Last year I wrote an academic paper formally solving Gary Gygax's Jousting game, as it appears in the original edition of Chainmail: Rules for medieval miniatures (1971; and prior to that in The Domesday Book newsletter).
This paper was presented at a conference of the International Computer Games Association (ICGA), namely, Computers and Games 2022 (CG2022) -- and recently published in the Springer collected lecture notes for the conference. You can see the abstract here:
Collins, D.R. (2023). Solving Chainmail Jousting. In: Browne, C., Kishimoto, A., Schaeffer, J. (eds) Computers and Games. CG 2022. Lecture Notes in Computer Science, vol 13865. Springer, Cham. https://doi.org/10.1007/978-3-031-34017-8_4
Solving the game involves a mix of mathematical game theory principles and computer solving techniques. In brief, it turns out that there are 3 tentpole "best strategies" (Nash equilibria), involving a particular probabilistic selection of a handful of possible moves in the game. And for more sophisticated play, you can mix-and-match these three tentpoles in an infinite number of ways (all having an equivalent expected win payoff). One of the tentpole strategies is extremely simple, and possibly identifiable on inspection of the game matrix by a smart player; the others are more obscure.
Fortunately, the organizers of the conference record all of the presentations and make them publicly available on YouTube. Here's the session where I made my presentation (first in the block): it goes fast, as we only have 15 minutes each to speak. I'll let you watch below to see the three magic best strategies. Thanks to the organizers for giving me a platform to present these results!
(Note: If you're an entrant to the Jousting tournament at the GaryCon game convention, this won't help you win there; the rules context is significantly changed there in terms of the number of rides, point-scoring, end-game, restricted-ride rule, etc.)
I watched the video; nicely explained. Are you able to share the paper? Currently it's behind a paywall.
ReplyDeleteHave you looked at the Garycon jousting winners to see how close to optimal they played?
Thanks! For publishing like this, the copyright gets turned over to the publisher, so unfortunately I can't distribute it publicly. And at GaryCon I'm not sure they distribute people's moves: I'm not even sure (and wildly curious) how they handle the 2nd round where everyone submits pre-selected moves on paper.
DeleteI have no more kidneys to sell for anything from Springer, used them up on social science stuff of interest and now, like so many lay academics without institutional access, I face my imminent demise from ESRD, but the video seems good.
ReplyDeleteThanks! I totally feel that. I do hope true open-access becomes the norm in the future.
DeleteThat's interesting. I'd like to read the paper, so I sent a request to you on ResearchGate. Probably faster than dealing with my uni's library.
ReplyDeleteActually, I'm not on ResearchGate. Feel free to ping me at the email at the start of the video?
DeleteFun stuff! I'm not educated in the various teams and processes you used to analyze the game, but it was pretty clear what your results were and that the single change of allowing Shield Low along with Steady Seat for restricted ride made it far harder to game the system.
ReplyDelete