New iPhone SE, 3 Apple Watch models coming? Moderna booster shot and omicron Popeye's meme kid now state football champ Squid Game: Most trending TV show in 2021 Best Christmas gifts 2021 PS5 restock tracker

Twitch Plays Pokemon: In conversation with the phenomenon's creator (Q&A)

One month after this 'social experiment' redefined online multiplayer, we virtually sit down with the unnamed creator to discuss the initial goals, what changed along the way, and what comes next.

Screenshot by Nick Statt/CNET

Thirty-six million views. 1.16 million unique players, with 121,000 connected simultaneously at one point. For a major videogame publisher, those numbers would be nothing to scoff at. (Indeed, that's more people than are playing Battlefield 4 online currently, across all five platforms.) For an independent port of an 18-year-old game originally intended to be played on the humble Nintendo Game Boy, that's nothing short of a marvel.

Built on top of the Twitch game streaming platform, Twitch Plays Pokemon (or TPP) was an attempt to play through Pokemon Red, where everyone in the chat room had a chance to control the game. It was all handled through the attached chat room, where viewers would enter commands. A straightforward IRC bot monitored the whims and fancies of the crowd, aggregating the results passing that on to the emulator running the game.

Its success was certainly a surprise to most, even to its creator, an Australian programmer who prefers to remain anonymous but who kindly agreed to give us an audience -- virtually, at least -- to look back at the project one month after it first went online on February 12, and to look forward toward what comes next.

Q: It's been just about a month since the initial Twitch Plays Pokemon began, where over a million people chipped in to play through Pokemon Red. Did you have any idea it would see this kind of success?

I had no idea it was going to be this successful, I was expecting ~300 concurrent viewers at peak as a best-case scenario.

I find the success of Twitch Plays Pokemon to be a little annoying. It's extremely unlikely anything else I would make would be this successful.

Screenshot by Nick Statt/CNET

Q: How long did it take you to implement the code necessary to make this work? About how much time would you estimate you've spent making updates to the system since then?

I wrote the code and put it up on a server all within a few days, I didn't put much effort into it because I didn't expect it to be successful; I just wanted to test to see if interacting with a game in this manner would have any interest at all. I was expecting to let it run for a short while (until interest died) and then take what I learnt and applied it to something more complex.

Since the explosion in popularity I've made extensive modifications to the software. It has a lot more functionality now, and it is also more reliable. I've also made many improvements to the overlay -- it's strange to go back and see the old one now.

I've lost track of how many hours I've spent working on TPP, and I expect there to be many more.

Q: Was it ever difficult to keep the system operational, given the volume of players?

No, not really -- my server just listens to a chat channel and sends a video stream to Twitch. It doesn't make much difference if there's one viewer or 100,000 viewers.

Twitch, on the other hand, has seen a a big strain on their servers due to the popularity of Twitch Plays Pokemon and its heavy usage of Twitch's chat functionality.

Q: After about a week, a change was made, adding a "democracy" mode in the hopes of giving more power to the majority to keep the game moving. This though, created a huge debate about the intrinsic nature of the project, with some saying that it compromised the core experience. Did you expect this sort of negative reaction?

Yes, I expected a negative reaction; people don't like change. Early on in TPP's life, I realized that the movement was too imprecise for parts of the game to be possible -- something needed to be changed.

The intrinsic nature of the project was to exist for a week before being shut down due to disinterest.

Q: Most people rate Pokemon Red as taking between 40 to 50 hours to complete when playing solo. It took the players on Twitch just over two weeks. Did you have an idea of how long it would take the group to complete the game? Did you think they could complete it at all?

I didn't think it was possible. I thought the number of viewers would be too low allowing for trolls to easily release every Pokemon, delaying the game's progression indefinitely.

Q: Now that the project has moved on to other Pokemon games, do you think the group will get faster at completing them, or even slower?

I think that now that many people in the group are experienced, it will be easier for them to communicate and perform (relatively) complex tasks.

Q: What are your thoughts on the other Twitch Plays games that this effort has spawned?

Seeing how other games worked with the same format made me feel right in saying that Pokemon is by far the most ideal game for such a simple method of input.

I was a little disappointed in seeing how many of the streams had the same functionality as TPP. I was hoping to see much more variation and experimentation.

Q: In your estimation, how long will gamers remain interested in this sort of social experiment? Will Twitch gamers still be collaborating to get through Pokemon and other games a year from now? Five years?

I was expecting to shut the stream down within a week or two due to disinterest; inputting button presses into a chat window and waiting to see the effect didn't sound like it would be all that appealing.

I'm not sure how much longer this sort of thing will be relevant, but I'm planning on running my stream for as long as there's still interest.

Watch live video from TwitchPlaysPokemon on