Rog-O-Matic, Game Playing Program - Computerphile

TL;DR

A group of individuals at Carnegie Mellon University developed a program, Rogomatic, to play the game Rogue using genetic learning algorithms.

Transcript

last time we were playing um a game on a terminal and uh you sort of teased a little bit about what was to come next so where did we go with this yeah we're playing the game of rogue which was one of these early terminal based games but had this terminal addressing so it could jump around the screen and draw you know a nice map and so on but i'd so... Read More

Key Insights

• 👾 Rogue, an early terminal-based game, inspired the development of Rogomatic at Carnegie Mellon University.
• 🎮 The Rogomatic program utilized genetic learning algorithms to make informed gameplay decisions.
• 👶 The program underwent several versions and improvements, including the addition of new genes to optimize gameplay.
• ❓ Implementing Lua and switching to C++ enhanced the adaptability and development of Rogomatic.
• 👾 Updates to the game Rogue posed challenges to maintaining the effectiveness of the Rogomatic program.
• 🏃 Rogomatic had a recording system and different run modes to suit varying needs.
• 📽️ The developers of Rogomatic received partial sponsorship for their project under the ARPA project.

Questions & Answers

Q: How did the Rogomatic program interpret and understand the game's codes?

The developers intercepted the codes sent to the screen and created their own map of the game's dungeon, allowing them to understand the game's mechanics and make informed decisions.

Q: How did the genetic learning component of Rogomatic work?

The program had genes for different gameplay decisions, such as aggression towards monsters, exploration time, and level progression. Through multiple iterations and playing the game, the program would determine the best decisions based on the given circumstances.

Q: How did the team at Carnegie Mellon University overcome the game's updates to make Rogomatic obsolete?

The developers of Rogue made subtle changes to the game's text and monsters, which required the Rogomatic program to be hard-coded with specific responses. However, as new versions of Rogue were released, the program became less effective.

Q: How did the implementation of Lua and the switch to C++ benefit Rogomatic?

The inclusion of Lua allowed for quick adaptation to new versions of the game, as it provided a scripting language to control game elements. Switching to C++ enabled the use of new language features and improved overall project development.

Summary & Key Takeaways

• The game Rogue, an early terminal-based game with complex gameplay mechanics, sparked the interest of a group at Carnegie Mellon University.

• The team developed the Rogomatic program, which could interpret and understand the game's codes, create its own map, and make decisions using a genetic learning algorithm.

• Rogomatic underwent several iterations and improvements, including the addition of new genes to optimize gameplay decisions.