Euchre is a popular trick-taking card game played a lot in the midwest. I play it with family all the time. Euchre has a small (but still big) state space as far as card games go (24 cards, 4 players, 5 tricks per round).
How would a reinforcement learning agent do at learning this game? From experience, there is a general strategy to play the game (there is usually one correct play at any given time), but this consists of lots of "rules" depending on the situation. I think a rule-based agent would be well, but there would be lots of rules to enumerate. So it would be interesting to see how an RL agent might learn this game.
Let's get the game written first, and maybe put a front end on it.
This looks relevant: https://web.stanford.edu/class/aa228/reports/2020/final165.pdf
The American Hoyle; or, Gentleman's hand-book of games on Library of Congress, apparently considered the holy grail of Euchre strategy: link. The guy named Hoyle predates the existence of Euchre though.
Someone made a JavaScript library to render cards (though it looks like it may not be maintained anymore): https://github.com/richardschneider/cardsJS