I Simulated 38,612 Countryle Games to Find the Best Strategy
2 days ago
- #Shannon-entropy
- #geography-games
- #Countryle-solver
- The author enjoys geography games like GeoGuessr, Globle, Travle, and Countryle.
- Countryle provides feedback on direction, continent, hemisphere, population, and temperature to help guess the target country.
- The author built a solver for Countryle using five modules, each handling a different feedback type.
- Shannon entropy was used to determine the most informative guesses by measuring how evenly a guess splits possible outcomes.
- Greece was found to be the most informative guess for directional feedback due to its balanced distribution.
- The solver initially used great-circle routes for direction but switched to rhumb lines (loxodromes) for more intuitive results.
- The Mercator projection, used in Countryle, preserves angles but distorts area, especially near the poles.
- Libya was identified as the best starting guess when balancing all feedback types, but other countries like the Republic of the Congo performed better in average solve depth.
- The solver achieves an average of 2.85 guesses to find the target country when all modules are enabled.
- Future improvements could include optimizing module weights and refining population and temperature bucket handling.