pith. sign in

arxiv: 1907.06562 · v1 · pith:QMDXIWY2new · submitted 2019-07-15 · 💻 cs.AI

The Many AI Challenges of Hearthstone

Amy K. Hoover , Julian Togelius , Scott Lee , Fernando de Mesentier Silva This is my paper

Pith reviewed 2026-05-24 21:24 UTC · model grok-4.3

classification 💻 cs.AI
keywords Hearthstonecollectible card gamesAI challengesgame AIvideo gamesplayer modelingprocedural content generation
0
0 comments X

The pith

Hearthstone analysis shows the full range of AI and games challenges through one collectible card game.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper surveys the AI problems that arise in the collectible card game Hearthstone, from winning play to content generation and player modeling. It argues that focusing on a single popular game reveals both familiar AI topics and new variations on them. The authors position this as a way to view the entire AI and Games field through one concrete example rather than abstract lists. Collectible card games have received less attention than board games despite their commercial reach and distinct mechanics. A reader would care because the approach organizes scattered research questions around an accessible, real-world game.

Core claim

Hearthstone presents a varied set of AI challenges that include playing to win, playing in particular styles, generating game content, and modeling players, and an in-depth analysis of this single game allows the field of AI and Games to be viewed through its lens, discovering a few new variations on existing research topics.

What carries the argument

Hearthstone as a case-study lens that surfaces multiple AI challenges and their interactions in one collectible card game.

If this is right

  • Collectible card games become a recognized source of AI problems distinct from classic board games.
  • Partial observability, deck construction, and hidden information receive targeted study within one ruleset.
  • Procedural content generation and player modeling gain concrete testbeds that combine multiple mechanics.
  • Case studies of individual commercial games can serve as organizing devices for the wider research area.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar single-game analyses could be applied to other popular titles to surface overlooked challenge combinations.
  • The method might produce more focused benchmark suites that test several AI capabilities at once.
  • Insights from card-game hidden information could transfer to real-world domains involving incomplete data and strategic choice.

Load-bearing premise

An in-depth analysis of challenges in one game is enough to reveal new variations across the broader AI and games field.

What would settle it

A review of subsequent AI-and-games papers that finds no new research directions traceable to the Hearthstone analysis.

read the original abstract

Games have benchmarked AI methods since the inception of the field, with classic board games such as Chess and Go recently leaving room for video games with related yet different sets of challenges. The set of AI problems associated with video games has in recent decades expanded from simply playing games to win, to playing games in particular styles, generating game content, modeling players etc. Different games pose very different challenges for AI systems, and several different AI challenges can typically be posed by the same game. In this article we analyze the popular collectible card game Hearthstone (Blizzard 2014) and describe a varied set of interesting AI challenges posed by this game. Collectible card games are relatively understudied in the AI community, despite their popularity and the interesting challenges they pose. Analyzing a single game in-depth in the manner we do here allows us to see the entire field of AI and Games through the lens of a single game, discovering a few new variations on existing research topics.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 1 minor

Summary. The paper surveys AI challenges in the collectible card game Hearthstone, mapping established categories (game playing, deckbuilding, opponent modeling, content generation, and player experience) onto Hearthstone-specific mechanics such as stochastic card draws, hidden information, and large combinatorial spaces. It argues that an in-depth analysis of challenges in a single game provides a lens for the broader AI-and-Games field and can surface new variations on existing research topics.

Significance. The survey fills a gap by focusing on an understudied but commercially prominent genre. A structured mapping of known AI problems to Hearthstone mechanics can guide researchers toward concrete testbeds for imperfect-information, stochastic, and multi-agent settings. The methodological observation about single-game analysis is presented as a perspective rather than a tested hypothesis and does not require additional empirical support within the manuscript's scope.

minor comments (1)
  1. Abstract: the phrase 'discovering a few new variations on existing research topics' is not followed by an explicit enumeration or summary of those variations in the conclusion or a dedicated subsection; adding a short list would strengthen the framing without altering the survey character of the work.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive summary, significance assessment, and recommendation to accept the manuscript. We are pleased that the structured mapping of AI problems to Hearthstone mechanics was viewed as a useful contribution to guiding research in imperfect-information and stochastic settings.

Circularity Check

0 steps flagged

No significant circularity

full rationale

This is a survey paper that catalogs known AI-and-games challenge categories (playing, deckbuilding, opponent modeling, etc.) and maps them onto Hearthstone mechanics. It contains no equations, fitted parameters, predictions, or derivations. The abstract's framing statement is a methodological observation rather than an empirical result that must be independently verified; the paper does not invoke self-citations as load-bearing uniqueness theorems or smuggle ansatzes. The derivation chain is therefore self-contained against external benchmarks and receives the default non-finding.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is a survey-style analysis paper with no mathematical content, free parameters, axioms, or invented entities. The central claim rests on domain knowledge of Hearthstone and prior AI/games literature.

pith-pipeline@v0.9.0 · 5701 in / 1076 out tokens · 19814 ms · 2026-05-24T21:24:36.593811+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

62 extracted references · 62 canonical work pages · 5 internal anchors

  1. [1]

    and Srikant, R

    Agrawal, R. and Srikant, R. (1994). Fast algorithms for mining association rules. In Proc. 20th int. conf. very large data bases, VLDB, volume 1215, pages 487– 499

    work page 1994

  2. [2]

    Aponte, M.-V., Levieux, G., and Natkin, S. (2009). Scaling the level of difficulty in single player video games

    work page 2009

  3. [3]

    G., Naddaf, Y., Veness, J., and Bowl- ing, M

    Bellemare, M. G., Naddaf, Y., Veness, J., and Bowl- ing, M. (2013). The arcade learning environment: An evaluation platform for general agents. Journal of Ar- tificial Intelligence Research, 47:253–279

    work page 2013

  4. [4]

    W., Togelius, J., and Hoover, A

    Bhatt, A., Lee, S., de Mesentier Silva, F., Watson, C. W., Togelius, J., and Hoover, A. K. (2018). Ex- ploring the hearthstone deck space. In Proceedings of the 13th International Conference on the Foundations of Digital Games , page 18. ACM

    work page 2018

  5. [5]

    Hearthstone

    Blizzard (2014). Hearthstone

    work page 2014

  6. [6]

    Browne, C. (2011). Yavalath. In Evolutionary Game Design, pages 75–85. Springer

    work page 2011

  7. [7]

    Bursztein, E. (2016). I am a legend: Hacking hearth- stone using statistical learning methods. In CIG, pages 1–8

    work page 2016

  8. [8]

    and Wunsch, D

    Cai, X. and Wunsch, D. C. (2007). Computer go: A grand challenge to ai. In Challenges for Computa- tional Intelligence, pages 443–465. Springer

    work page 2007

  9. [9]

    Coulom, R. (2006). Efficient selectivity and backup operators in monte-carlo tree search. In Interna- tional conference on computers and games , pages 72–

    work page 2006

  10. [10]

    Cully, A., Clune, J., Tarapore, D., and Mouret, J.- B. (2015). Robots that can adapt like animals. Na- ture, 521(7553):503

    work page 2015

  11. [11]

    de Mesentier Silva, F., Canaan, R., Lee, S., To- gelius, J., and Hoover, A. K. (2019). Evolving the hearthstone meta. In IEEE Conference on Games (CoG)

    work page 2019

  12. [12]

    de Mesentier Silva, F., Isaksen, A., Togelius, J., and Nealen, A. (2016). Generating heuristics for novice players. In 2016 IEEE Conference on Computational Intelligence and Games (CIG) , pages 1–8. IEEE

    work page 2016

  13. [13]

    de Mesentier Silva, F., Togelius, J., Lantz, F., and Nealen, A. (2018a). Generating beginner heuris- tics for simple texas hold’em. In Proceedings of the Genetic and Evolutionary Computation Conference , pages 181–188. ACM

    work page

  14. [14]

    de Mesentier Silva, F., Togelius, J., Lantz, F., and Nealen, A. (2018b). Generating novice heuristics for post-flop poker. In 2018 IEEE Conference on Com- putational Intelligence and Games (CIG) , pages 1–8. IEEE

    work page 2018

  15. [15]

    Dockhorn, A., Frick, M., Akkaya, ¨U., and Kruse, R. (2018). Predicting opponent moves for improving hearthstone ai. In International Conference on Infor- mation Processing and Management of Uncertainty in Knowledge-Based Systems, pages 621–632. Springer

    work page 2018

  16. [16]

    Drachen, A., Sifa, R., Bauckhage, C., and Thu- rau, C. (2012). Guns, swords and data: Clustering of player behavior in computer games in the wild. In Computational Intelligence and Games (CIG), 2012 IEEE Conference on, pages 163–170. IEEE. The Many AI Challenges of Hearthstone 11

    work page 2012

  17. [17]

    S., Garfield, R., and Gutschera, K

    Elias, G. S., Garfield, R., and Gutschera, K. R. (2012). Characteristics of games . MIT Press

    work page 2012

  18. [18]

    Ensmenger, N. (2012). Is chess the drosophila of artificial intelligence? a social history of an algorithm. Social Studies of Science , 42(1):5–30

    work page 2012

  19. [19]

    C., Lee, S., Soros, L

    Fontaine, M. C., Lee, S., Soros, L. B., Silva, F. D. M., Togelius, J., and Hoover, A. K. (2019). Map- ping hearthstone deck spaces through map-elites with sliding boundaries

    work page 2019

  20. [20]

    M., Squillero, G., and Merelo, J

    Garc´ ıa-S´ anchez, P., Tonda, A., Mora, A. M., Squillero, G., and Merelo, J. J. (2018). Automated playtesting in collectible card games using evolu- tionary algorithms: A case study in hearthstone. Knowledge-Based Systems, 153:133–146

    work page 2018

  21. [21]

    Garc´ ıa-S´ anchez, P., Tonda, A., Squillero, G., Mora, A., and Merelo, J. J. (2016). Evolutionary deckbuild- ing in hearthstone. In Computational Intelligence and Games (CIG), 2016 IEEE Conference on , pages 1–8. IEEE

    work page 2016

  22. [22]

    and Simon, H

    Gobet, F. and Simon, H. A. (1996). The roles of recognition processes and look-ahead search in time-constrained expert problem solving: Evidence from grand-master-level chess. Psychological science, 7(1):52–55

    work page 1996

  23. [23]

    G´ oes, L. F. W., Da Silva, A. R., Saffran, J., Amorim, A., Fran¸ ca, C., Zaidan, T., Ol´ ımpio, B. M., Alves, L. R., Morais, H., Luana, S., et al. (2017). Honingstone: Building creative combos with honing theory for a digital card game. IEEE Transactions on Computational Intelligence and AI in Games , 9(2):204–209

    work page 2017

  24. [24]

    C., Khalifa, A., Barros, G

    Green, M. C., Khalifa, A., Barros, G. A., Machado, T., Nealen, A., and Togelius, J. (2018). Atdelfi: automatically designing legible, full instructions for games. In Proceedings of the 13th International Con- ference on the Foundations of Digital Games, page 17. ACM

    work page 2018

  25. [25]

    Guzdial, M., Liao, N., and Riedl, M. (2018). Co- creative level design via machine learning. arXiv preprint arXiv:1809.09420

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  26. [26]

    Hodge, V., Sephton, N., Devlin, S., Cowling, P., Goumagias, N., Shao, J., Purvis, K., Cabras, I., Fer- nandes, K., and Li, F. (2019). How the business model of customisable card games influences player engage- ment. IEEE Transactions on Games

    work page 2019

  27. [27]

    C., Liapis, A., and To- gelius, J

    Holmgard, C., Green, M. C., Liapis, A., and To- gelius, J. (2018). Automated playtesting with proce- dural personas with evolved heuristics. IEEE Trans- actions on Games

    work page 2018

  28. [28]

    Holmg˚ ard, C., Liapis, A., Togelius, J., and Yan- nakakis, G. N. (2014). Evolving personas for player decision modeling. In Computational Intelligence and Games (CIG), 2014 IEEE Conference on , pages 1–8. IEEE

    work page 2014

  29. [29]

    Jakubik, J. (2018). A neural network approach to hearthstone win rate prediction. In 2018 Feder- ated Conference on Computer Science and Informa- tion Systems (FedCSIS), pages 185–188. IEEE

    work page 2018

  30. [30]

    Janusz, A., ´Swiechowski, M., and Tajmajer, T. (2017). Helping ai to play hearthstone: Aaia’17 data mining challenge. arXiv preprint arXiv:1708.00730

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  31. [31]

    Janusz, A., Tajmajer, T., ´Swiechowski, M., Grad, L., Puczniewski, J., and ´Sl¸ ezak, D. (2018). Toward an intelligent hs deck advisor: Lessons learned from AAIA’18 data mining competition. In 2018 Feder- ated Conference on Computer Science and Informa- tion Systems (FedCSIS), pages 189–192. IEEE

    work page 2018

  32. [32]

    Kasparov, G. (2017). Deep Thinking: Where Ma- chine Intelligence Ends and Human Creativity Begins. PublicAffairs

    work page 2017

  33. [33]

    N., and Togelius, J

    Liapis, A., Yannakakis, G. N., and Togelius, J. (2013). Sentient sketchbook: Computer-aided game level authoring. In FDG, pages 213–220

    work page 2013

  34. [34]

    Latent Predictor Networks for Code Generation

    Ling, W., Grefenstette, E., Hermann, K. M., Koˇ cisk` y, T., Senior, A., Wang, F., and Blunsom, P. (2016). Latent predictor networks for code genera- tion. arXiv preprint arXiv:1603.06744

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  35. [35]

    LLC, D. W. D. (2016). Eternal

    work page 2016

  36. [36]

    Lucas, S. M. (2009). Computational intelligence and ai in games: a new ieee transactions. IEEE Transactions on Computational Intelligence and AI in Games, 1(1):1–3

    work page 2009

  37. [37]

    Machado, T., Bravi, I., Wang, Z., Nealen, A., and Togelius, J. (2016). Shopping for game mechanics

    work page 2016

  38. [38]

    Mahlmann, T., Togelius, J., and Yannakakis, G. N. (2012). Evolving card sets towards balancing domin- ion. In Evolutionary computation (CEC), 2012 IEEE congress on, pages 1–8. IEEE

    work page 2012

  39. [39]

    A., Veness, J., Bellemare, M

    Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., Graves, A., Ried- miller, M., Fidjeland, A. K., Ostrovski, G., et al. (2015). Human-level control through deep reinforce- ment learning. Nature, 518(7540):529

    work page 2015

  40. [40]

    Illuminating search spaces by mapping elites

    Mouret, J.-B. and Clune, J. (2015). Illuminat- ing search spaces by mapping elites. arXiv preprint arXiv:1504.04909

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  41. [41]

    of the Coast, W. (1993). Magic: The gathering

    work page 1993

  42. [42]

    Openai five

    OpenAI (2018). Openai five. https://blog. openai.com/openai-five/

    work page 2018

  43. [43]

    SA, A. (2017). Faeria

    work page 2017

  44. [44]

    A., and Melo, F

    Santos, A., Santos, P. A., and Melo, F. S. (2017). Monte carlo tree search experiments in hearthstone. In Computational Intelligence and Games (CIG), 2017 IEEE Conference on , pages 272–279. IEEE

    work page 2017

  45. [45]

    Shaker, N., Shaker, M., and Togelius, J. (2013). Ro- possum: An authoring tool for designing, optimizing 12 Amy K. Hoover et al. and solving cut the rope levels. In AIIDE

    work page 2013

  46. [46]

    Shannon, C. E. (1950). Xxii. programming a com- puter for playing chess. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Sci- ence, 41(314):256–275

    work page 1950

  47. [47]

    J., Guez, A., Sifre, L., Van Den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., et al

    Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., Van Den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., et al. (2016). Mastering the game of go with deep neu- ral networks and tree search. Nature, 529(7587):484

    work page 2016

  48. [48]

    Smith, G., Whitehead, J., and Mateas, M. (2011). Tanagra: Reactive planning and constraint solving for mixed-initiative level design. IEEE Transactions on Computational Intelligence and AI in Games , 3(3):201–215

    work page 2011

  49. [49]

    Spronck, P., Sprinkhuizen-Kuyper, I., and Postma, E. (2004). Difficulty scaling of game ai. In Proceed- ings of the 5th International Conference on Intelligent Games and Simulation (GAME-ON 2004) , pages 33– 37

    work page 2004

  50. [50]

    Stiegler, A., Dahal, K., Maucher, J., and Living- stone, D. (2017). Symbolic reasoning for hearth- stone. IEEE Transactions on Computational Intel- ligence and AI in Games

    work page 2017

  51. [51]

    Stiegler, A., Messerschmidt, C., Maucher, J., and Dahal, K. (2016). Hearthstone deck-construction with a utility system. In Software, Knowledge, Information Management & Applications (SKIMA), 2016 10th In- ternational Conference on, pages 21–28. IEEE

    work page 2016

  52. [52]

    Summerville, A. J. and Mateas, M. (2016). Mysti- cal tutor: A magic: The gathering design assistant via denoising sequence-to-sequence learning. In Twelfth Artificial Intelligence and Interactive Digital Enter- tainment Conference

    work page 2016

  53. [53]

    ´Swiechowski, M., Tajmajer, T., and Janusz, A. (2018). Improving hearthstone ai by combining mcts and supervised learning algorithms. arXiv preprint arXiv:1808.04794

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  54. [54]

    Turing, A. M. (1953). Digital computers applied to games. Faster than Thought

    work page 1953

  55. [55]

    Van Lankveld, G., Spronck, P., and Rauterberg, M. (2008). Difficulty scaling through incongruity. In AI- IDE

    work page 2008

  56. [56]

    Vinyals, O., Babuschkin, I., Chung, J., Mathieu, M., Jaderberg, M., Czarnecki, W. M., Dudzik, A., Huang, A., Georgiev, P., Powell, R., Ewalds, T., Hor- gan, D., Kroiss, M., Danihelka, I., Agapiou, J., Oh, J., Dalibard, V., Choi, D., Sifre, L., Sulsky, Y., Vezh- nevets, S., Molloy, J., Cai, T., Budden, D., Paine, T., Gulcehre, C., Wang, Z., Pfaff, T., Pohle...

    work page 2019

  57. [57]

    Ward, C. D. and Cowling, P. I. (2009). Monte carlo search applied to card selection in magic: The gath- ering. In 2009 IEEE Symposium on Computational Intelligence and Games , pages 9–16. IEEE

    work page 2009

  58. [58]

    N., Liapis, A., and Alexopoulos, C

    Yannakakis, G. N., Liapis, A., and Alexopoulos, C. (2014). Mixed-initiative co-creativity. In FDG

    work page 2014

  59. [59]

    Yannakakis, G. N. and Togelius, J. (2018). Ar- tificial Intelligence and Games . Springer. http: //gameaibook.org

    work page 2018

  60. [60]

    Yee, N. (2006). Motivations for play in online games. CyberPsychology & behavior, 9(6):772–775

    work page 2006

  61. [61]

    and Buro, M

    Zhang, S. and Buro, M. (2017). Improving hearth- stone ai by learning high-level rollout policies and bucketing chance node events. In 2017 IEEE Con- ference on Computational Intelligence and Games (CIG), pages 309–316. IEEE

    work page 2017

  62. [62]

    Zopf, M. (2015). A comparison between the usage of flat and structured game trees for move evalua- tion in hearthstone. Master), Technische Universit¨ at Darmstadt, Darmstadt, Germany

    work page 2015