Minority Game, explained.
A minority game rewards players for choosing the less crowded of two sides, making an action's value depend on how many others choose it.
Why it happens
An attractive side can cease to be attractive when others imitate it. The payoff cannot be determined from the action alone. The participant also contributes to attendance, so evaluating the crowd without counting your own choice misses part of the rule.
A minority game rewards participants for being on the less crowded side. A strategy that is attractive to one player can lose its advantage when enough others adopt it.
Read the result
Choose A or B and compare the 101-person totals. The exact winning probabilities use the fixed bot behavior. A few victories do not demonstrate a strategy that would succeed against adapting opponents.
A worked example
Choosing the quieter service
One hundred other users choose between two equivalent services; you choose too.
If most others tend to choose A, B is more likely to be the minority. If everyone learns that pattern and switches, the advantage can move.
The fixed-bot game demonstrates crowd-dependent payoff but leaves that adaptive response out.
OPTIONAL DEEPER DETAILGo deeper: inside the model
Inside this model
One human and 100 bots choose two sides. Each bot independently chooses A with the stated probability. The 101-player total prevents a tie; everyone on the smaller side wins one point. Bots do not learn. Exact binomial probabilities compare the chance of winning on either side.
Where this idea is useful
A practical use
When choosing a shared facility or service time, predicting others' choices can matter more than an option's intrinsic appeal.
A common misconception
“Always oppose the crowd and you will win.”
You need a relevant prediction of the actual crowd, and others can respond. Being contrarian is not an advantage by itself.
What this explanation leaves out
- The fixed bots omit the learning and adaptive strategies studied in the original minority-game research. Choosing a minority is not a general-purpose rule for social or market decisions.
Why are there 101 participants?
An odd total prevents equal attendance between the two sides. Your own choice is counted with the hundred bots when deciding which group is smaller.
Would this option still be attractive if everyone noticed the same advantage?
Associated thinkers
Further reading
Explore the original research or the teaching reference behind this experiment.