Caster of Magic — Spell Counter-Graph

CoM counters form a directed graph where each tier’s standard answer ages out and demands a more expensive successor. Same structural pattern as Monopoly’s efficient-frontier dominance map, applied to spell economy.

Links: Master of Magic — Economic Analysis, CoM Anti-Colossus Counter Analysis, Efficient Frontier Trade Theory, BattleValue, The Dominance-Frontier Lens (the general lens this counter-graph instantiates) Project: masterofmagic — data/com/spells.json is the source

Thesis

CoM’s spell book is not a flat list of effects — it’s a dominance graph. The cheap counter to today’s threat is the threat that ages out tomorrow, demanding a more expensive successor. Mapping the graph reveals (a) which research investments are forced, (b) where asymmetric counters create attrition wars, and (c) where the only “counter” is behavioral discipline rather than another spell.

This is the same methodological pattern as the Monopoly Efficient Frontier Trade Theory: identify dominance edges to collapse a brute-force decision space into a navigable structure.

Scope Categories

Counters operate at different scopes, and a complete graph needs nodes for each:

Scope	Mechanism	Examples
Single-target active	Removes one effect on one unit	Dispel Magic, Disjunction
Area active	Removes all effects in an area	Disenchant Area (overland), Dispelling Wave (combat-mass)
Passive pool	Buffer that auto-counters until drained	Counter Magic (70-pt pool, -10 per cast)
Resource denial	Starves the spell pool itself, not the spells	Mana Leak (5 MP/turn drain)
Lock/protect	Increases the cost of removal	Spell Lock (~2× dispel cost)
Behavioral	No spell, just discipline	Don’t-stack vs Dispelling Wave; delay-don’t-counter vs Spell of Mastery

The Warp Wood Frontier (canonical example)

Tier 0:  Bows                                  free with unit
   ↓
Tier 1:  Warp Wood (Chaos C, 5MP/320RP)        strips all ammo
   ↓
Tier 2:  Magic ranged units come online        Warp Wood doesn't apply
   ↓
Tier 3a: Resist Elements (Nature C, 5MP/440RP) +4 def vs magical ranged (cheap partial)
Tier 3b: Magic Immunity (Sorc R, 36MP/5500RP)  100 def+res vs spells & magical ranged
   ↓
Tier 4:  Disenchant / Dispel Magic on the buff back to dispel taxonomy

The cost ratchet is roughly 5-10× per tier. The cheap base counter (Warp Wood, 5 MP) becomes irrelevant in the magic-ranged meta; the late counter (Magic Immunity, 36 MP + 5500 RP) requires a different realm AND deep research investment.

Dispel Taxonomy and Asymmetric Edges

The dispel formulas show that counters are taxes, not bargains — equal-cost = 50% chance:

Spell	Realm	Cost range	Formula	Notes
Dispel Magic	Arcane	10-50 MP	`cost / (orig + cost)`	Single-unit, in-combat
Disenchant Area	Arcane	50-250 MP	same	Overland enchantments
Disjunction	Arcane	375-1875 MP	`cost / (1.5 × orig)`	Globals — worse ratio (asymmetric edge)
Dispelling Wave	Sorc Uncom	25-125 MP	`(0.5×cost) / (orig + 0.5×cost)`	Mass combat dispel — half-power, area

Asymmetric edge example — Spell of Mastery vs Disjunction. SoM costs 15,000 MP. To get 50% Disjunction success: 11,250 MP. The counter is nearly as expensive as the spell. Conclusion: SoM cannot be reliably dispelled — only delayed (chip-attack the caster, attack their cities, pressure them to interrupt). The “counter” is behavioral, not magical.

Spell Lock as a cost-doubler. Mind Storm (35 MP) without Spell Lock: opponent’s expected dispel cost = 35 MP. With Spell Lock (100 MP) on top: opponent must dispel two layers, expected cost ≈ 270 MP. You paid 135, they pay 270 — net 2× ratio, exactly as feedback_spell_lock style discipline predicts.

Counter Magic as Zone Defense

Counter Magic (Sorc Uncom, 50 MP, 1280 RP) creates a 70-point pool. Each enemy spell rolls cost/pool chance to succeed; on failure, pool drops by 10. Over the lifetime of a combat, that’s ~7 enemy spells potentially denied — passive, no targeting required. Different category from active dispel: pays once, denies many. A natural complement to Spell Lock for stacking-based wizards.

Mana Leak as Resource Denial

Mana Leak (Death Common, 20 MP) does not target a spell or an enchantment — it drains the opponent’s capacity to cast at all. 5 MP/turn off caster units, 5 mana crystals/turn from the wizard, 1 ammo/turn off magical-ranged units. Over a 10-turn combat that’s 50+ MP worth of denied casting. Counters the spell pool, not the spells — orthogonal to dispel. Pairs especially well against caster-heavy stacks (Demon Lord, Djinn, Efreet).

Open Chains Worth Mapping (next exercises)

Anti-fantastic chain. Resist Magic / True Sight / Magic Immunity tier; how Holy Word / Banish / Disintegrate each hit different points on the resist curve.
Anti-stack chain. Black Prayer / Mass Invisibility / Wrack on the offense; Heroism / Bless / Holy Armor on the defense; Dispelling Wave as the anti-stack cleanup.
Anti-caster chain. Mana Leak / Counter Magic / Spell Blast / Possession-on-the-mage; the chain that targets the caster instead of the unit.
Global enchantment race. Eternal Night / Tranquility / Time Stop / Spell of Mastery; which globals dominate which others, and where Disjunction’s asymmetric cost forces tempo investment.

Limits and Caveats

The graph is wizard-build-conditional — Life ↔ Death incompatibility removes whole chains; multi-realm wizards open combo paths impossible to a single-realm caster.
Casting-rate constraint (per-combat-turn skill cap) means a chain with N spells takes N turns minimum. The cheapest chain may not be the fastest, and time-to-effect matters as much as total MP.
Stacking rules (do -2 res from Black Prayer and -5 res from Mind Storm sum to -7?) need wiki verification before the graph’s combo-edges are trusted.
The graph is also map-state-conditional — Disenchant Area is decisive when you can reach the buffed enemy stack between turns; useless when geography blocks you.