The Art of Not Being There

Light Armor is the most-worn, least-committed school in the game. Its five castable skills were used by more distinct players in the last thirty days than any weapon school's, and at a third of any weapon school's volume: everyone samples it, nobody mains it. Eighty percent of every character ever created learned Dodge; one in twenty of them kept leveling it. The passives out-invest the actives by more than double, which means players buy this school as a stat package, not a fighting style. Its Specialization converts worse than any of the twenty-one schools in the game, fifty-two grandmasters ever, because it mostly buys longer durations on buffs that already outlast their own cooldowns. Its most-cast skill is a farm-loop tool for a handful of players that the forum calls useless. Its skill-ID block still carries the scars where the stealth limb was amputated into Subterfuge. And this week, unprompted, the community sat down and wrote this page's design brief in two forum threads. The school's fantasy is being missed. The school's problem is being skipped.

Executive summary

The genre has run the light-armor experiment continuously since 1997, and the results have a shape. Ultima Online, this game's direct ancestor, made leather the armor of the dexterity fighter because plate destroyed the stat his damage ran on, then accidentally made leather the mage's armor too, because only leather let mana meditation work; the light fighter's identity drowned in a decade of everyone wearing his suit. EverQuest gave its leather-class monks the genre's founding aggro trick, Feign Death, and watched an exploit become a beloved raid profession. World of Warcraft let random dodge grow until tanks approached untouchable, clamped it with diminishing-returns curves in 2008, then deleted avoidance from gear entirely in 2014, on the stated grounds that random defense makes damage spiky and healers miserable. The Elder Scrolls Online, which runs an actual medium-armor skill line, the same structural object as a Shroud school, converted its signature 15 percent random dodge buff into a flat 25 percent damage reduction against area attacks in 2018, because a single-target attack randomly vanishing felt terrible to the person who aimed it. And Path of Exile kept random-looking evasion honest with an entropy counter that guarantees a 25 percent chance to be hit means exactly one hit in four. The through-line: pure dice-roll avoidance is the genre's most-retrofitted mechanic, and every survivor either capped it, converted it, activated it, or made the dice tell the truth.

Shroud's Light Armor school predates all of those verdicts and inherited none of their fixes, but its actual condition is stranger than neglect. Thirty days of live telemetry and a census of the entire live database sort the school's problems into three clusters:

1. The commitment gap. More distinct players cast Light Armor skills than any weapon school's, and the school still has no mains: 581 casts per caster against a weapon school's thousands, passives invested at double the rate of actives, and a Specialization that converts worst of all twenty-one schools because what it sells is duration nobody needs. The school is priced and shaped as seasoning, and a cloth chest piece alone turns all of it on, so its largest constituency is mages who will never press Dodge.
2. The counterplay gap. The kit has no answer to the things that actually kill a light fighter. A root cuts dodge in half, a stun deletes it, mesmerize has had no practical counter since smelling salts were slowed, and Escape's threat drop fails at its one job when you are alone, because the monster simply picks you again. Meanwhile the school's own defensive stat is itemized backwards: base plate carries dodge-flavored avoidance on the armor itself, base leather carries none.
3. The rot. Two decades of drift: a parenthesis error that quietly made raw Dexterity the main input of the Evasion buff, a stat named Armor Movement Effects that offsets no armor movement penalty anywhere in the game, dodge math that halves a rooted defender's chance and then skips the roll it was halved for, a creature copy of Dodge missing half its duration, two creature skills no monster uses, dead stats, unreachable set-bonus rows, tooltips that have answered a decade of forum questions wrongly or not at all, and zero dedicated tests. Six defect tickets were filed from this page's research before a word of it was written.

Thirty proposals follow, in four tiers.

The school we have

Eight skills: five castable, three passive. The five castables are all self-buffs or utility; the school deals no damage, participates in no combo, and interacts with an enemy exactly once, when Escape edits a threat table. In the skills window it looks like this.

                    DODGE                          (1)
                    (school root, granted at
                     character creation)
        ______________|________________________
       |              |                        |
   EVASION         ESCAPE                 SWIFTNESS    (5)
   (Dodge 10) (3)  (Dodge 10)  (4)        (Dodge 10)
   ____|_____                                  |
  |          |                            ACROBATICS   (6)
FLURRY    SPRINT                          (Swiftness 40)
(Evasion  (Evasion                             |
 40) (7)   40) (8)              LIGHT ARMOR SPECIALIZATION  (9)
                                          (Acrobatics 80)

• 1Dodge. Learned by 95,889 of the 120,340 characters ever created, 80 percent, because starter templates hand it out. 5.2 percent of learners took it past level 40, the worst conversion in the school. It is also the only skill here with no locked-bar cooldown discount: 30 seconds either way, where every sibling halves. The community's verdict on the skill itself: good as it is. Cards S9, S8.
• 3Evasion (internal name Tumble, and the game still carries an "Evasion Power" stat named for a retired ancestor). Deliberately sought: 64 percent of its 12,931 learners passed level 40. Its buff amount looks like the school’s worst formula defect, the Dexterity term escaping its divisor; measured, it is not one, because the evaluator has no operator precedence. Cards S3, S12.
• 4Escape. The school's most-cast skill, 40,384 casts in thirty days, by its fewest hands, 59. One player accounts for 36.6 percent of every Escape cast on the server: it is a farm loop's aggro-drop, and the forum's verdict for everyone else is "currently useless." It already removes breakable roots; the tooltip has never said so. Cards S6, S10, S11, S21.
• 5Swiftness. The school's real product. 12,154 characters at level 40 or higher, more than any castable, for a passive 13.3 percent move and attack speed at level 100. It vanishes silently if your chest piece breaks. Cards S8, S17.
• 6Acrobatics. The school's most-grandmastered skill, 2,521 GMs, and half of what it grants is a mystery stat: Armor Movement Effects offsets no armor penalty anywhere in the game, because no armor in the game has a movement penalty. Its entire function is making Sprint faster. Cards S16, S8.
• 7Flurry. The volume cellar, 18,470 casts in thirty days, and the school's priciest cast. Its specialization rider speeds up card draw, which is worth nothing to every locked-bar player. Card S14.
• 8Sprint. The healthiest skill in the school: 158 casters, and the top five of them account for only 29 percent of casts, the broadest organic use of any skill here. Its buff survives zoning and even logout. Only its specialization misfires, buying seconds for a buff that already runs 50 seconds on a 15-second cooldown. Card S13.
• 9Light Armor Specialization. 586 characters ever learned it. 114 reached level 80. 52 grandmastered it. That is the worst specialization conversion of all twenty-one schools, 19 percent, and the explanation is on the label: Heavy Armor's specialization says "at 80, it will double the damage resistance," a concrete number that is true; Light Armor's buys longer durations on buffs that already outlast their cooldowns. Cards S15, S12, S13.

Two more facts about the tree that no tooltip states. First, the school used to be bigger: two skill identifiers inside its contiguous numeric block now belong to Subterfuge's Camouflage and Silent Movement. The stealth limb was carved off deliberately, and this page does not try to regrow it; that territory is the Subterfuge page's. Second, the whole school keys off one gate: you are "wearing light armor" if and only if your chest piece is light, and cloth counts. A mage in a cloth robe satisfies every gate in this school without ever slotting a skill from it. The other four armor slots only feed a set-count bonus that tops out at 25 percent with four matching pieces.

How a swing actually resolves

Every attack in the game carries flags naming which defenses may answer it: block, parry, dodge. If the attack is dodgeable and the attacker is not behind you and you are not stunned, your dodge chance is your dodge stat plus half your Dexterity minus a quarter of the attacker's. Then everything is settled by one roll: dodge, glancing blow, normal hit and critical are bands on a single line, and the roll walks them in order. Dodge here is proportional, not an additive percent chance, which means it naturally self-dilutes as the attacker's own hit chance grows. That is the school's quiet, built-in answer to "is dodge capped": no cap, but the bigger the attacker's number, the smaller your band's share.

flowchart TD
  ATK["Incoming attack: flag bits say
  what may answer it
  (block / parry / dodge)"] --> PRE["Block and parry
  roll separately, first"]
  PRE --> DC["Dodge chance =
  dodge stat + your Dex/2
  minus attacker Dex/4"]
  DC --> GATE{"Attacker behind you,
  or you stunned?"}
  GATE -- "yes" --> ZERO["dodge band removed"]
  GATE -- "no" --> ROOT{"Rooted?"}
  ROOT -- "yes" --> HALF["chance halved, then the
  band is skipped anyway
  (defect: the halved number
  still widens the roll)"]
  ROOT -- "no" --> SUM
  ZERO --> SUM
  HALF --> SUM["one roll line:
  dodge + glancing + hit + crit"]
  SUM --> ROLL["single random roll
  walks the bands in order"]
  ROLL --> R1["Dodged: zero damage"]
  ROLL --> R2["Glancing: 1/3 damage
  (from Damage Avoidance:
  the Evasion buff + gear)"]
  ROLL --> R3["Normal hit"]
  ROLL --> R4["Critical
  (divided by Evasion's
  crit-defense rider)"]
  R2 --> DR["then damage resistance
  subtracts, physical only;
  spells never touch it"]
  R3 --> DR
  R4 --> DR
  DR --> OUT["final damage"]

Note the small print that stayed wrong for years: a glancing blow did 25 percent damage in the code while the player-facing text said "one third," until the constant was raised to match the tooltip in August 2026. And the rooted branch was a genuine defect this page filed: rooted halved your dodge chance, but the dodge band was skipped entirely while rooted, so the halved number did nothing except pad the roll and dilute the attacker's crit share. Both halves are closed now, and a root takes your dodge away outright. Card S4.

The gate: one chest piece

flowchart LR
  TORSO["Chest slot"] --> Q{"Light or universal
  chest, unbroken?"}
  Q -- "no" --> OFF["wearing light armor = NO
  every buff amount multiplies to zero
  Swiftness and Acrobatics switch off"]
  Q -- "yes" --> ON["wearing light armor = YES"]
  PIECES["Head, legs, feet, hands:
  unbroken light pieces"] --> CNT["count 0 to 4
  (chest not counted)"]
  CNT --> TBL["set bonus: 1 piece +4%,
  2 +8%, 3 +15%, 4 +25%
  (table rows for 5 and 6
  pieces are unreachable)"]
  ON --> LAB["light armor bonus =
  1 + set bonus + artifact set stat"]
  TBL --> LAB
  LAB --> USE["multiplies Dodge and Evasion
  amounts and Sprint's duration;
  divides Escape's cooldown"]

Three consequences. A cloth robe turns the school on, which is the free-rider question card S28 puts to the owner. A broken chest piece silently deletes 13.3 percent move speed, 13.3 percent attack speed and four points of dodge, with no message of any kind, card S17. And Escape's cooldown is the only one in the school that gets better the more of the set you wear, the school's best-kept secret, which card S8 finally prints on the label.

What the numbers say

Thirty days of live skill telemetry, ending the day this page was written, plus a census of all 120,340 character sheets ever created. Passives never appear in the cast log by design; their story is told by the census alone. One line frames everything: the school's five castables were used by 286 distinct players, more than any weapon school (Blades: 162), for a total of 166,083 casts, less than half of Blades' volume. That is 581 casts per caster against a weapon school's roughly 2,840. Everyone samples the school. Almost nobody lives in it.

[Escape]     40,384 casts    59 casters   (1)
[Sprint]     36,629 casts   158 casters   (2)
[Evasion]    35,549 casts    89 casters
[Dodge]      35,051 casts   153 casters
[Flurry]     18,470 casts    69 casters   (3)

school total ... 166,083 casts, 286 casters
                 1.87% of all skill use

Escape    top 1: 36.6%   top 5: 70.8%   (1)
Evasion   top 1: 24.0%   top 5: 53.9%
Dodge     top 1: 18.8%   top 5: 37.4%
Flurry    top 1: 13.8%   top 5: 37.2%
Sprint    top 1:  7.3%   top 5: 29.4%   (2)

healthy skills spread wide;
farm loops spike narrow

• 1Escape is a farm loop wearing a skill's name. Most casts in the school, fewest hands, and a single player is over a third of all of them. For everyone else the forum's verdict stands: the aggro drop fails at its one job when you are alone, because the monster simply retargets you. Compare the mirror image on the other side of the armory: Heavy Armor's aggro-gaining tools, Taunt and Provoking Shout, saw 15 and 12 casters respectively. Threat tools on both sides of the game are broken quiet.
• 2Sprint is what a working skill looks like. Broadest caster base in the school, lowest concentration, cast by farmers and fighters and everyone between. The kit around it should envy its shape.
• 3Flurry is priced out of its own school. The most focus, the longest lock-in, for an attack-speed buff whose specialization rider pays only random-deck players.

The census: who actually invests

Read the two passive rows again. Swiftness at 12,154 and Acrobatics at 9,161 dwarf every castable in the school; Acrobatics converts 92 percent of the people who touch it and is the school's most-grandmastered skill. The school's actual product, as purchased by the player base, is a passive speed-and-dodge stat package that runs while they play some other school's buttons. The buttons themselves, and the Specialization meant to crown them, go unbought. Sibling weapon schools run 200 to 500 specialists at level 80; Light Armor has 114, and 52 grandmasters, ever, out of a hundred and twenty thousand characters.

The community's brief

On August 12th and 13th, the week of this page's data pull, the official forum produced two threads that together read like a commissioned design review: "Light Armor Ideas" and "armor trees - light/medium/heavy." The asks were specific, unanimous where it matters, and in one case asked for a feature the code already half-contains. This page credits them by name on the cards; here is the ledger, every ask mapped to a card or an explicit answer.

The historical record deepens the picture. In 2016 a thread asked what Acrobatics does and the accepted answer of its day was wrong. In 2016 a thread asked whether Swiftness affects cast speed and never got an authoritative answer. In 2017, when avoidance stacking got its nerf, one player observed that "heavy armor currently has more avoidance while light has resistance... Odd!", an inversion in the base itemization that is still true today. In 2019, light armor plus a bugged artifact set briefly made archers the terror of PvP, and the school has carried that reputation tax since. And through all of it, casters have worn cloth, which counts as light armor, collected the school's gate for free, and skipped its skills entirely, because, as one put it flatly in 2016, there is "no reason to wear leather as a pure mage."

Lessons: Ultima Online to Path of Exile

Twenty-five years of the genre repeatedly discovering the same truths about avoidance, mobility, and the suit you can move in.

And Shroud's own lane: 2015, dodge learns to avoid magic and Evasion becomes damage avoidance. 2017, the avoidance-stack nerf, "avoidance is the #2 mitigation stat in the game." 2017, bone armor ships as the light-armor tank loophole. 2019, stun deletes dodge and root halves it. 2019, the Bandit set bug and the archer scare. August 2026, the community writes this page's brief. The school itself has never had a dedicated pass. This is it.

Capability matrix

What the genre's light-armor systems can do, what Shroud's can do today, and what it could do if this page ships. Full circle: first-class mechanic. Half: partial or indirect. Empty: absent.

Six lessons carried into the catalog. One: entropy accounting (Path of Exile) is the cheapest honesty upgrade in the genre: keep the dodge chance, delete the streaks; it needs no visible change at all. Two: two studios independently converted random dodge buffs into flat reduction against area or unavoidable damage (ESO 2018, WoW's tertiary avoidance); when two rivals converge on the same fix, it is load-bearing. Three: pay heavy, never tax light: UO spent a decade with dead plate before Publish 81 paid heavy armor its own compensations; this page proposes no Heavy Armor nerfs and hands the Heavy page its own dossier. Four: the light suit's genre identity is a bundle, tempo, crit, resource economy, escape artistry, not a single dodge stat; ESO's medium line is the proof it works as an armor school. Five: threat tools made a class raid-essential exactly once in genre history, and it was the leather class; Escape's ceiling is high if threat becomes legible. Six: scripted-encounter design and random defense are natural enemies (FFXIV deleted evasion over it); the cheap insurance is flag discipline, attacks that must not be dodged say so, and the school gets a floor instead of a whiff.

Where light armor fits

Heavy Armor is the wall: it soaks, it taunts, its specialization doubles damage resistance and says so. Cloth is the battery: zero fizzle, focus and intelligence masterworks, the caster's uniform. Light Armor's seat at that table is tempo: you act more often, move faster, recover quicker, and are briefly, deliberately, not where the blow lands. The genre canon for the archetype is the swashbuckler, the fencer who wins by footwork, the skirmisher in the treeline. Shroud already owns every piece of that identity on paper: attack speed in Flurry and Swiftness, movement in Sprint and Swiftness, the sidestep in Dodge and Evasion, the vanishing act in Escape. What it never built is the connective tissue: the kit does not feed itself (a dodge changes nothing about your next action), does not protect its own premise (the mobility school is the easiest to root), and does not pay commitment (the deepest investment, Specialization, buys the least). The catalog below is that tissue. What this school is not, and stays not: a stealth school (Subterfuge's), a third armor tree (ruled out), a tank via the bone-armor loophole (an itemization question for the armor program), and a nerf delivery vehicle aimed at plate (the Heavy page pays heavy in its own coin, card S30).

The thirty proposals

Tier 1 · Read the school (S1–S8)

Truth before change. Every card in this tier makes the school legible: corrected documentation, pinned tests, verified defect fixes, and tooltips that state their numbers. No design changes; six of these began as defect tickets filed from this page's research.

Tier 2 · Kit repairs (S9–S17)

Every existing skill's specific fix, driven by the telemetry verdicts and the community's brief. The specialization cards, S12, S13, S15, share one doctrine: the spec pays in power, not seconds.

Tier 3 · New verbs (S18–S25)

The school learns to fight back. Every new mechanism here reuses plumbing the game already has, the dodge timestamp, the threat table, the innate-row system, the combo data format, and every new skill obeys the identity ruling in S29.

Tier 4 · Doctrine (S26–S30)

Rulings, not code. Everything structural above hangs on one of these five; they cost the owner an opinion each, and they are cheap only if made before the building starts.

Considered and rejected

• A third armor tree (medium armor, robes). Asked by Jack SinAssist this week; already ruled out by the armor program's deep dive, as Cora Cuz'avich immediately noted in the same thread. The energy lands in parity instead: S15, S18, S19.
• Stealth or detection passives. The genre's light-armor bundle includes sneaking, but Subterfuge owns stealth, visibility and detection here, and this school's stealth limb was amputated into it deliberately. Acknowledged in S29, not regrown.
• A literal panache charge pool. The Tactics page fences new charge-pool mechanisms, and the Blades page already declined one for the same reason. S20 delivers the loop with a timestamp instead.
• A dodge-triggered reaction glyph. Reactions ride the Tactics catalog; dodge-triggered card draw is the Focus page's claimed territory. S20 is a normal glyph with a time-window bonus, which is neither.
• Sprint focus-cost changes and the free-run sprint keybind. Both live on the Focus page (its S14) and in movement UI territory respectively; cited, not duplicated.
• Evasion as a timed invulnerability window. The genre's fairest dodge (Monster Hunter, GW2) and the most expensive possible rebuild of this game's buff model; S22 plus S24 buy most of the fairness for a fraction of the cost. Revisit after the engine migration.
• Armor pieces granting deck glyphs. Albion's classless signature and a lovely fit for decks, but items are the armor program's, and ability-bearing gear multiplies the balance surface by the wardrobe. Referenced in the genre section only.
• Weight-ratio gating. Elden Ring's equip-load thresholds would replace the equipment-requirement system game-wide; per-piece counting (S23) reaches the same identity inside existing plumbing.
• Heavy armor nerfs. Including the tempting one, stripping plate's itemized avoidance to fix the 2017 inversion. S30 is the page's promise not to; the inversion is documented for the armor program.
• An obsidian cloth and leather line. Real itemization gap, no obsidian light armor exists, and entirely the armor program's to fill; flagged for its cards.
• A new attacker accuracy stat. The proportional roll self-dilutes dodge, and the dodge-shredding counter already shipped in the Ranged page's territory. S26 records the reasoning; the genre's unbounded-evasion arms race is the warning label.
• Server-side dodge verification. Combat resolves on the client and the server holds no player positions; structurally impossible, and noted in S24's client-side construction.
• Escape as vanish. Stealth is Subterfuge's instrument. Escape stays a threat verb: S10, S21.
• Un-nerfing smelling salts. Barugon's grievance is real, but it is a consumables item question. S10's specialized mesmerize removal answers the need inside the school.

The composed school

The same tree, after the catalog ships. Three new skills at most appear on the chart, the tree stays readable, and every change traces to a card.

                    DODGE
                    (locked-bar discount: S9;
                     undodgeable floor rider: S22)
        ______________|________________________
       |              |                        |
   EVASION         ESCAPE                 SWIFTNESS
   (dex paren      (removes Root, says so;     |
    fixed: S3;      spec removes Mesmerize;    |
    spec adds       solo drop works: S10;  ACROBATICS
    power: S12)     can split a pull: S21) (honest reward: S16)
   ____|_____________                          |
  |          |       |          LIGHT ARMOR SPECIALIZATION
FLURRY    SPRINT  MOMENTUM (new)  (concrete per-piece
(crit     (spec =  (S20: strike     avoidance grant: S15;
 rider:    speed:   pays for        stated in the tooltip: S8)
 S14)      S13)     footwork)
                                UNFETTERED (new innate)
   DEXTERITY BUFF (new)         (S19: root + mesmerize
   (S18: the engine stat)        steadiness in leather)

And the same fight, before and after. Left is today, right is the catalog; nothing on the right is invented, every line traces to a card in this page.

A bandit swings. You dodge.
  (the game records the moment;
   nothing in this school uses it)

You cast Flurry. Your swings
speed up.

A root catches your feet. Your
dodge is cut in half and the
half does nothing.

Three more bandits close in.
You cast Escape. The bandits
pick you again at once.

You run for the treeline at
the same speed as the man
in full plate behind you.

A bandit swings. You dodge.
Momentum is ready.          [S20]

You cast Flurry. Your swings
speed up and bite harder.   [S14]

A root catches your feet.
Your training holds and the
root breaks early.          [S19]

Three more bandits close in.
You cast Escape. Threat
dropped on 3 enemies. They
lose interest long enough
to lose you.           [S10,S11]

You sprint. The full suit of
leather is why you make
the treeline.          [S13,S15]

Sequencing

graph LR
  subgraph P1["Phase 1: read"]
    T1A["S1 book (Built)"]
    T1B["S2 tests"]
    T1C["S8 tooltips + S11 visibility"]
  end
  subgraph P2["Phase 2: defects"]
    D1["S3+S12 Evasion (atomic)"]
    D2["S4 rooted roll"]
    D3["S5 S6 S7 S9 S16 sweep"]
  end
  subgraph P3["Phase 3: kit"]
    K1["S10 Escape rework"]
    K2["S13 S14 S15 S17"]
  end
  subgraph P4["Phase 4: new verbs"]
    N1["S18 S19 S20"]
    N2["S21 pull split"]
    N3["S25 combo"]
  end
  subgraph P5["Phase 5: structurals"]
    X1["S22 floor"]
    X2["S23 gate"]
    X3["S24 entropy"]
  end
  R1["S26 fairness ruling"]
  R2["S27 flag contract"]
  R3["S28 free-rider ruling"]
  R4["S29 identity ruling"]
  TAC["Tactics combo card"]
  T1B --> D1
  T1B --> D2
  T1B --> D3
  D1 --> K2
  D3 --> K1
  K1 --> N2
  R4 -.-> N1
  TAC -.-> N3
  R2 -.-> X1
  R3 -.-> X2
  R1 -.-> X3
  D2 --> X3

Appendix

Data sources

Constraints honored

• Subterfuge owns stealth, visibility, detection: Escape's cards are threat bookkeeping only; the stealth limb stays amputated.
• Tactics owns reactions, stances, combos, the deck: S20 is a plain glyph on a shared timestamp; S25 rides the Tactics combo mechanism; no new charge pool anywhere.
• Focus owns the focus economy: sprint focus cost and dodge-triggered draw cited, not touched.
• The armor program owns items, durability, and the third-tree ruling: this page changes school-side mechanics only and hands the itemized-avoidance inversion and the obsidian gap to its cards.
• The avoidance flag vocabulary is the Polearms page's; this page reuses it (S22, S27) without re-derivation.
• Every number ships in a balance changeset; the S2 goldens make silent drift impossible.
• No Heavy Armor nerfs (S30); the Heavy page is owed its own paying-in-kind pass.
• Client-authoritative combat, no server positions: nothing here requires the server to know where anyone stands.

Test surface

The S2 suite at a glance: ten expression goldens (five amounts, five durations), the two-effect Dodge structure, Escape's radius, threat and cleanse counts with and without specialization, the chest-decides gate with broken-chest transitions, the set-count table including the unreachable rows' fate, five cooldown pairs, the rooted-defender outcome distribution (pins S4), creature-versus-player Dodge parity (pins S5), and, as cards ship: the S12/S13 spec-rider relocations, S15's per-piece grant, S19's resists, S20's window, and S24's entropy distribution. Existing adjacent suites, dodge timestamps, swing-roll sharing, creature mastery reads, stay authoritative for their own ground.

Sources

Forum: "Light Armor Ideas" (Aug 13, 2026), "armor trees - light/medium/heavy" (Aug 12, 2026), "Nerf to Evasion / Glancing Blows / Fortify Defense" (Sep 2017), "Archer OP?" (Nov 2019), "Mage Build - Leather vs Cloth Armor" (Sep 2016), "damage resistance and damage avoidance" (Aug 2024), "Did Flurry get nerfed?" (May 2019), "Acrobatics in the 'light armour tree'..." (May 2016), "Swiftness" (Aug 2016), "Bandit armor questions" (May 2020). Wiki: Light Armor school, Dodge, Escape, Release 21, R65 instruction changes, Bone Armor, Patch notes. Steam: developer reply on avoidance caps (Nov 2014), solo-build threads (2016–2017). Genre: UO armor and Publish 81 documentation, EQ monk and Feign Death references, WoW Warlords itemization changes and avoidance-attribute documentation, GW2 dodge and vigor documentation and the June 2013 balance coverage, ESO medium-armor line and the Murkmire Evasion change, FFXIV 4.0 parry-to-Tenacity notes, BDO evasion-versus-DR guides, Albion armor-ability documentation, Elden Ring equip-load references, Monster Hunter evade-window references, Path of Exile evasion-entropy documentation, Pathfinder 2e swashbuckler rules.

Written from thirty days of live telemetry, a full census of the live database, the shipped skill data and combat code, ten years of forum record, and the genre's twenty-five-year avoidance experiment. Six defect tickets were filed from this research before the page was published. Discussion and votes below; every card has its own thread.

General comments

No account needed. Every proposal above also has its own Discussion box for talking about that one alone; this thread is for the plan as a whole. Comments are plain text, held to basic decency, and may be trimmed.