Rock, Paper, Scissors is, when it comes right down to it, a silly, unwinnable two-player game: if one player makes random moves, then the game, on average, ends in a tie.
Humans, however, are somewhat predictable. (So are computers, for that matter, but let's ignore that for the moment.) Therefore, we at essentially.net thought it would be interesting to see what happens when we set humans against computers in a massive, ongoing Rock, Paper, Scissors tournament.
We're using a custom, "intelligent" algorithm that ought to adapt well to the way people play the game. Now, keep in mind we don't even hope that it'll win all of the time; that's not even remotely likely. But if we get the computer to win even 55% of the time (note the "win percentage" below), we'll be thrilled, since that'd be a convincing demonstration if enough individual rounds of the game are involved.
So far, here's how the computer is doing:
| Total rounds: | 3864879 |
| Total wins: | 1745184 |
| Total losses: | 1073658 |
| Win/loss ratio: | 1.63 |
| Win percentage: | 61.91% |