The Neighbor's Shield

The hoplite's shield covered the left half of its bearer and the right half of the man beside him: the one piece of the panoply that existed for someone else. Shroud's Shield school remembers the idea. Testudo grows stronger for every shield-carrying ally in the wall; Dig In shares its avoidance with the group. And yet the school is the least played thing in Novia. Its six actives were 0.52 percent of all skill casts in the last thirty days, the smallest share of any combat school. Its specialization ranks dead last of all twenty-one adventuring trees: 79 players ever learned it, 24 of them active this year. Testudo has thirteen living users and one of them casts seventy percent of it. The school's one combo was used by exactly one player. Meanwhile nearly every melee character in the game owns a shield. This page is about that gap: the audit of a defense that misfires in its own math, the manual the school never shipped, and thirty-four proposals to give the wall a reason to be manned.

Executive summary

Shield design has a twenty-five-year arc and Ultima Online walked all of it first. In 1997 a shield was flat armor soak, a bigger number on a ladder where the tower was strictly better than the buckler and blocking had no moment. In 2003 Age of Shadows swung to the opposite pole: block became a pure chance to negate, shield type stopped mattering, and a failed roll meant the shield did nothing at all. The rest of the genre spent two decades reconciling those failures. World of Warcraft made block a guaranteed percentage and then made mitigation something you spend for. The Elder Scrolls Online priced blocking in stamina. Guild Wars 2 reduced it to a countable charge. Final Fantasy XIV's own patch notes retreated from chance-block wording to a flat percent. Elden Ring made the block the setup for the counter. The modern consensus is a deterministic baseline, player-controlled spikes, and a block event that produces output. Shroud's shield sits astride the two abandoned extremes at once: a chance roll gating a flat subtraction, with a block event that produces nothing. The reaction system already stamps the time of your last block; not one skill in the game reads it.

Thirty days of live telemetry, the specialization census, the school's complete data surface, a defect audit of the block math, and nine years of forum record put the school's condition in three clusters:

1. A defense that misfires in its own math. The 75 percent block cap is really a floor in the roll's denominator, and the character sheet displays a cap that does not exist. Ranged Defense, the level-80 innate, multiplies block against player bows and crossbows and nothing else: not spells, not monster archers, not dragon spit, and its tooltip is a word-for-word copy of a different skill's. The entire rear half-circle skips block and parry while doubling the attacker's hit chance. An empty off-hand out-blocks every shield item in the game. Bolster Balance resists stuns at double its authored value. Shield Bash's specialization debuff does nothing against most of the game. The top craftable shield wears out three times faster than the starter shield. Seven defects from this audit were filed this week.
2. A block that produces nothing, and gear that cannot buy one. A successful block is a silent subtraction: no counter, no proc, no moment, no rotation consequence. The genre converged on the opposite years ago, and the plumbing is already half-built here: the reaction catalog registers the block timestamps with zero consumers anywhere. Itemization compounds the silence. Two of the school's six trained stats are written by no item, masterwork, or enchantment in the game. Exactly two crafted shields grant any block chance at all; the round, triangle, and Acara shields grant none. The shield enchant pool is empty and the shield masterwork pool is the shallowest in the weapon table. Ultima Online's oldest failure, the interchangeable shield, imported whole.
3. The lonely wall. The school's stated identity is cooperative, and it has no cooperative verb beyond two auras nobody runs. There is no way to guard an ally, the oldest and best idea in shield design. In the community's own "Good Tank Build" thread, nobody recommends the Shield specialization and Testudo is never mentioned; when players asked for a real tank role in 2018, the answer shipped in Heavy Armor. The numbers agree: thirteen Testudo users, seventeen for Dig In, one player carrying the school's only combo, and the top eight casters producing seventy percent of the school's entire volume. This is not a school with an unpopular kit. It is a school with no public.

The deal that runs through all thirty-four proposals: Tier 1 makes the school legible, the block math published, the seven filed defects fixed, the manual corrected, the machine pinned by tests. Tier 2 completes the existing kit, a Deflect with an identity, a Dig In whose ramp actually ramps, a Testudo whose specialization reaches the aura players receive. Tier 3 gives the school its missing verbs: the guarded ally, the guaranteed block, the answered blow, the shield worth choosing. Tier 4 puts the doctrine questions on the owner's desk. First the shield tells the truth; then it learns to answer back.

The funnel, and the audience that never arrived

Nearly every melee character owns a shield; about a hundred characters cast a Shield glyph last month, and seventy-nine accounts in the game's history ever bought the school's specialization. The specialization itself is rated mid-tier in payoff, one of the broader improvement sets in the game. The spec is fine. The school under it is what nobody plays.

Nine cards are pure client and content work, eight ship as changesets through the balance program, and the new verbs price out as buff-registry and data work. Seven cards add new skill identifiers, batched into single shared-enum edits; Arcane Defense spent the first of them. Nothing here requires server combat logic, because there is none: combat is computed on the client.

The school today

Twelve skills on the Combat tab: six actives in two short branches off Deflect, five equipment-gated innates in a diamond, and the specialization. The Bash branch controls, the Dig In branch holds ground, and the innate spine stacks resistance, avoidance, durability, and block power. There is also a thirteenth skill the tree never shows: Shield Smash, the combo of Shield Bash and Earth magic's Stone Fist, hinted at by nothing in any interface. And there is a block formula underneath all of it that has never once been shown to a player.

 [ Deflect ] 17,551 casts ........ (1)
    | L10                | L10
 [ Bolster Balance ]   ( Angles ) ....... (6)
   5,874 (2)            /       \
    | L40    | L40   ( Redir- ) ( Shield  )
 [ Shield  ] [ Dig  ]( ection )( Preserv- ) (8)
 [ Bash   ]  [ In  ]    \      ( ation   )
  12,453 (3)  1,512 (4)  \       /
    | L40      | L40   ( Stand Your )
 [ Charge ] [ Testudo ]( Ground ) ....... (7)
  7,399 (5)  1,204 (4)   | L80      | L80
                    ( Special- ) ( Ranged  )
 hidden combo:      ( ization  ) ( Defense ) (9)
  Bash + Stone Fist
  = [ Shield Smash ]  34 casts, 1 player (10)

  [ ] active glyph   ( ) innate

 Incoming hit
   |
   Blockable flag set?
   (about half the world) ... (11)
   |
   Attacker in your front
   half-circle? ............. (12)
   |
   Roll block chance ........ (13)
   |
   Subtract a flat amount
   from the damage .......... (14)
   |
   Nothing else happens ..... (15)

• 1The school's front door and busiest verb. Two block windows in one cast: a large four-second spike and a sustained lift that outlasts its own thirty-second cooldown, so the buff is effectively permanent. It is also the only Shield active with no locked-slot cooldown discount. Cards S10, S2.
• 2The stance that resists twice. Bolster Balance writes one resistance to four crowd-control channels, and the game adds two of them together: stuns, knockdowns and knockbacks are resisted at double the authored value. Filed as #4570. Card S4.
• 3The shield's one attack. A left-arm strike with a stun and a knockback chance. Its specialization payoff, an avoidance debuff, is clamped to zero against most of the game. Filed as #4571. Cards S5, S27.
• 4The ground-holding branch nobody runs. Dig In: seventeen users in thirty days, and its escalating specialization bonus is a single five-second blip that never escalates. Testudo: thirteen users, one of them seventy percent of the casts, and the specialization widens a carrier volume instead of the aura allies receive. Cards S12, S13.
• 5The gap-closer that arrives empty-handed. Charge deals half a weapon swing that players measured as one point of damage, cannot be blocked or parried, and carries the school's best stun. Card S14.
• 6The resistance spine is physical-only. Angles and Redirection contribute exactly nothing against Fireball, Lightning, or any spell, and no tooltip says so. Card S8.
• 7The one innate that feeds the block number. Stand Your Ground's block power is a multiplier that doubles the shield's stopping power at level 100, and no item, masterwork, or enchantment in the game can raise the stat. Cards S2, S23.
• 8The durability passive on a random trigger. Durability damage picks a random equipped slot on every hit; blocking never costs the shield anything extra, so Shield Preservation pays out on a fraction of hits, unrelated to blocking. And the top craftable shield wears out three times faster than the starter shield, filed as #4572. Card S6.
• 9The innate that watches the wrong door. Ranged Defense multiplies blocked damage only when the attacker's weapon is a player bow or crossbow. Spells and creature ranged attacks bypass it entirely. It also wears Angles' icon and a tooltip copy-pasted from Stand Your Ground. Filed as #4567. Cards S16, S7.
• 10The combo one player knows. Shield Smash: Shield Bash plus Stone Fist, block and earth-attunement stacks scaling with both specializations, a stun players report never lands. Thirty-four casts in thirty days, all by one person. Card S15.
• 11About half the world is blockable. Of every attack in the game's data that can hit you, 54 percent carry the flag that lets a shield respond. No ground effect is blockable. No polearm hit is blockable. Neither are the Shield school's own attacks. Cards S21, S31.
• 12The rear half-circle. Attacks from behind skip block and parry and double their hit chance, and "behind" is a full 180 degrees. The wiki says that is the rule; a 2018 bug report says the opposite happened. The layers disagree. Filed as #4568. Card S29.
• 13The roll with a floor dressed as a cap. Block chance is your block modifier over 25 plus itself, floored at 100. Past a modifier of 75 the real chance keeps climbing while the character sheet clamps the display at 75. Filed as #4566. Card S3.
• 14The subtraction with a strange fallback. Damage stopped scales off the shield item's armor rating, times block power, plus a strength term, capped at 75 percent of the hit. With no off-hand item at all, the formula substitutes a value that exceeds every shield in the game. Filed as #4569. Card S30.
• 15The silence. Blocked hits skip weapon procs, produce no event, and feed nothing. The timestamps exist; nothing reads them. Cards S19, S22, S27.

What a shield is allowed to answer

Every attack carries flags that decide which defenses may respond. Decoded from the full rune census, this is the world a shield lives in today:

The flag map is a patchwork of one-off rulings: R26 let spells be blocked, R90 took Lightning back out, and one two-handed blade ships a polearm's flag set and is silently unblockable. Card S31 asks the owner to decide the table on purpose; card S21 is the capstone that widens it for the player who commits to the school.

The pipeline every card plugs into

flowchart TD
  A["Incoming effect"] --> B{"Blockable flag?
54% of the attack world"}
  B -- no --> Z["No block roll ever happens"]
  B -- yes --> C{"Attacker in the front
half-circle, target upright?"}
  C -- no --> Y["Block and parry skipped
hit chance doubled"]
  C -- yes --> D{"Roll: modifier over
25 + max(75, modifier)"}
  D -- fail --> X["Full damage; the shield
contributed nothing"]
  D -- block --> E["Subtract flat amount:
shield rating x block power x 5
+ strength share, cap 75% of hit"]
  E --> F["Bows and crossbows only:
times ranged block power"]
  F --> G["Timestamp stamped.
Read by: nothing"]

The school's mechanics live almost entirely in data and one function of client combat math. There is no server-side combat: the server holds identifiers and never learns a player's position, which is why the front-arc test is client-authoritative and why every proposal on this page resolves as a flag, a formula term, a buff, a charge counter, or a timestamp, never as geometry.

What the canon knows

Ultima Online is the ancestor, and it walked both dead ends first. At launch a shield was flat armor rating on a pure ladder, buckler 7 to Chaos shield 32, bigger strictly better, blocking indistinguishable from wearing more armor. Age of Shadows in 2003 abolished the ladder and made Parrying a chance to negate the hit entirely, about thirty percent at grandmaster, and created the opposite failure the same day: shield type stopped mattering at all. The player guides are blunt about it: which shield you carry makes no difference. Twenty years later the live formula is a sprawl in which training Bushido actively subtracts from your shield block, an anti-synergy players still file complaints about. The two lessons this page inherits whole: a shield must be more than armor, and shields must differ from each other, and UO never managed both at once.

Dark Age of Camelot built the deepest shield line the genre has produced, and every piece of it resolves without geometry. Block chance came from shield specialization. Shield size decided how many simultaneous attackers you could block: small one, medium two, tower three, the cleanest answer ever shipped to the interchangeable-shield problem. Slam, at specialization 42, was a nine-second stun entire builds existed to reach. Engage bought thirty percent block against one designated enemy at the price of your entire offense. And Guard, trained at level five, let you stand beside an ally and block the blows aimed at them: the mechanic that made the shield a group role instead of a selfish stat, and the single biggest hole in Shroud's school today. Base block against arrows started at thirty percent, six times the melee base, because a wall of wood is what you raise against the sky going dark.

World of Warcraft ran the full arc on one stat. Block began as a flat value that raid content inflated into irrelevance, blocking two hundred of a twenty-thousand-point hit. Cataclysm made a block a guaranteed thirty percent of the hit and gave Protection warriors a mastery that could critically block for double. Mists of Pandaria completed the pivot to active mitigation: Shield Slam generated the resource and Shield Block spent it, offense funding defense every few seconds, because the designers concluded that a tank's skill should live in choices, not in waiting out a three-minute cooldown. Spell Reflection remains the genre's best statement that shields answer casters. The Elder Scrolls Online made blocking universal and priced it in stamina, a resource conversion rather than a roll, with the One Hand and Shield line tuning the exchange rate; its taunt lives on the shield line, and permablocking became the balance tail it chased for years. Guild Wars 2 reduced the whole question to a boon: Aegis, a charge that blocks the next attack, countable, grantable to allies, deterministic. Final Fantasy XIV gave the paladin a block stat nobody else has, Sheltron to spend a gauge on guaranteed blocks, Cover to take an ally's hits, and Passage of Arms, the party-wide shield wall; by patch 6.3 its notes had quietly rewritten chance-block into a flat percent. New World added the smallest good idea on this page: block an attack, gain a Fortify stack. The block event produces something.

The action lineage is where blocking finally felt good. Dark Souls made the shield's real stat stability, how much stamina each caught blow costs, with per-damage-type negation turning shield choice into a build decision, and a guard break with teeth when the resource ran dry. Elden Ring added the Guard Counter: the block is not the end of the exchange, it is the wind-up for the answer, and two or three of them stagger most bosses. Monster Hunter's lance blocks as a way of attacking, and its Guard Up skill does something very specific to this page's interests: it flips otherwise-unblockable attacks into blockable ones, a designed override of exactly the kind of flag Shroud's combat already runs on. And tabletop supplied the action-economy verdict: D&D's 2024 rules made the shield bash a free rider on the attack you were already making, because a decade of play proved that a small shield attack that competes for the action slot never gets used. Shroud's deck is the same economy, and Shield Bash sits in it today exactly where the bonus-action bash sat.

The imagery the school already owns

The aspis covered the man beside you; the phalanx was a machine built from that overlap, and its collapse began the moment one man's nerve broke the line. The Roman testudo, the school's own late-game verb by name, was the anti-missile formation: a box of shields raised against the arrow storm. The Norse skjaldborg was shallower and looser, individual shields locked at need. In every telling the shield is the communal weapon, the one you carry for the line, and Shroud's Testudo already computes exactly that: its aura grows per shield-carrying ally in the wall. The mechanic is faithful. It has thirteen players. The stories also supply the missing verb this page keeps circling: standing over the fallen, taking the blow meant for another. The school that owns Testudo has no way to guard a single ally, and that is the strangest absence in its kit.

Capability matrix

Cell judgments worth defending: SotA's published-math cell is empty even though the project's own reference documents the formula, because no player-facing surface does. The blocks-spells cell is half full: bolt spells carry the flag, but the anti-ranged innate ignores them and two lightning spells were removed from the table by hand. The stun cell stays half full in the proposed column on purpose: a marquee stun without diminishing returns is a trap, and that decision belongs to the Bludgeon page's owner card. The enemies-with-shields row is the quiet one: fighting a thing teaches it, and nothing in Novia blocks.

The thirty-four proposals

Tier 1 · Read the shield (S1–S9)

The truth wave. Seven defects from this page's audit were filed this week (#4566 through #4572); this tier fixes them, publishes the block math for the first time, corrects the manual, and pins the machine in tests before any later card moves a number.

Tier 2 · The block made right (S10–S17)

The existing kit, its gaps closed: the school's first balance wave since R90. No new skill identifiers; every number ships as a changeset through the balance program against the S9 pins, and nothing any player owns today gets smaller.

Tier 3 · New verbs (S18–S28)

The missing vocabulary. Every new skill here takes a fresh identifier, batched into single shared-enum edits; the ghost Block identifier in the tree's data is spent, claimed by the Polearms page's Haft Guard, and nothing below touches it. Every verb resolves as a flag, a buff, a charge, or a timestamp, because the server knows no positions and never will for this work.

 Guard a party member with
 your shield.

 Blockable attacks against
 them use your block chance
 at 60 percent effect.
 Guarded damage is split:
 they take none, you take a
 reduced share.

 Duration: 20 sec
 Breaks if you are stunned,
 knocked down, or slain.
 ---------------------------
 Requires: Shield
 Focus: 8    Cooldown: 30 sec

Tier 4 · The owner's desk (S29–S33)

Five decisions only the owner can make. Each is written as the question, the evidence, and the options, and several earlier cards state which way their wording bends once the ruling lands.

Tier 5 · Added after the fact (S34)

One card that was not in the audit. It came from the owner's desk after this page was published, and it is here because it shipped and the page should say so.

Considered and rejected

• Reviving the ghost Block identifier as a shield skill. Spent: the Polearms page's Haft Guard claims it, publicly. Every verb here takes a fresh identifier.
• A held active-block stance (hold to mitigate, ESO-style). The held-guard pattern belongs to the Polearms page's Haft Guard, and stance-channel tenancy belongs to the Tactics programs; the deterministic-defense need ships as S20's charges instead.
• Block-triggered card draw. The Focus page's Mental Reflex owns the block-to-draw hook, verbatim fence; S19 and S22 route around it.
• A block charge-pool or meter. Charge-pool mechanics are the Tactics page's Poise card.
• Taunt or aggro work. Bahamut Arc's 2017 taunt thread is real and his credit stands, but aggro redesign is the Tactics page's vanguard card, and the dev answer (Provoking Shout) already lives in Heavy Armor.
• Guard-break and posture-damage systems. The Blades page's Sekiro-lineage cards own posture stacks; a second posture system in the ring would compete, not complete.
• Stun immunity changes or a marquee stun. Both ride the Bludgeon page's owner decisions; S32 is the brief, not the verdict.
• Full spell reflection. A duel-warping swing with a long PvP grievance record everywhere it shipped; S25 keeps the absorb half.
• Directional or three-stance guard postures. No server positions, no action timing, and Age of Conan's allocation interface was ignored by its own players; the salvageable idea, category allocation, lives inside S21's ranks.
• Stillness stacks on Dig In. The Ranged page's Steady Aim owns stillness stacking; S12 repairs the ramp Dig In already has.
• A brace-versus-charge verb. The Polearms page's set-against-the-charge card owns bracing.
• Shield itemization cards. Gem pools, masterwork depth, named shields, wooden crafting lines: all weapons-program ground (#2550), cited from S23, S24 and S33 rather than re-proposed.
• Magic shield spells. Shield of Ice, Chaotic Shield, a Shield of Focus: their schools' pages, their channels; the ally-shield ask already spent on the Moon page.
• Server-verified formations, line-of-sight cover, or a wall that physically stops projectiles. No server positions exist; the line-of-sight gate is a separately parked owner decision; cover remains a damage cut. Said plainly here so nobody reads the shield wall as geometry.
• The polearm blockability inversion. Mobs blocking polearms while players cannot (#1299) is parked, with the whole shields-versus-polearms doctrine, under the Polearms page's desk.

The composed wall

The proof the catalog composes: one fight, three players, every proposed verb doing its job beside the kit that exists. The two named shield-bearers below are real: the last devotee of Testudo and the only living user of Shield Smash, per thirty days of telemetry.

 EBIPIRAFU'S BUFF BAR
 [Deflect 41s] [Deflect Surge 3s] [Aegis: 2 charges]
 [Bulwark x3] [Testudo: 3 shields in the wall, 8s]
 [Counter Ready 6s] [Guarding: Merryn 22s]

 COMBAT LOG
  You cast Deflect. Block bonus 60 for 4 seconds,
    then 20 for 42 seconds.
  Bone Horror hits you. Blocked. Your shield stops
    54 damage. You take 11.
  Your block readies a counter.                    <- S19
  Bulwark rises. Damage resistance +2, 3 stacks.   <- S22
  You use Shield Bash. Countered: 96 damage.
    Bone Horror is stunned.
  You cast Ward of the Aspis on Merryn.
    2 guaranteed blocks, 20 seconds.               <- S20
  Skeletal Archer shoots Merryn. Guarded. You
    block in her place and take 9.                 <- S18
  You cast Testudo. 3 allies bear shields.
    The wall gains 12 damage resistance.           <- S13
  Death Ray strikes Merryn. Aegis consumed.
    The hit is fully blocked. 1 charge remains.
  Bone Horror strikes you from behind.
    Cannot be blocked.                             <- S29
  CHEIRON NYMPHEAN uses Shield Smash.
    The target is stunned.                         <- S15

Worth noticing: the from-behind line stays, because the mockup shows the constraint instead of hiding it, whichever way S29 rules. The guard line names its price, a redirected reduced share, because a guard without a cost is an aura. And the Testudo line finally says the number the formation earns, which today is computed faithfully and shown to no one.

Sequencing

flowchart LR
  A["Wave 1
truth + pins"] --> B["Wave 2
defect changesets
S4 S5 S6"]
  A --> R["Rulings
S29 S30 S31"]
  B --> C["Wave 3
kit repairs
S12-S17"]
  B --> D["Wave 4a
counter loop
S19 S20 S22 S27"]
  D --> E["Wave 4b
guard verbs
S18 S25 S26"]
  R -- "S31 rules the table" --> F["S21 Guard Up"]
  C --> G["S23 S24
with weapons W2"]
  H["Bludgeon stun desk"] -. "prices the stuns" .-> I["S32 brief"]

Appendix

Where every number comes from

Constraints this page honors

Combat is computed on the client; the server holds identifiers and no player positions, so nothing here proposes geometry, server-side facing, or line-of-sight, and the front-arc test is named as client-authoritative wherever it appears. The line-of-sight gate is a separately parked owner decision this page does not touch. The ghost Block identifier is spent by the Polearms page. No shipped number gets smaller; every specialization change is strictly upward at rank 80. All final numbers ship through the combat balance program against pinned tests. New identifiers are appends, batched, mirrored to both trees in the same commit. Reaction-gated skills follow the Riposte precedent and update the pinned gate set; block-triggered draw belongs to the Focus page; stances, aggro, offhand and charge pools to the Tactics programs; stun regimes to the Bludgeon desk; stillness to the Ranged page; itemization to the weapons and enhancement programs; magic shields to the magic schools. Existing tracker issues are cited, never re-filed.

Test surface

New: a ShieldBalanceTests golden suite (asset expression pins for all seven runes, the sixteen-term mastery census, the innate table, the shield archetype table) plus block-math unit cases (roll shape, front-arc skip, empty-hand fallback, bow-only branch, damage ceiling, the resistance double-count) and the first Blockable-flag manifest sweep. Extended: the reaction gate catalog's exact-set test when S19 lands; the held-stance and combo-window suites already covering Dig In and Shield Smash. Every tester-verification task for this school states that a working shield must be equipped, because the tester skill overlay grants levels, not gear.

Sources

Game data and code: the project repository at the date of writing. Telemetry: the live shard's skill-use log. Census: the live database report dump of 2026-08-14. Genre: the published wikis, patch notes and design posts of Ultima Online (launch armor tables, Publish 16/17, the current Parrying formulas), EverQuest (Bash, Shield Block), Dark Age of Camelot (Blocking, Engage, Slam, Guard, shield sizes), World of Warcraft (Shield Slam and Shield Block histories, the Cataclysm block redesign, the active-mitigation essays), The Elder Scrolls Online (block mechanics, One Hand and Shield), Guild Wars 2 (Aegis, shield skills), Final Fantasy XIV (Sheltron, Cover, Passage of Arms, the 6.3 notes), New World (Sword and Shield), Dark Souls and Elden Ring (stability, guarding, Guard Counter), Monster Hunter (Guard, Guard Up, Power Guard), and the 2024 D&D rules revision (Shield Master). Community record: sixteen threads on the official forums, 2014-2023, and the player wiki's shield pages as of R109.

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.