Drafting “magic: The Gathering” With High-Dimensional Card Embeddings
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Drafting is a game mode in collectible card games where players build their decks from a restricted pool of cards. Throughout one draft, players are offered a series of selections, from which they must build their deck. Although drafting is a popular game variant in <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Magic: The Gathering</i>, few machine learning models have been developed to learn card selection strategies. We model drafts with a Siamese neural network that is trained on real-world data and predicts human expert selection. Our model learns an embedding space of preferences by comparing cards in the context of a deck. We examine card representations, evaluate our model on a large-scale dataset, and show that our model achieves 45% zero-shot drafting accuracy on cards that are completely unseen in training. This suggests that the model understands general card semantics and is able to evaluate their strength. In addition, we provide an in-depth exploration of the embedding space. We find that card embeddings capture a significant amount of interpretable information, such as the sizes of decks, and the strengths of individual cards. We also find that the preference-conditioned embedding space learns the similarity of cards, which can enable downstream tasks in the future.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it