more players. Twitch Plays Pokemon is hovering around 100,000 active watchers. My peak was like 1,500. The dynamics of Twitch Plays Pokemon versus Twitch Plays Zelda are completely different just because of the player count.
Isn't part of the fun that it is frustrating? That Link doesn't always do what you individually want him to do?
There's a difference between frustration that's interesting and frustration that's not. I had to write a script modifying the game to allow item-switching. Normally, you have to bring up the pause menu, scroll over to an item, and then un-pause. But I made it so you could just type the name of the item in the chat and it'll switch. Having the game in the pause menu the whole time would just be too boring.
These are games that lots of people probably remember playing in their original forms. Is nostalgia part of the appeal here?
Yeah, I think Pokemon is really perfect for that in a lot of ways. The original game came out in the late '90s, and there's been a pretty steady release of Pokemon games ever since then. Anyone born from pretty much the late '80s on probably grew up with Pokemon.
Are you planning to host different games?
I've been trying to get Dragon Quest to work. It's similar to Pokemon in a lot of ways. But the one fatal flaw is that every time you die, it asks if you want to keep playing or not. So there's a chance the players would say no or erase the save file. I'm also working on Link's Awakening, another game in the Zelda series.
What has been interesting to you about this whole trend?
It's brought people together in an interesting way that other online games haven't. Everyone's sort of in it together. And no matter how well-intentioned everyone is, there's going to be a lot of mishaps. I wouldn't be surprised if we saw games designed around a concept like this in the future.
