For this month’s repost, I’m publishing up an article I wrote in explanation of a programming project in 2018. In theory you could find it on Github, but to maintain a layer of pseudonymity I’m not linking it directly. A few minor revisions have been made to adapt to the audience.
Introduction
The goal of this project is to create Markov Chain simulations showing that the card game Dominion contains phase transitions, much like the physical phase transition between liquid and solid.
Dominion is a popular card game created in 2008. In Dominion, each player has their own deck, and they add/remove cards from their deck over the course of the game. Each game has a unique set of cards available to be added to players’ decks, making the optimal strategy in each game different. However, there are two archetypical strategies, based on two fundamentally different decks. The “Big Money” deck makes the best of the 5 cards drawn each turn. The “Engine” deck includes cards that draw more cards, and tries to draw itself in its entirety each turn.
Because of my background in physics, I recognized that the line between “Big Money” and “Engine” strategies is a phase transition. More specifically, it’s a one-dimensional percolative transition. That explains why there is such a strong dichotomy between the two strategies over a wide range of conditions.