Empiric Bans Only — A Legacy Philosophy

Magic: The Gathering · by Nathan Flachaire (a.k.a. AronGomu)  · March 8, 2025

Ban season is approaching. The next scheduled Ban and Restricted (BnR) announcement at the time of this article's publication is March 31, 2025. I want to delve into a possible ban philosophy for the Legacy format that Wizards of the Coast (WotC) could adopt.

Why Ban?

Win Rate

Playing a specific combination of cards leads to a better average win rate, assuming players are of similar skill level. This is problematic if that specific combination becomes the only viable competitive strategy, warping the format around it.

A Tier 0 format, in which a single strategy comprises the overwhelming majority of top 32s and 64s, leads to a lack of diversity and player expression in deck choice. Either play the Tier 0 strategy or a strategy specifically built to beat it. This excludes all other macrotypes, strategies, and cool cards, making gameplay repetitive.

A Tier 0 format is not a problem if remedied swiftly.

Poor Gameplay Experience

A poor gameplay experience occurs when a specific combination of cards leads to an unhealthy pattern of play. Notable examples from diverse formats are:

Not all cards that promote unhealthy gameplay are banned. Eggs-type strategies and undeterministic combos are still available to players. I played a deck with the sole win condition of naturally decking my opponent via Thwart and 3 Mystic Sanctuary — most opponents were willing to deck themselves before conceding.

Unhealthy play patterns are subjective. Data can be informative, but without a format philosophy to contest that data against, it may not be meaningful.

The Sowing Mycospawn Issue

Bans are often subjective. Sowing Mycospawn is a current example of what most players want banned for unhealthy play patterns, even though its win rate wouldn't warrant action. Some believe its existence prevents control from being competitively viable; I agree, though it is challenging to find data to directly support causation.

Defining control is opaque. I've had success playing control strategies, meaning I won games by going long and outcarding them with Up the Beanstalk and Tamiyo, Seasoned Scholar's ultimate. I play four Daze and four Wasteland, but the gameplay isn't as aggressive as a Delver strategy.

Incoherence

Banning based on win rate is justified more often than not. It is odd that concrete data is rarely shown. WotC doesn't typically justify bans solely through data; they also include community sentiment.

WotC banning cards for format health reasons is incoherent without defining what is considered unhealthy.

Nadu, Winged Wisdom, creates an unpredictable combo that dominates most of the gameplay and is legal, but Sensei's Divining Top is banned, supposedly due to tournament logistics. Consistent turn one kills via Oops, All Spells and Dark Ritual, Entomb, Reanimate packages remain untouched, but cards like Dreadhorde Arcanist and Expressive Iteration that incentivize longer games are over the line.

Why are decks that only care about "Force-checking" legal? Why is that an acceptable play pattern? Of course, some players enjoy glass-cannon combo or turbo prison strategies. Others, like me, like the long, grindy interaction of control shells like Miracles.

Which strategies are we trying to protect? Presumably, it should be more challenging to execute wholly degenerate strategies capable of fast goldfishes than strategies dependent on an opponent's actions. Currently, the banlist supports the opposite.

I despise the lack of coherence across Legacy's banlist in the absence of a format philosophy. It's littered with double standards.

Banning Only via Empirical Data

In our digital era, everything can be quantified. Most Legacy tournaments have decklists and win rates. I am advocating for banning cards based solely on objective criteria that cannot be contested on subjective grounds.

Defining Objective Criteria

We have to define the axioms on which the banlist would operate:

For example, in Legacy, some cards would obviously be banned based on representation and win rate — Brainstorm, Ponder, Force of Will — typically comprise half the format. Ancient Tomb and Daze/Wasteland shells each often comprise a third or more of any given meta. All of these cards have the community sentiment of being Legacy format pillars, though WotC has only ever alluded to agreeing with that notion; they've never said so explicitly.

Guardrails for what is considered acceptable for Legacy might look something like:

If the above guardrail can be verified through empirical data, it should be stated as format philosophy.
This data could also help navigate format health parameters to Legacy's banlist philosophy.

Empiric Banning offers total transparency — no committee needed, just an analyst fetching the data into digestible form. No obscurity from opinions or subjectivity.

Debunkings

Below are common arguments against banning cards that warrant action.

Overpowered by Underrepresented

Sometimes, a deck has lower representation than its high win rate would suggest. This is often due to its complexity, how fun it is to play, or cost, without cards that overlap with other playable strategies.

If there is not enough data — in this case, enough pilots to collect sufficient data — we cannot conclude anything. If the deck is as good as it appears based on limited data, competitive players should adopt it, thereby increasing its representation above the threshold at which empirical data would warrant a ban.

You would never ban an average win-rate deck even if its meta representation is high. If everyone decided to play Storm and it became 95% of the format with a 550% win rate against the remaining 5% of the meta without any format-warping cards, that should be okay.

A format-warping card is anything that keeps the deck sufficiently unaffected by the deck-specific hate players adopt. Grief is a good example — it didn't matter what hate was played to combat it since it would be stripped from hand before being relevant.

A ban-worthy problem is when a deck has the most representation and a lopsidedly favorable win rate. It's on WotC to explicitly inform the Legacy community what those thresholds are.

Bannings for Gameplay Health

Health is subjective and should not be valued in ban discussion. What if I don't like playing fair and interactive magic? I want to play a linear combo that ignores my opponent. That should be as okay as the players who want to play win-conditionless control, even though some players might hate that.

Just because some players don't like something is insufficient to warrant a ban.

What if the vast majority of players hate a particular card for health reasons, but it does not cross any threshold that would warrant a ban?

Let's assume the thresholds WotC implements are reasonable. The card is not played that much,h nor do those who do play it win at a high enough clip. If the vast majority of players dislike the card's presence in the format because of the experience it creates, most won't play it, and the problem will resolve itself

You could argue that theoretical turn-one combo decks that kill through interaction 50% of the time would not be banned under Empirical Banning. That currently exists via Force-check decks like MoMono-Rtompy and Oops, All Spells. Solving for matchups that are a small percentage of the meta carries the opportunity cost of preparing for more represented decks.

If 51% of people play a 50% win-rate interactive combo deck, meaning a majority of players prefer this type of gameplay, then my personal involvement in the format would fall off.

Personally, I would ban every turn-one Force-check version of Soturn-one Force-checkrs, Teferi, Time Raveler because I think they promote unhealthy gameplay. But, of course, not everyone agrees with that.

If diversity matters, WotC should explicitly say so, and then quantify what diverse means. How many different archetypes, at what meta percentages, over what time frame, would define a sufficiently diverse format?

The problem always comes back to the thresholds and/or axioms not being established and communicated.