Dune/docs/ECAZ_OCCUPY_RULES.md

Ecaz Occupy: Basic and Advanced source, and historical conflict

Historical primary-source audit, 2026-09-07, corrected after independent verification. The publisher/designer odd-force conflict remains unresolved as a publisher correction. Under the 8 October Basic-first prototype instruction, Basic now provisionally uses the main p8 paragraph: ceiling contribution/loss and floor survivors, with a visible control warning. This is not a numerical ruling by the user or designer. Advanced retains its separately authorized source. The earlier false attribution to Jack Reda remains withdrawn.

Advanced Ecaz special Karama first version — 8 October 2026

The authorized supplied revision, physical p30, Ecaz G, permits Karama before plans reveal to add the leader-disc difference to the number dialed when Ecaz plays neither weapon nor defense. The native Ecaz combatant now declares this through the existing normalized special-use, real-card cost and once-per-game pipeline. Declare before sealing or after an unarmed own plan, without inspecting the opponent's hidden leader.

Provisional interpretation: absolute printed/copied disc-value difference is virtual dial strength, not new counters, extra support or additional physical casualties. KH/skill modifiers are excluded. Ordinary traitor/explosion precedence remains unchanged. Later plans enforce both empty card slots; existing Voice/inspection/Truthtrance constraints must leave a legal unarmed plan before the cost can commit.

game/ecaz-special-karama.ts supplies the owner-only quote and numeric bonus; authoritative resolution adds the virtual strength independently from physical dial/losses. components/ecaz-special-karama.tsx and minimum legal AI are connected. A direct smoke uses Easy's declaration, rejects an armed plan, changes the winner through the disc difference and observes two actual losses/two spice support. Live isolated React controls submit the exact current event/card in Chromium. Types pass; no broad suite or complete-mode/deployed acceptance follows.

Original live Discovery coalition — 5 October 2026

The original standalone Ecaz Discovery profile marks its native Occupy path, but its former combined-battle gate still rejected the mandatory encounter. The live supported Discovery configuration now feeds original lead selection, owner-labelled fixed/variable armies, counters, payments and aftermath, including optional original Tech/Advanced Strongholds. No battle is skipped, no alliance/plan is synthesized and no extra coalition source is inferred. Unsupported rosters, Skills/Nexus/HW/variants/previews and Basic odd-force rounding remain separate. A fresh Occupy initializer does not gain a Discovery conversion path merely from this live composition.

Meaningful original coalition cases fail before the fix and pass afterward: Ecaz-led support3 pays bank2/own1; allied Guild lead pays its own3 and does not borrow Ecaz's held card. Actual Ecaz5/Guild4 loses3 per owner and leaves2/1. The original genuine six-seat seed20265041 capture after616 accepted actions resumes without changing its saved position, completes557 more/no rejection/ 15 JSON and retains genuine Discovery decisions. Related39, types/lint/build pass. Discovery evidence and limits.

Standalone Nexus / Skills coalition — 5 October 2026

The fresh standalone Ecaz/classic skill-Nexus profile now connects mandatory Advanced coalition battles, including selected original Tech/Strongholds. The chosen faction supplies the plan and its skill; the ally's training is not lent to that plan. The existing selected-lead and owner-labelled army contract remains. Ecaz-led Suk redirects its own fixed casualties after the ally's original variable casualties. Diplomat retreats only the selected faction's undialed own counters: Ecaz uses the fixed Ecaz commitment, while an ally lead uses the variable dial. Cancellation restores the selected lead's ordinary own-pool calculation. These owner-pool compositions are explicit implementation inferences from the card and selected-plan contract, not a publisher combined-Skills ruling. Shared-Duke training and other unresolved effects are not enabled.

Three genuine Advanced captures stopped at the original module guard; each now finishes without rejected actions or skipped battles (918/752/21 more, original scenario seeds; resume does not reconstruct the earlier stream). Meaningful regressions reproduced wrong-army Suk rescue and missing Diplomat retreat before repair, then preserve raw support, separate losses and JSON choices. The coalition regressions retain the separate native source pools and actual rescue/retreat decisions. Human390px L7C8HTQT uses an actual Nexus alliance: Ecaz5/ally4, total dial5/fixed3/variable2, held Arrakeen bank2/own0, trained Sanya4. It wins 9–6, loses ally2, rescues Ecaz3 into one kept/two reserves, then receives original Emperor Axlotl after card cleanup and separate city2 income. Exact standalone source/program evidence. Basic's separate even-force profile and odd-count ruling below are unchanged.

Basic even-force composition — 4 October 2026

Fresh local ecaz-occupy now admits Basic as well as Advanced. Basic uses the exact ecaz Treachery deck with classic or optional Moritani opponents and no optional cards or modules. Genuine setup, original Nexus/worm alliance, shipments, Karama and Collection remain native; older played Basic tables are not retrofitted.

For an even Ecaz force count the publisher page-8 general paragraph and the page-15 example agree: contribution and ordinary-win Ecaz losses are E/2, and E/2 survive. The two pages still contradict each other for odd counts, so an uncanceled odd Basic count is rejected at the original battle choice before a lead frame is saved. It is not silently rounded, canceled or replaced with another fight. Advanced keeps its separately authorized ceiling arithmetic.

Basic has no Emperor/Fremen elite inventory (elites require Advanced or Homeworlds), and the engine normalizes a supplied Basic support amount to zero rather than inventing paid Basic support. The shared editor therefore blocks nonzero support; no additional Basic payment rule is created. Native Karama after the lead choice, selected-lead own-pool cancellation, original winner cards, traitor/mutual/explosion outcomes, joint three-stronghold coexistence and original Mentat victory all use the existing handlers.

Focused source-band cases plus native runtime cases cover E2/E4 under both leads, free full-strength ordinary commitments, typed Basic Fedaykin/Sardaukar strength-one casualties, cancellation under both leads, and the four legal policies. Five genuine Basic two-through-six-seat games were run: four complete with 2,090 accepted actions, no rejections and 54 JSON continuations; the four-seat game is captured at the preserved odd-count conflict and is not rerun or bypassed. Four completed natural games contain no coalition lead choice, so the actual coalition arithmetic evidence comes from the controlled programs and the human path rather than those games.

Authenticated human390px FEJ6SBFB v14 kept the admitted Ecaz/Guild/Emperor seats and native wallets, then staged a conserved even-force board after the genuine Nexus alliance. The table showed the correct Basic labels: a total-dial slider bounded 2..6 (2 fixed Ecaz strength for four Ecaz forces plus Guild's four-force variable pool) with no spice-support control. Real sealed plans — Ecaz Sanya Ecaz 4 with total dial 4 against Emperor Hasimir Fenring 6 — produced an 8–6 Ecaz win. The log and saved state record Ecaz's two mandatory Occupy casualties, Guild's two variable casualties, and the Emperor's eight territory forces to the Tanks; every seat remains at 20 physical forces after refresh.

Authorized Advanced cutover — 2 October 2026

The user authorized the root UNOFFICIAL_Revised_Dune_Rulebook_v.2.3_web.pdf as Advanced source of truth on1 October, including its identified unofficial rulings. Exact revision/SHA and metadata2.3/editorial2.2 are bound by the source amendment. Its physical p32 / logical p30 Occupy B explicitly supplies:

Advanced mandatory Ecaz dial contribution = ceil(E / 2), fully supported
Advanced ordinary-win Ecaz survivors = floor(E / 2)
Advanced ordinary-win Ecaz losses = ceil(E / 2), separately from ally dial losses

Ecaz chooses the lead faction; that faction supplies its actual leader/cards/ plan and spice, while the active variable dial always uses Ecaz's ally's force pool plus the fixed Ecaz contribution, regardless of lead. Physical p19 / logical p17 Ecaz Karama row explicitly changes a prevented Occupy to the SELECTED LEAD'S OWN force pool, with its other ally contributing zero. Keep the selected lead and recompute legality before plans; do not silently inherit the ally's pool when Ecaz leads.

Storm Obstruction (physical p6) and Battle Determination (physical p11) forbid battle if any participating forces are in storm or separated by it. Advanced eligibility therefore checks the whole participating territory armies, including both coexisting allies; it does not invent component-only battle plans or casualties. The canonical board quote retains actual co-side membership when one seat represents the coalition, so an unlocked BG advisor may still voluntarily become a fighter through its native pre-movement opportunity.

This is an authorized Advanced software/source cutover, not a new GF9/ designer erratum or a Basic ruling. Basic's earlier conflict stays separate. The specific user Ecaz-only Duke revival/Ghola decision remains higher-priority than the contrary revival exception printed on that sheet.

Advanced combined-army implementation is Partial / Development Verified: native selection, owner-labelled force/support/loss settlement, printed/BG prevention, human controls and all four minimal legal policies. Ordinary canceled-win non-dialed ally retention and canceled-defeat whole-side destruction compose the base same-side outcome, not a newly located special casualty FAQ. Exotic/optional combinations remain explicit. Source/controls/rule runtime are developed now; comprehensive save/privacy/custody assurance follows all rules under the2 October priority.

Bounded native rule and control proof

The first checkpoint admitted Ecaz/classic/optional Moritani,2..6-seat Advanced, ordinary33 Treachery Cards and the E3 faction setup alone. It does not select the independent three-card variant, other overlays, public expansion starts or an existing-game retrofit.

Seven actual native-program cases exercise E1/E5 free fixed contribution, both plan/payer leads, native Fremen/Fedaykin and Emperor/Sardaukar, and both canceled lead-own branches through original Collection continuation. Exact odd/even, storm-relative order, traitor, mutual-traitor, explosion, printed/ BG Karama and next-turn cases have meaningful rule regressions. A reproduced storm-isolated joint battle now has no offer; clear whole armies retain one offer and their actual co-side membership. The existing voluntary BG advisor offer is preserved, not re-pinned to absence. Final scoped reviews find no defect.

Original CLI/human390px Q5Z2WH8X v22 chose Ecaz lead and total9: mandatory Ecaz3 plus free Fremen/Fedaykin6. Native resolution left Ecaz2, Fremen0, Ecaz spice unchanged5 and original Collection. LZHMY454 v14 used real printed Karama after Ecaz lead, recomputed minimum0/own pool3, sealed dial1/support1, and resolved with Ecaz2/Guild4, one actual Ecaz support debit and original Collection. All four policies actually seal native legal plans for either lead, active or canceled.

The coherent affected rule/guide batch passes 173/173. Current-source five unstaged2..6 games complete with2,123 accepted actions, no rejections and54JSON continuations, sourceUnchanged:true. Full results.json records one actual combined lead choice. Native cases and human rooms provide the explicit arithmetic, casualty and counter-path evidence.

Native E1/E2 force-owner integration — 2 October 2026

The existing profile now admits native Ixians, Tleilaxu and CHOAM with Ecaz, classic opponents and optional Moritani. Distinct selected decks include ecaz and each native ix/choam family:33 ordinary with Ecaz alone,47 with Ix, 35 with CHOAM alone,47 with both. Genuine setup preserves original Ix offers/ HMS, Tleilaxu Face Dancers and CHOAM Auditor. Richese mixed No-Field planning, independent E3 Treachery and optional overlays remain explicitly unfinished.

Physical Ix variable losses lead to the actual Ixian's Suboid substitution decision/counter, including when Ecaz formed the plan. Accept, decline and Karama prevention retain the original selected lead's cards and winner suffix. CHOAM receives only its original eligible paid support shares; mandatory free Ecaz strength mints no support payment, donor share or traitor income. An actual Tleilaxu winning co-side cannot Face Dance itself. An external Face Dancer returns both surviving Ecaz/allied armies, with replacement maximum equal to their combined survivors, while original selected winner rewards/ cards remain. Sources are real reserves/board choices; no fake merged player.

Native34 rule cases, coherent affected191/191, types/lint and19 actual program cases pass, including Ix substitution decline/prevention and zero/full whole co-side replacement. Twenty-three unstaged classic/paired/mixed2..6 games complete with13,004 accepted actions/no rejection and339JSON continuations, sourceUnchanged:true. Full results.json records three actual combined lead choices. Native cases and human controls additionally exercise the typed/ replacement boundaries. Two scoped rule reviews find no defect.

Original CLI/human390px 5B4FW8D3 v12 let physical Ixians sacrifice two Suboids to recover two Cyborgs after an Ecaz-led plan; Ecaz retained its Shield and the real Collection opened with Ecaz2/Ix4 including two Cyborgs. UVGT4DLN v10 exposed and used the full three-force replacement maximum, returned Ecaz1/Guild2 to their own reserves, placed three Tleilaxu forces and retained the original Ecaz Slip Tip and bounty. Both continue the original phase rather than manufacturing another battle or card owner.

Independent cards with combined armies — 2 October 2026

The existing fresh ecaz-occupy CLI accepts explicit --ecaz-treachery, independently selecting all three physical cards before the original setup: 36 Ecaz-only,50 Ix,38 CHOAM,50 both. No default/public start or saved retrofit.

The printed E3p16 Harass FAQ returns only the card user's own undialed forces. An active Ecaz card user returns floor(original E/2) and retains the original ceil(E/2) free/full physical commitment; no ally support is charged to those fixed fighters. A selected ally uses its typed variable commitment after subtracting fixed strength. Canceled Occupy uses the lead's ordinary own pool. Opposing successful Traitor cancels returns; withdrawn reserves are outside later loss, explosion and FaceDance replacement. This is a composition of the own-only FAQ and authorized Advanced arithmetic, not a newly retrieved detailed Ecaz casualty FAQ.

Reinforcements uses the card user's own three normal-first reserve counters and existing +2 score-only modifier, not ally reserves, extra physical dial losses or support. Existing provisional all-revealed-outcome cost/disposal is unchanged and remains explicitly not a publisher ruling. Both slots and opposing holders work. Same-plan Harass/Reinforcements now follows the provisional original-reserve-cost-first policy, preserving actual card-user ownership and original fixed losses. Original or copied Stone now composes with same-plan/opposing Reinforcements under the physical-token contract. Its +2 normal score adds no undialed counters; original reserve cost and fixed versus variable on-board losses stay separate. Other optional modules and Basic/unprofiled co-side card battles remain guarded.

Native/pure67 cases, relevant284-case batch and27 actual native-program cases pass, including all four legal policies. Two natural captured games exposed an existing mandatory-discard callback comparing casualties to the logical winner rather than physical owner. Correct owner and explicit fully committed Harass pool constraints now preserve the selected loss allocation. Both original captures continued to completion; no database reset or new recovery-audit campaign. Obsolete wording/log/default pins were deleted. Final scoped source/native reviews find no defect.

Current-source46 ordinary/variant2..6 games complete26,619 accepted actions/ no rejection/696JSON continuations, source unchanged. Full results.json records9 actual combined lead choices; compact report.json lacks counters. Original CLI/human390px T8T3ZYWH v12 sealed Ecaz Harass total5/support2, returned Ecaz2 reserves, sent original fixed3 to Tanks and left Guild2. WW54BZGM v12 sealed total5/support2 with Reinforcements; its +2 won the real equal-leader comparison, paid three Ecaz reserves, left Ecaz2/Guild2 and sent six total Ecaz counters to Tanks. Both reach original Collection.

Native Richese, No-Field and Stone composition — 3 October 2026

Advanced native tables now admit Richese alongside the already supported families. The actual six-native failure was Ecaz+Ixians versus CHOAM at Carthag with Richese elsewhere, not a Richese opponent or an illegal bot offer. The roster-wide guard and fresh-label-only runtime restriction prevented a mandatory source-clear battle. Native composition now selects the real lead and original force owners; it does not suppress that battle or fall back to an ordinary one-army plan. Optional overlays remain guarded.

The supplied Advanced PDF physical p36, Richese No-Field B says a concealed 0/3/5 marker counts as one force for all effects until reveal, materializes only available reserves, and reveals with Battle Plans. A marker-only Richese ally therefore remains original battle presence even with a zero prospective pool. The selected plan actor receives its private variable pool; rivals do not receive that pool or an Occupy profile containing the secret denomination. The physical Richese owner reveals once after both seals. Dial prescience and whole-plan restrictions follow the actual variable-force owner rather than the original chooser. Preventing Occupy with Ecaz lead makes only Ecaz's own pool inspectable; the unrelated marker stays concealed until original reveal timing. Richese's own ordinary-plus-marker dialing and broader disclosure rulings remain.

Stone Burner compares floor(E/2) + native ally physical undialed tokens for active Occupy, or the selected lead's own pool only after prevention. Its total dial offset is removed before native typed casualty enumeration. Mirror copies use the same comparison and original physical-card retention. Public opposing No-Field preflight enumerates possible native pools, never the actual secret denomination. Clean selected CHOAM/Ix now uses the existing provisional original/copy priority; winner-changing allocations and Ix combined modules retain their guards. Harass/Stone now uses the explicitly provisional post-withdrawal policy: remove only the actual card user's undialed pool, never recompute original fixed commitment and never inspect concealed opposing cards for preflight. Native Ecaz/Moritani Mirror admission is Advanced only; Basic is unchanged.

The affected rule batch passes 363/363, with types/lint and build passing. 26 actual native programs cover ten ordinary/marker owner cases, eight Stone/Mirror active/prevented lead cases and eight four-policy concealed-pool cases, with original Collection and JSON continuation. These are genuine setup followed by explicitly conserved board/card/bank staging, not unstaged games. All four policies remain minimal legal participants; strategy is deferred. The formerly stalled genuine Advanced combined seed20260932 now completes 2,424 accepted actions, no rejection and65 JSON continuations, at turn9.

Original CLI GQ4FEWNW v8 and human T78EFNZK v8 admitted fresh native setups with real authenticated seats. Human390px T78EFNZK v22 selected Ecaz lead and sealed total5/support2 with Sanya Ecaz. Both seals revealed the original No-Field5 into only two available Richese reserves; settlement left Ecaz2, Ecaz Tanks3 and Richese Tanks2. Native CHOAM received its original one-spice support income, and original Collection returned Ecaz's wallet to30. Refreshing retained the actual continuation. No reset, re-deal, public start, deployment or new comprehensive assurance campaign.

Encounter evidence correction — 2 October 2026

Earlier “no combined lead choice” statements were incorrect: the compact report.json deliberately omits used counters. The immutable full results.json records1 choice in the five-game checkpoint and3 in the 23-game checkpoint. Neither historical report is rewritten. The current 46-game original/variant batch records9 actual combined lead choices.

Provenance and attribution correction

Source	Observed wording or correction	Authority and conclusion
GF9 E3 rulebook, p.8	Contribution rounds upward; survivors after a win round downward.	General printed rule, supporting ceil contribution/loss and floor survivors.
Same publisher PDF, indexed p.15	Five contribute two; two die and three remain.	Contradicts the general paragraph. A current publisher URL does not prove every page includes later revisions.
Publisher-authored English mirror, p.15, extracted in /tmp/dune-rules/ecaz-audit.txt:686	Five contribute three, but two still die and three remain.	Contribution agrees with p.8; casualty sentence still conflicts.
Rulebook file submitted by Jack Reda / The Warp	File description mentions numerical corrections.	Supports the existence of a revised file, but the description alone does not identify the casualty correction or establish corrected contents.
Jack Reda, 26 August 2023, on that file page	Explains face-up Ambassador placement and face-down Terror placement.	Does not correct Occupy casualties.
Jeff C / @brigade, 28 November 2023, on that file page	Reports remaining number errors and proposes three of five Ecaz forces die, with two surviving.	Community interpretation, not publisher or designer erratum. It cannot resolve this task's source conflict.

The exact fresh query site:boardgamegeek.com/filepage/263742 "First Q&A" returned the full comment boundaries. An earlier broader result had omitted the intervening Jeff C author/date block, causing the false attribution. Direct BGG opening returned 403. The complete indexed result takes precedence over that incomplete excerpt. Do not cite Jeff C's sentence as a Reda ruling or infer author approval from his hosting the file.

For participating Ecaz force count E, the p.8 reading, not an adjudicated correction, is:

Ecaz contribution = ceil(E / 2), fully supported without spice
Ecaz losses on ordinary allied win = ceil(E / 2)
Ecaz survivors on ordinary allied win = floor(E / 2)

The historical publisher/mirror example still conflicts for E=5. The 8 October first-version implementation uses the p8 reading in Basic, including odd counts, and labels it provisional in the lead-choice and plan controls. Advanced uses the separately user-authorized source cutover. Fixed Ecaz losses remain independent of the ally's variable dial losses. One actual Basic five-counter coalition smoke reaches Collection with three Ecaz casualties and two survivors; broader interaction refinement is deferred.

Occupation, controller and battle order

Ecaz and its current ally form one occupation side and may share a territory. Their actual counters retain separate ownership. This allows Ecaz to enter a stronghold already holding its ally and an enemy; it does not allow a third unrelated occupation side. Storm restrictions still apply. Homeworlds are expressly excluded because they are not territories. Existing discovery locations and the mobile stronghold require their own admission rules; Occupy does not itself create permission to ship directly into the mobile stronghold. E3 pp.8,10

Jack Reda independently confirms the stronghold exception and explains the two meanings of aggressor in Understanding Ecaz Occupy, 16 October 2023: ordinary storm order still determines who selects battles; the faction ultimately fighting matters for tie priority. His answer also confirms that selecting Ecaz does not exchange the roles of the two force pools. BG advisors need not flip and do not participate in Occupy combat while they remain advisors.

Implementation must therefore distinguish:

• Battle-order actor: determined by the normal battle sequence before Ecaz's selection.
• Selected battling faction: Ecaz chooses this after battle order is determined. It supplies its own permitted leader, hand and Battle Plan; it is not permission to browse or play the ally's hand.
• Dialed force owner: always Ecaz's ally while Occupy operates.
• Fixed-contribution owner: always Ecaz, with the printed half-force contribution and separately attributed casualties.

Do not represent these as one interchangeable attacker identifier. In a storm order ally → opponent → Ecaz, selecting Ecaz must not silently inherit the ally's tie priority merely because the ally selected this battle. The precise API should retain the battle-order actor and derive the selected combatant's ordinary tie relation separately. This follows the designer's distinction; no Sapho or Stronghold-card exception is inferred from it.

Battle calculation, support and losses

The ally's chosen dial uses its normal force rules; the Ecaz increment costs no support spice in Advanced play. The E3 p.15 Fremen FAQ confirms Fedaykin can be dialed and Fremen forces retain free full strength in this co-occupied battle. Its answer to the first of two questions is not a numerical one-Fedaykin limit. Preserve physical force types and actual force ownership when computing support and casualty alternatives. E3 pp.8,15

For a normal win, settle the dialed ally's legal losses under Basic or Advanced rules, then the independent Ecaz loss under the explicitly selected rounding rule. For a normal loss, both participating allied force pools lose their forces: retaining Ecaz's fixed survivors is conditional on winning, not a guaranteed retreat. BG advisors remain outside these pools. A lasgun/shield explosion destroys all affected territory forces under its general rule; neither half-rounding nor the unchosen-faction label makes a safe army. Base pp.10–11,13,22

The chosen faction owns the plan and is the ordinary plan-payment actor under base p.13. Apply only established funding permissions, such as CHOAM's battle aid; shared occupation does not merge spice balances or grant arbitrary allied funding. The fixed Ecaz increment is not an amount of spice secretly paid by Ecaz and must not create CHOAM payment income. Preserve normal force-owner properties, including the explicitly confirmed Fremen support exemption. This payment composition is inferred from the ordinary plan rule and Ecaz's explicit selection, rather than a retrieved worked payer example.

General battle effects still need their own scoped application. A traitor's ordinary winner-loss exception, defeated-side destruction, leader deaths, and card ownership must be considered across the actual allied side instead of processing Ecaz as an unrelated third army. Neither the conflicting example nor the community correction is an explicit ruling on every traitor, Face Dancer, Stone Burner, Suboid exchange, Auditor, capture or Stronghold-card combination. These integration checks are separate from the still-open rounding conflict.

E3 p.16 specifically says that Harass & Withdraw returns the card player's undialed forces, while the ally's forces follow their normal treatment. Therefore a side-wide indiscriminate retreat is incorrect. Preserve owner-labelled force pools before adding this card. E3 p.16

Collection, control and victory

Desert collection is a shared allocation: the allies may agree on the split; absent agreement it is as equal as possible with the odd spice going to Ecaz's ally. Determine the actual collectible amount once, respecting force availability and collection rate, then allocate it. Do not pay the entire spice deposit separately to both factions. The p.15 FAQ also confirms that co-occupation in Arrakeen gives both factions ornithopters and collection rate three. E3 pp.8,15

The Advanced Collection advantage is separate: in co-occupied Arrakeen and Carthag each ally receives two bank spice; in co-occupied Tuek's Sietch each receives one. It applies at Spice Collection, and does not multiply desert spice or grant Basic stronghold income. E3 p.8, base p.13

At turn end a co-occupied stronghold counts once for the alliance and Ecaz controls it. Three strongholds suffice for their allied victory only when both factions co-occupy all three; otherwise ordinary allied victory conditions still matter. A two-and-one distribution is not three co-occupied strongholds. Advisors cannot manufacture a second fighting occupier. The p.15 FAQ forbids combining split ownership of the three Tech Tokens into an extra stronghold and confirms Ecaz/Fremen co-occupation in Sietch Tabr does not spoil Fremen's special victory. E3 pp.8,15, base p.12

A single holder of all Tech Tokens can still gain their ordinary stronghold equivalent. That does not itself establish a third territory occupied by both allies; the special three-co-occupied test must retain its physical co-occupation requirement. This is the direct composition of the threshold wording with the token FAQ, not a new permission to combine token custody.

Karama: native and converted

The Occupy cancellation is explicitly before Battle Plans are formed. Once canceled, the chosen battling faction dials its own forces and the other ally contributes zero. It does not revoke their three-stronghold victory permission or evict legally co-occupied forces. Preserve the selected combatant; do not quietly select whichever faction now has more forces. A separate Collection cancellation denies Ecaz's co-occupation bank income while the ally collects normally. It is not a cancellation of desert collection, territory access or the whole alliance. E3 p.16

An Advanced BG Worthless conversion uses the same substantive timing and result as a printed Karama. Conversion is not permission to cancel after a plan has become committed, to see the selected opponent's private plan, or to add a second Occupy choice after cancellation. The existing BG conversion response can decide whether the Worthless operates as Karama; after that decision the original source-bound Occupy request resumes. An allowed cancellation must recompute force ownership, support and dial options before any plan becomes valid. Base p.18

The p.16 row explicitly states zero combat contribution for the non-battling ally but does not itself give a full casualty sentence for that canceled case. Ordinary-winning-side retention and losing-side destruction are a plausible composition; do not advertise an independently retrieved designer ruling on that consequence. This is a separate source check from native rounding.

Historical bounded recommendation

Implement a pure, owner-labelled Occupy battle profile shared by validation, AI, public preview and actual scoring: current Ecaz/ally side; selected combatant; battle-order actor; dialed force owner; exact Ecaz contribution; ordinary winner casualty split; and canceled profile. Keep the rounding policy explicit until resolved; retain storm-separated and advisor exclusions, and never read the non-selected ally's hand. Integrate a durable public combatant-selection stage and the pre-plan Karama opportunity before admitting plans. The p.8 arithmetic can be represented as a labelled candidate in an isolated pure helper, but an execution path must not silently treat it as a designer-corrected rule. The independently supported allied-entry and battle-side identity work is the safer immediate production slice.

Acceptance examples should include Ecaz counts 1–5, either selected faction, all three relative storm positions, both win/loss outcomes, Basic/Advanced supported dials, Fremen free strength, advisor-only BG exclusion, printed and BG-converted cancellation, and JSON recovery after selection. Separate targeted follow-ups should cover the source-scoped special battle effects above, agreed/default desert allocation, Advanced Collection, and actual three-co-occupied victory. Keep the optional Homeworld/Nexus/Discovery modifiers explicitly outside the initial profile until their own faces and outcomes are wired; no ordinary Occupy rule may be used to bypass them.

The September audit changed no runtime and corrected the false designer attribution. The October Advanced cutover and the October Basic even-force composition above are separate; optional/exotic interactions and independently unanswered cancellation details retain their historical boundaries.