El Farol is a bar in Santa Fe, New Mexico. The bar is popular but becomes overcrowded when more than 60 people attend at any given evening. Everyone enjoys himself/herself when fewer than 60 people go, but no one has a good time when the bar is overcrowded. Each customer decides independently whether or not to go to the bar, based on his/her knowledge of previous bar attendance and a finite set of strategies. This scenario, originally proposed by the economist Brian Arthur, provides a simplified mathematical model of a class of congestion and coordination problems that arise in modern ICT systems. In this talk I shall review previous mathematical solutions and recent extension to the El Farol bar problem on networks, and discuss possible applications to the emerging field of cognitive radio networks.