Skip to Content

Machine learning is about to revolutionize the study of ancient games

AI, computer modeling, and data mining are tools for a new field focusing on how games have evolved.
An image from Libro de los Juegos
An image from Libro de los JuegosWikimedia Commons | Libro de los Juegos

In 1238, the medieval Spanish ruler Alfonso X of Castile published a tome called Libro de los Juegos,  or The Book of Games. It consisted of 97 parchment pages, many with beautiful color illustrations, and contains the earliest descriptions of games such as chess, dice, and backgammon.

Alfonso went on to classify games into three categories: games that are played on horseback, games played dismounted (such as fencing and wrestling), and games played seated. He divided this third category even further into games that rely on the brain, games of chance, and games that rely on both.

In making these distinctions, Alfonso is the unofficial founder of a field of science known as ludology—the study of games, which has attracted much interest among mathematicians, computer scientists, sociologists, and others.

But there is an area of ludology that has been neglected. That is the study of ancient games, those played by Alfonso and other historical figures, which has never evolved into an independent discipline.

Today, that looks set to change thanks to the work of Cameron Browne at Maastricht University in the Netherlands and colleagues. They are pioneering a new area of archaeology focused purely on games. The goal is to better understand these ancient games and their role in human societies, to reconstruct their rules and to determine how they fit into the evolutionary tree of games that has led to the games we play today. They call this discipline archaeoludology.

The researchers have ambitious plans for their incipient science. They say the new techniques of machine vision, artificial intelligence, and data mining provide an entirely new way to study ancient games and to build a better understanding of the way they have evolved.


The first task in archaeoludology is to define the activities of interest. Browne and co are interested in traditional strategy games in which good decisions defeat bad ones, and games that reward mental rather than physical skill. In other words, no sports. They are also particularly interested in games that have historical cultural relevance.

However, they do not entirely exclude games of pure chance.  A good example is the family of games related to Snakes and Ladders. Though they are based on pure chance, the team say the games are culturally important and can shed light on the development of other strategy games in society.

The team model games as mathematical entities that lend themselves to computational study. This is based on the idea that games are composed of units of information called ludemes, such as a throw of the dice or the distinctively skewed shape of a knight’s move in chess.

Ludemes are equivalent to genes in living things or memes as elements of cultural inheritance. They can be transmitted from one game to another, or they may die, never to be seen again. But a key is that they can be combined into bigger edifices that form games themselves.

The modeling process attempts to capture the forms in which ludemes combine. For example, the team’s approach is to create a “tree” of ludemes for each game they study. This allows the game play to be recorded and compared in an objective way.

That should be an important advance. A significant problem for those who study ancient games is that the rules aren’t always known or clear. That can lead to strange game play.

A good example is the Nordic game Hnefatafl, say Browne and co. “For many years the ‘definitive’ rule set for Hnefatafl was biased to be strongly in favour of the king’s side due to a translation error,” they say. When the error was spotted and corrected, the game became much more symmetric.

So this kind of asymmetry in game play is a clue that all may not be right with the current understanding of the game. It suggests that the evidence should be checked for translation errors and other problems. But of course, the game may have been played in this asymmetric way.

This approach can also pick up other problems. For example, games are less plausible when they drag on when the outcome is clear, or when they frequently end in a draw or are too long or too short. Any games that fall into these categories can be studied in more detail

The other advantage of the ludeme approach is that it allows researchers to think about the evolution of games in a similar way to the evolution of living things. For example, the best aspects of one game can be combined with the best parts of another to form a new game. This is like sexual recombination. Or a rule can be mistranslated, which is equivalent to a point mutation in genetics.  

The process of dividing games into ludemes is equivalent to gene sequencing. It allows researchers to study families of games in a different way and work out links between them. They can even draw up hypothetical family trees to suggest how games might have evolved from each other.

Of course, it is likely that certain rules or ludemes are so obvious that many games will have evolved independently with similar rules. That will be part of the challenge for this emerging discipline.

This is where the techniques of machine learning, data mining, and AI are all set to make important advances. It even raises the possibility of using evolutionary algorithms to invent new games.

This would start with randomly selected sequences of ludemes that the system tests for game play. Of course, most of these would be nonsensical. But by chance a few of them might contain interesting game features.

These successful games are then preferentially reproduced in a new generation of ludeme sequences but also changed via point mutations and sexual recombination. The best from this generation is selected for reproduction, and so on. The types of games that emerge from such a system might be interesting.

So this field looks well set to develop well. That’s thanks in part to this paper, which reviews a conference set up to kick-start the field. New fields of study are not so uncommon, but their evolution is always something to watch with interest. Alfonso X would surely be fascinated.

Ref: : Foundations of Digital Archæoludology

Deep Dive


It’s time to retire the term “user”

The proliferation of AI means we need a new word.

How ASML took over the chipmaking chessboard

MIT Technology Review sat down with outgoing CTO Martin van den Brink to talk about the company’s rise to dominance and the life and death of Moore’s Law.


How Wi-Fi sensing became usable tech

After a decade of obscurity, the technology is being used to track people’s movements.

Why it’s so hard for China’s chip industry to become self-sufficient

Chip companies from the US and China are developing new materials to reduce reliance on a Japanese monopoly. It won’t be easy.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.