Strategy of Rock Paper Scissors

When I did my regular browse on BGG tonight, I ran into an interesting discussion thread (you may follow the title link). It talked about Rock Paper Scissors (RPS). Some readers were surprised by the claim of existence of strategy at playing RPS. They misbelieved that RPS was a mere random act, and could not admit the use of strategies.

I have studied a bit about this before. From a psychological view, there are well-known techniques to influence a RPS decision. There are also well-perceived patterns of RPS decisions, say a tendency to play Rock more often. These make RPS decisions no longer pure random results. It is particularly true if multiple games are counted than playing just a single game.

Game theory also suggests that a strategy of pure random actions is an inferior one. Randomized decisions generally makes below-average results, because such strategy does not take into account your opponent's model. Most people cannot understand this well. They even believe random is optimal at the situation of uncertainty. There are many researches and studies showing that it is not the fact. There are some good players who are capable of winning RPS more often.

To further support the idea, there is an annaul programming competition about computer RPS algorithms. Random is hence justified by the competition results that it is not an optimal strategy. It may not be the worst one, but never the best one neither.

As a side note, there are often situations in gaming which your opponent is asked to randomly draw a card from your hand. At the very first glance, it sounds like a random pick. However, there usually exists behavior pattern as discussed in social psychology. I sometimes make use of this piece of knowledge to arrange my hand of cards and "direct" him to "randomly" pick the card I intented. It works fine more than 50% of the time.