Dr. Cameron Browne, a researcher in the Computational Creativity Group at the Imperial College of London and the creator of Yavalath, was recently honored as the winner of the 2012 Human-Competitive awards, also known as the “Humies”. Each year, the Genetic and Evolutionary Computation Conference (GECCO) hosts the Humies Awards to honor the use of evolutionary and genetic programming (the process of applying the Darwinian principles of natural selection to software programming) to create computer programs that rival those created by humans. Entries range from software debugging systems to tools that control the motions of atomic force microscopes and systems that allow computers to play board games.

Yavalath particularly excited the Humies judges because it demonstrated a case where evolutionary programming allowed a computer system to develop a tool that was not only successful, but also interesting for humans.

“I thought Yavalath was particularly worthy of an award because it is extraordinary that a computational system can find something that is engaging for humans in the way that this game is," said Lee Spector, a professor of computer science at Hampshire College and a Humies judge. “One of the reasons that Yavalath stood out to me is that while designing games may not have much practical significance, it is striking the way that a computer system is able to take on such a human task. The computer system that designed Yavalath is not playing or following rules, but designing a game in a way that is engaging to humans. That stands out to me as a new kind of thing and is very exciting to me.”

According to Browne, Yavalath and another game he created using genetic programming are the first fully computer-generated games to be commercially published.

“Game design is an art as much as a science, it’s a very human craft,” said Browne. “I think the novelty and creativity of having a computer system design a game is what impressed the judges. Game design is a hard problem to model computationally, and it hasn’t been done successfully before. It’s a very subjective thing.”

Yavalath is a simple mind game similar to tic-tac-toe that seems to engage players because of an interesting twist: Players win when they are able to make four marks in a row, but lose if they are only able to make three. This twist is what keeps players coming back, according to Spector, who tested the game.

The game was developed by a new computer system Browne developed called LUDI. The system employed standard genetic programming techniques to develop different games and then test them. Browne looked at a number of board games, and then encoded the rules of the games into LUDI. LUDI then ran these rules in different combinations, crossing over rules from different games to create innovation, and then fine-tuned the results into a game that is challenging and interesting enough for a human to play numerous times.

“The evolutionary approach worked really well in producing this game,” said Browne.

LUDI was also able to make predictions about which games humans might find interesting by playing a game against itself and then measuring and comparing how well the game worked compared to other games. What most surprised Browne about LUDI is not only that it was able to design such a compelling game, but also that it was able to accurately predict if humans would enjoy playing the game.

Browne plans to apply a similar technique to his next project, in which he will explore the concept of computational creativity employing systems similar to LUDI.

According to Spector, Yavalath is an interesting example of the different types of results evolutionary computation can produce.

“The Darwinian idea is very powerful and it can do surprisingly interesting things,” said Spector. “Cameron’s game was rated highly by game enthusiasts, and I also thought that it was fun to play. It’s exciting to me that evolutionary computation, properly employed, can uncover things like this."

For information on past Humies (which are awarded annually), visit the genetic-programming.org(external link) website. For more information on Yavalath, please visit: cameronius.com(external link). To purchase the game, visit nestorgames.com(external link).