pith. sign in

arxiv: 2604.18314 · v1 · submitted 2026-04-20 · 📊 stat.ME

Embarrassingly Causal: Causal Use of Associational Data in Magic The Gathering Drafts

Mark Louie F. Ramos , Ph.D This is my paper

Pith reviewed 2026-05-10 03:57 UTC · model grok-4.3

classification 📊 stat.ME
keywords causal inferenceobservational datadirected acyclic graphsMagic the Gatheringdraftsconfoundingselection effects
0
0 comments X

The pith

When the causal link is obvious, observational data alone justifies causal estimates despite confounding.

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

The paper introduces embarrassingly causal scenarios as cases where the exposure-outcome relationship is so uncontroversial that a causal edge can be added to a directed acyclic graph without strong debate. It illustrates this through Magic the Gathering booster drafts, where players routinely rely on large observational datasets to make decisions even though confounding, selection effects, and post-treatment conditioning are present. The central argument is that this obviousness quality is enough to support building causal estimands and using observational data to estimate them. The work also supplies practical guidance for authors, reviewers, and readers on how to evaluate the relevant assumptions in such settings.

Core claim

We argue that the embarrassingly causal quality is a sufficient condition for justifying the construction of causal estimands and the collection of observational data to estimate them, using the case of Magic the Gathering booster draft decisions and gameplay outcomes where purely observational data from 17Lands are widely and effectively used to guide draft choices despite substantial confounding, selection effects, and post treatment conditioning.

What carries the argument

The embarrassingly causal criterion, which treats an exposure-outcome relationship as uncontroversial enough to include its directed edge in a DAG.

If this is right

  • Causal estimands can be constructed and observational data collected to estimate them when the relationship is obviously causal.
  • Purely associational data can guide decisions in domains like game drafts without needing stronger causal assumptions.
  • Reviewers and readers gain concrete criteria for accepting observational causal claims in uncontroversial settings.
  • The same logic applies to other areas where exposure-outcome links are widely accepted as direct.

Where Pith is reading between the lines

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

  • This criterion may reduce the evidentiary bar for causal claims in data-rich recreational domains with strong intuitive mechanisms.
  • It suggests testing whether the same approach holds in other games or sports where large observational records exist.
  • If adopted, it could encourage more papers to frame observational analyses explicitly around obvious causal edges rather than avoiding them.

Load-bearing premise

The existence of an exposure outcome relationship is so uncontroversial that the assumptions needed to include the corresponding causal edge in a DAG can be reasonably made.

What would settle it

A randomized experiment on Magic the Gathering draft choices showing that the associations found in observational data do not match the causal effects obtained under controlled conditions.

read the original abstract

Observational data are often used to answer causal questions, yet the legitimacy of doing so is often argued to hinge on strong, domain supported assumptions about underlying causal structure with limited guidance on how much domain knowledge support should exist to justify including a causal edge of interest in a directed acyclic graph. We introduce the criterion of embarrassingly causal scenarios, where the existence of an exposure outcome relationship is so uncontroversial that the assumptions needed to include the corresponding causal edge in a DAG can be reasonably made. Using the case of Magic The Gathering booster draft decisions and gameplay outcomes, we show how purely observational data from 17Lands are widely and effectively used to guide draft choices despite substantial confounding, selection effects, and post treatment conditioning. We argue that the embarrassingly causal quality is a sufficient condition for justifying the construction of causal estimands and the collection of observational data to estimate them. Correspondingly, we provide guidance on evaluating observational causal inference assumptions for authors, reviewers, and readers.

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

2 major / 2 minor

Summary. The paper introduces the 'embarrassingly causal' criterion, under which domain knowledge renders the existence of an exposure-outcome causal relationship sufficiently uncontroversial to justify including the corresponding directed edge in a DAG. Using Magic: The Gathering booster draft decisions and purely observational 17Lands data as the running example, the manuscript argues that this criterion alone suffices to license construction of causal estimands and their estimation from associational data, even in the presence of substantial confounding, selection, and post-treatment issues. It closes with guidance for authors, reviewers, and readers on evaluating the relevant observational causal inference assumptions.

Significance. If the central claim holds, the criterion supplies a pragmatic, domain-knowledge-based threshold for when observational data can support causal questions without requiring the full suite of strong structural assumptions typically demanded in causal inference. The MTG draft illustration is a concrete, accessible case study that highlights how practitioners already act on such data. The work also earns credit for explicitly acknowledging the confounding and selection problems in the example rather than eliding them.

major comments (2)
  1. [Abstract and §1] Abstract and §1: the claim that the embarrassingly causal quality is a sufficient condition for justifying causal estimands and estimation from observational data is not accompanied by an explicit identification argument. The manuscript notes 'substantial confounding, selection effects, and post-treatment conditioning' in the 17Lands MTG data yet does not show how the uncontroversial exposure-outcome edge renders any target causal contrast identifiable; standard identification (adjustment, front-door, etc.) is not supplied.
  2. [§1] §1 (discussion of the weakest assumption): the assertion that an uncontroversial exposure-outcome relationship licenses inclusion of the edge does not address the remaining identification requirements. The paper correctly flags confounding and selection but leaves the mapping from the observed associational quantities to a causal contrast unaddressed, so the sufficiency claim remains unsupported.
minor comments (2)
  1. The manuscript would benefit from a short formal definition or set of necessary conditions for an 'embarrassingly causal' scenario, perhaps as a boxed definition or short subsection, to make the criterion easier to apply or critique.
  2. Figure or table summarizing the DAG for the MTG draft example (with the uncontroversial edge highlighted) would improve clarity; the current textual description alone makes it harder to see which paths are claimed to be blocked by domain knowledge.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed comments on our manuscript. We respond point-by-point to the major comments below, indicating where we agree that revisions are warranted to strengthen the presentation of the embarrassingly causal criterion.

read point-by-point responses

  1. Referee: [Abstract and §1] Abstract and §1: the claim that the embarrassingly causal quality is a sufficient condition for justifying causal estimands and estimation from observational data is not accompanied by an explicit identification argument. The manuscript notes 'substantial confounding, selection effects, and post-treatment conditioning' in the 17Lands MTG data yet does not show how the uncontroversial exposure-outcome edge renders any target causal contrast identifiable; standard identification (adjustment, front-door, etc.) is not supplied.

    Authors: We appreciate the referee highlighting the lack of an explicit identification argument. The embarrassingly causal criterion is proposed as a domain-knowledge threshold that justifies including the exposure-outcome directed edge in a DAG when the causal relationship is uncontroversial, thereby licensing the construction of causal estimands from observational data. In the MTG example, this is illustrated by the widespread practical use of 17Lands data for draft decisions despite the acknowledged biases. However, we agree that the manuscript does not derive or present a formal identification result showing how the observed associations identify a specific causal contrast under these conditions. We will revise the abstract and §1 to clarify the scope of the criterion, explicitly distinguish edge justification from full identification, and discuss how the uncontroversial edge can be combined with other assumptions (or accepted in practice) to support causal estimation. revision: yes

  2. Referee: [§1] §1 (discussion of the weakest assumption): the assertion that an uncontroversial exposure-outcome relationship licenses inclusion of the edge does not address the remaining identification requirements. The paper correctly flags confounding and selection but leaves the mapping from the observed associational quantities to a causal contrast unaddressed, so the sufficiency claim remains unsupported.

    Authors: We acknowledge that §1 centers on the embarrassingly causal criterion as the key weak assumption enabling edge inclusion, without fully mapping the remaining identification requirements or deriving the associational-to-causal link. The paper's argument is that the uncontroversial nature of the relationship in such scenarios provides sufficient justification for causal estimands even when confounding, selection, and post-treatment issues exist, as demonstrated by MTG practitioners. That said, the referee correctly notes that this leaves the sufficiency claim without a complete identification strategy. We will expand the discussion in §1 to address how the criterion interacts with standard identification approaches and to qualify the limits of what the embarrassingly causal property alone establishes. revision: yes

Circularity Check

0 steps flagged

No circularity: new criterion introduced and applied without reduction to inputs or self-referential logic

full rationale

The paper defines 'embarrassingly causal scenarios' as cases where domain knowledge renders an exposure-outcome relationship uncontroversial enough to include the corresponding edge in a DAG. It then applies this criterion to the MTG booster draft setting using 17Lands observational data, arguing the criterion suffices to justify causal estimands despite confounding. No equations, fitted parameters, or derivations are present that reduce by construction to the inputs. No load-bearing self-citations, uniqueness theorems from prior author work, or renamings of known results appear. The central claim is a methodological argument resting on the strength of external domain assumptions rather than any self-definitional or fitted-input loop, making the derivation self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The paper introduces a new conceptual entity and relies on a domain assumption about uncontroversial relationships to support its central claim.

axioms (1)
  • domain assumption Exposure-outcome relationships can be uncontroversial enough to justify including a causal edge in a DAG with minimal additional assumptions.
    This underpins the definition of embarrassingly causal scenarios.
invented entities (1)
  • embarrassingly causal scenario no independent evidence
    purpose: A category of cases where observational data can legitimately support causal inference due to uncontroversial relationships.
    Newly defined term in the paper to guide causal analysis.

pith-pipeline@v0.9.0 · 5469 in / 1389 out tokens · 56564 ms · 2026-05-10T03:57:49.680142+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

36 extracted references · 36 canonical work pages

  1. [1]

    and SCHÖLKOPF, B

    PETERS, J., JANZING, D. and SCHÖLKOPF, B. (2017). Elements of Causal Inference: Foundations and Learning Algorithms. MIT Press, Cambridge, MA, USA

    work page 2017

  2. [2]

    and DROUIN, A

    BROUILLARD, P ., LACHAPELLE, S., LACOSTE, A., LACOSTE-JULIEN, S. and DROUIN, A. (2020). Differentiable Causal Discovery from Interventional Data. In Advances in Neural Information Processing Systems vol 33 pp 21865–77. Curran Associates, Inc

    work page 2020

  3. [3]

    M., PETERS, J., JANZING, D., ZSCHEISCHLER, J

    MOOIJ, J. M., PETERS, J., JANZING, D., ZSCHEISCHLER, J. and SCHÖLKOPF, B. (2016). Distinguishing Cause from Effect Using Observational Data: Methods and Benchmarks. Journal of Machine Learning Research 17 1–102

    work page 2016

  4. [4]

    PEARL, J. (2014). The Science and Ethics of Causal Modeling. In Handbook of Ethics in Quantitative Methodology. Routledge

    work page 2014

  5. [5]

    HERNAN, M. A. and ROBINS, J. (2024). Causal Inference: What If. Taylor & Francis, Boca Raton

    work page 2024

  6. [6]

    DAWID, A. P . (2010). Beware of the DAG! In Proceedings of Workshop on Causality: Objectives and Assessment at NIPS 2008 Causality: Objectives and Assessment pp 59–

    work page 2010

  7. [7]

    correlation does not equal causation

    STEVENS, C., WITKOW, M. R. and ISBELL, J. (2025). Improving the teaching of “correlation does not equal causation” in Introductory Psychology. Front. Psychol. 16

    work page 2025

  8. [8]

    M., HARRINGTON, M., MICHAL, A

    SEIFERT, C. M., HARRINGTON, M., MICHAL, A. L. and SHAH, P . (2022). Causal theory error in college students’ understanding of science studies. Cogn. Research 7 4

    work page 2022

  9. [9]

    A., WIETEN, S

    HABER, N. A., WIETEN, S. E., ROHRER, J. M., ARAH, O. A., TENNANT, P . W. G., STUART, E. A., MURRAY, E. J., PILLERON, S., LAM, S. T., RIEDERER, E., HOWCUTT, S. J., SIMMONS, A. E., LEYRAT, C., SCHOENEGGER, P ., BOOMAN, A., DUFOUR, M.-S. K., O’DONOGHUE, A. L., BAGLINI, R., DO, S., TAKASHIMA, M. D. L. R., EVANS, T. R., RODRIGUEZ-MOLINA, D., ALSALTI, T. M., DU...

    work page 2022

  10. [10]

    and ZIVICH, P

    FEENEY, T. and ZIVICH, P . N. (2025). Causal inference is hard, and advanced statistical analysis is not enough. BMJ 391 r2618

    work page 2025

  11. [11]

    HILL, A. B. (1965). The Environment and Disease: Association or Causation? Proceedings of the Royal Society of Medicine 58 295–300

    work page 1965

  12. [12]

    IMBENS, G. W. and RUBIN, D. B. (2015). Causal Inference for Statistics, Social, and Biomedical Sciences: An Introduction. Cambridge University Press, Cambridge

    work page 2015

  13. [13]

    and LAMB, L

    ZILIO, F ., PRATES, M. and LAMB, L. (2018). Neural Networks Models for Analyzing Magic: The Gathering Cards. In Neural Information Processing (L. Cheng, A. C. S. Leung and S. Ozawa, ed) pp 227–39. Springer International Publishing, Cham

    work page 2018

  14. [14]

    and IBSEN-JENSEN, R

    CHATTERJEE, K. and IBSEN-JENSEN, R. (2016). The complexity of deciding legality of a single step of magic: The gathering. In vol 285

    work page 2016

  15. [15]

    and HERRICK, A

    CHURCHILL, A., BIDERMAN, S. and HERRICK, A. (2019). Magic: The Gathering is Turing Complete

    work page 2019

  16. [16]

    and HUNG, T

    HAU, R., PLOTKIN, E. and HUNG, T. (2015). Magic: The Gathering Deck Performance Prediction

    work page 2015

  17. [17]

    Play-to- Earn

    RAMOS, M. L. F . (2024). The MTGO Virtual Economy: A Case Study on a “Play-to- Earn” Game That has Existed Since Before Cryptocurrency Was Even Invented. Games and Culture 15554120241273867

    work page 2024

  18. [18]

    SCHMIDT, G. (2023). Magic: The Gathering Becomes a Billion-Dollar Brand for Toymaker Hasbro. The New York Times

    work page 2023

  19. [19]

    Magic World Championship XXIX Viewers Guide.MAGIC: THE GATHERING | ESPORT

    WIZARDS OF THE COAST. Magic World Championship XXIX Viewers Guide.MAGIC: THE GATHERING | ESPORT. Available at http://magic.gg/news/magic-world-championship- xxix-viewers-guide

    work page

  20. [20]

    MTG Booster Draft.MAGIC: THE GATHERING

    WIZARDS OF THE COAST. MTG Booster Draft.MAGIC: THE GATHERING. Available at https://magic.wizards.com/en/formats/booster-draft

    work page

  21. [21]

    The Ultimate Guide to Drafting on MTG Arena - Draftsim.Available at https://draftsim.com/mtg-arena-draft-guide/

    VAN DER VLUGT, N. The Ultimate Guide to Drafting on MTG Arena - Draftsim.Available at https://draftsim.com/mtg-arena-draft-guide/

    work page

  22. [22]

    17Lands.com.Available at https://www.17lands.com/

    17LANDS. 17Lands.com.Available at https://www.17lands.com/

    work page

  23. [23]

    Metrics Definitions.Available at https://www.17lands.com/metrics_definitions

    17LANDS. Metrics Definitions.Available at https://www.17lands.com/metrics_definitions

    work page

  24. [24]

    SIERKOVITZ. (2021). Using Win Rate Data.17Lands Blog. Available at https://blog.17lands.com/posts/using-win-rate-data/

    work page 2021

  25. [25]

    BLACK, S. (2021). Using 17Lands.com As A Resource To Improve At Limited.Star City Games. Available at https://articles.starcitygames.com/magic-the- gathering/select/using-17lands-com-as-a-resource-to-improve-at-limited/

    work page 2021

  26. [26]

    NIKOLIC, A. (2023). How to Get the Most Out of 17Lands for MTG Limited.TCGplayer Content. Available at https://www.tcgplayer.com/content/article/How-to-Get-the- Most-Out-of-17Lands-for-MTG-Limited/a3994007-91ab-4fbb-9fe4-034f96741bf4/

    work page 2023

  27. [27]

    Limited Level-Ups.Available at https://www.youtube.com/@limitedlevel- ups

    NIKOLIC, A. Limited Level-Ups.Available at https://www.youtube.com/@limitedlevel- ups

    work page

  28. [28]

    Sam Black - MTG Wiki.Available at https://mtg.fandom.com/wiki/Sam_Black#Pro_Tour_Results

    MTG WIKI. Sam Black - MTG Wiki.Available at https://mtg.fandom.com/wiki/Sam_Black#Pro_Tour_Results

    work page

  29. [29]

    and YANG, S

    COLNET, B., MAYER, I., CHEN, G., DIENG, A., LI, R., VAROQUAUX, G., VERT, J.-P ., JOSSE, J. and YANG, S. (2024). Causal Inference Methods for Combining Randomized Trials and Observational Studies: A Review. Stat Sci 39 165–91

    work page 2024

  30. [30]

    US FDA. (2023). Considerations for the Use of Real-World Data and Real-World Evidence To Support Regulatory Decision-Making for Drug and Biological Products.Available at https://www.fda.gov/regulatory-information/search-fda- guidance-documents/considerations-use-real-world-data-and-real-world-evidence- support-regulatory-decision-making-drug

    work page 2023

  31. [31]

    HILL, M. A. and ULLRICH, R. L. (2019). Ionizing radiation. In Tumour Site Concordance and Mechanisms of Carcinogenesis IARC Scientific Publications (R. A. Baan, B. W. Stewart and K. Straif, ed). International Agency for Research on Cancer, Lyon (FR)

    work page 2019

  32. [32]

    B., LEURAUD, K., LAURIER, D., GILLIES, M., HAYLOCK, R., KELLY-REIF, K., BERTKE, S., DANIELS, R

    RICHARDSON, D. B., LEURAUD, K., LAURIER, D., GILLIES, M., HAYLOCK, R., KELLY-REIF, K., BERTKE, S., DANIELS, R. D., THIERRY-CHEF, I., MOISSONNIER, M., KESMINIENE, A. and SCHUBAUER-BERIGAN, M. K. (2023). Cancer mortality after low dose exposure to ionising radiation in workers in France, the United Kingdom, and the United States (INWORKS): cohort study

    work page 2023

  33. [33]

    and YANG, H

    GAUDINO, G., XUE, J. and YANG, H. (2020). How asbestos and other fibers cause mesothelioma. Translational Lung Cancer Research 9

    work page 2020

  34. [34]

    A., DONOHUE, T

    OSNA, N. A., DONOHUE, T. M. and KHARBANDA, K. K. (2017). Alcoholic Liver Disease: Pathogenesis and Current Management. Alcohol Res 38 147–61

    work page 2017

  35. [35]

    FREEDMAN, D. (1997). From Association to Causation via Regression. Advances in Applied Mathematics 18 59–110

    work page 1997

  36. [36]

    HO, F . (2025). Regression adjustment for causal inference. BMJ Med 4 e000816

    work page 2025