A group of more than one million people have successfully joined forces to complete the same Gameboy game. The experiment allowed participants to control the game one button-press at a time, taking 16 days to complete it.

The game was Pokemon Red/Blue from the original GameBoy, played via an emulator and displayed on Twitch, a site that lets users stream their gaming experience so that others can watch them in action.

As an experiment, the site decided to hook its chat feature up to operate the controls. Users simply typed in one of eight commands to match the original controls (A, B, Start, Select and four arrow keys) and the instructions were immediately applied.

The big question was whether the number of players working together and pooling their collective wisdom would outweigh those players who were deliberately trying to sabotage the project.

After a few days it was becoming hard to judge that because players were issuing commands so rapidly that there was a delay of up to 40 seconds for each command to take effect.

Organizers then decided to offer users an extra option, typing either “anarchy” or “democracy” and keeping a running total. Whenever anarchy was in the lead, the original set-up continued with every instruction applied in order.

Whenever democracy was in the lead, only the most-typed instruction in each five second period was carried out.

It took a combined 122 million instructions from players to complete the game, though the actual number of commands carried out will have been considerably lower.

The experiment is now being rerun with Pokemon Crystal (pictured above). This time round democracy mode is automatically restored at the start of every hour and the “voting period” has been extended to 30 seconds for each command.

There are plans to try it out with more complicated games, though anything that requires too rapid a response from players is unlikely to work, particularly in democracy mode. One possibility might be to divide players into two teams to remove the disadvantage of what might be termed crowdsourcing lag.