Skip to content

Latest commit

 

History

History
84 lines (63 loc) · 2.98 KB

Insights.md

File metadata and controls

84 lines (63 loc) · 2.98 KB
Raw

🤖 AI insights about The Royal Game of Ur

We've used our AI agents to develop some insights about The Royal Game of Ur, which are listed below. If you have any other metrics you'd like to see measured about the game, open an issue, and we can discuss the best way to measure them!

🍀 How much of the game is luck? 🍀

To quantify this question, we chose three different AI agents to represent an okay player, a good player, and a great player.

Skill Level AI Agent
Okay Player Greedy Agent
Good Player Expectimax Agent with a search depth of 3
Great Player Panda Agent with a search depth of 7 and full search for 2 moves

We then played all of these agents against one another to get an idea of how lucky you would have to be to win against players of different skill levels.

📊 Results

Okay Player vs. Good Player

Agent Win Percentage
Expectimax Depth 3 83.2%
Greedy 16.8%

Good Player vs. Great Player

Agent Win Percentage
Panda Depth 7 77.0%
Expectimax Depth 3 23.0%

Okay Player vs. Great Player

Agent Win Percentage
Panda Depth 7 92.0%
Greedy 8.0%

Conclusions

These results suggest that even if you are a world-class Game of Ur player, there is always a chance that someone with very meagre skills could still win! This is part of what makes The Royal Game of Ur such an approachable and exciting game, with matches often coming down to the wire even with big skill disparities. However, this chance is still very small when the skill gap between players is large.

Run the Luck target yourself

If you'd like to run the above tests yourself, you can run them through the RoyalUrAnalysis CLI using:

java -jar target/CLI.jar Luck

🐇 Does playing first give you an advantage? 🐇

Another question that people may wonder is: Does playing as light or dark give you an advantage?

Well, we ran the numbers by playing two depth-5 panda agents against one another for 20,000 games, and here are the results:

Colour Win Percentage
Light 58.4%
Dark 41.6%

From these results you can see that playing as light gives you a significant advantage when playing The Royal Game of Ur, with a 16.8% higher chance of winning! That said, ending up playing second with dark is not the end of the world, with a still significant 41.6% chance of winning.

Run the First-Move-Advantage target yourself

If you'd like to run the above tests yourself, you can run them through the RoyalUrAnalysis CLI using:

java -jar target/CLI.jar First-Move-Advantage