decisions/spine/SECTION5_NUMERICALS_BRIEF.md
Status: ✅ ALL 13 RATIFIED BY JOSH (2026-07-14) AND ALL 13 WIRED (verified 2026-07-14). The 10 DERIVABLE were Claude-applied; the 3 JOSH-CALL are now confirmed as-recommended — Josh: "okay on #9, okay on #12, #13 okay agreed." So: #9 HUD stack/layer ceilings = VRIL 7 layers / HEALTH 5 stacked bars / STAMINA-RAGE 1 non-stacking bar; #12 split-N cap 5 bonded / 3 standard + merge-arity 2→4; #13 enchantment source=both, slots 1/2/3/4 by rarity, re-roll cheap-rising with a lockable-affix floor. The wiring LANDED the same day (docs/spine/OPUS_HANDOFF_TIERDOC_AND_NUMERICALS.md item 2: the RN_ rune registry at commit a80b169 — see RN_0008/RN_0017 "(§5 #12 ratified)" — plus the T1_Combat_System_Spec §3 combat/HUD cluster); the only residue is the DISTINCT downstream per-ability grammar-column population (2/185), which is build-time design, not a §5 value. This brief is a closed record; no work remains here.
Prior status: BRIEF for Josh-final-yes / Fable-5 confirm (2026-07-14). Scope: the two §5-deferred numerical clusters that the combat/spellcraft engines need — the combat 4-bar-vs-canon reconciliation and the rune-system numbers. Driver: docs/FOUNDATION_COMPLETION_PLAN.md §5; ledger: docs/proposals/systems/RULINGS.md §5 (J7/J8) + docs/proposals/systems/combat-and-spellcraft.md (§6/§7 + Corrections C2/C5/C6). Indexed in docs/DOC_MAP.md.
These are consumer-layer values (counts / tiers / curves the downstream engines consume), not world-lore forks. They honor every locked count untouched — 77 ch · 22 threads · 72 weapons · 10 trades · 12-tier (120 gates) · 33 species · 22 familiars · 185 abilities · 69 regions — and the design laws (real-science immersive-sim, vril-altered physics, rated-R AAA fun-first).
Two classes, per the standing decision protocol:
Each row carries a concrete value/scheme (never a hand-wave), a one-line reasoning, and the strongest objection.
---
The core reconciliation (5 archetype-keyed systems → 4 canonical types → 3 always-on bars) is already constitutional: T1_Combat_System_Spec §3 lines 33-49 + HL_0074/HL_0075 landed it verbatim in commit b27da74. Items 1-4, 7, 8 are therefore confirm-the-landed-string; items 5-6 concretize canon fuzzy-strings; item 9 is a genuine ceiling.
| # | Item | Recommended value | Class | Reasoning (1-line) | Strongest objection |
|---|---|---|---|---|---|
| 1 | 5-systems → 4-types → 3-bars reconciliation | 4 canonical TYPES (HEALTH, VRIL, STAMINA, RAGE) surfaced as 3 always-visible bars: HEALTH always-on, VRIL always-on, and ONE archetype-swapped slot (STAMINA on Trickery/Sorcery, RAGE on Brute). RUNES (Sorcery off-hand) + AMMUNITION = contextual weapon overlays, never a 4th/5th always-on bar. | DERIVABLE | Landed verbatim in Combat Spec §3 line 41 + HL_0074 (J13, commit b27da74); this is a confirm. | A Sorcery off-hand shows Vril + Runes + Stamina + Ammo at once (~4 meters) — the clean "3 bars" story understates real HUD density; may need an overlay-declutter (fade-when-full) rule. Does not change the type-count. |
| 2 | HUD spatial layout of the 3 persistent bars | Left cluster, fixed: VRIL = a layered blue bubble (concentric deeper-blue rings, OoT-style) far-left; HEALTH = the main bar joined to the bubble's lower-right, stacking full bars MK-style as max grows; STAMINA/RAGE swap-bar joined directly under Health. Overlays (Runes/Ammo) render by the relevant weapon indicator, not in the core cluster. | DERIVABLE | Verbatim the J7 ruling now in Combat Spec §3 line 41; constitutional confirm. | Bubble-left / bars-right is unconventional for Western action-RPG players; a Fable-5 A/B pass could show it hurts stress-readability. But it is a deliberate Josh aesthetic (Zelda-bubble homage), an identity choice, not a defect. |
| 3 | RAGE behavior + the two canonical thresholds | Model = build-in-combat / decay-out (not deplete-from-full). BUILD to peak at ~30.0s sustained trading (front-loaded low, avg ~3.3%/s). On ceasing to attack, FREEZE 5.0s at current level, then DECAY to empty over ~10s (~10%/s). Fully empties out of combat. | DERIVABLE | Combat Spec §3 line 35 ("strongest after roughly 30 seconds") + line 41 ("freezes for roughly five seconds before it begins to deplete") — de-fuzzing landed canon strings; the ~10s decay slope is the only free parameter. | 5s-freeze + 10s-decay is generous — a Brute can disengage ~5s and re-open near-peak, trivializing hit-and-run vs slow bosses. But "roughly five seconds" is landed canon; only the decay slope is genuinely tunable. |
| 4 | STAMINA regen curve | OUT-OF-COMBAT: fast — full refill ~2.5s (~40%/s). IN-COMBAT: moderate — ~15%/s (~0.35-0.4× the out-of-combat rate), gated behind a ~0.5s post-action delay so it "recovers between abilities," not mid-cast. Spent by Trickery abilities, mobility, block-hold, parries (parry can refund at Mastery 8). | DERIVABLE | Combat Spec §3 line 36 ("regenerates quickly outside combat, moderately during combat") + line 41 ("spent in combat and recovered between abilities") fix the shape; rates are standard action-RPG feel. | Pure balance numbers with no anchor beyond "quickly/moderately"; real magnitudes come from the Phase-5M audit against actual ability costs, which do not exist yet. The 40%/15% split is correct in ordering, not necessarily magnitude. |
| 5 | VRIL pool regen model | Three income channels, density-scaled: (1) PASSIVE atmospheric absorption ~2-3%/s at neutral density, scaling with local vril-density (deadzone → ~0%/s with a small floor; vril-rich site → up to ~8%/s), attenuated for negative-signature players (canon 5.7); (2) REST (safe-site/camp) = full restore; (3) RECHARGE SOURCES (waterfalls/obelisks/pyramids/tombs/orbs) = rapid channel to full, the Orb strongest sustained income. Never auto-fills to max mid-combat in a deadzone (the intended difficulty knob). | DERIVABLE | Grounded in T1_Vril §1 (depletable/rechargeable pool) + §5.1/5.6/5.7; density-scaling ties regen to the canonical vril-density difficulty dial, so regen IS the cosmology knob, not a flat number. | Density-scaled passive regen risks a frustration spiral in the intentionally low-density human world (75/77 chapters) — near-zero passive regen makes the pool a chore. The small floor exists precisely so "constrained" never reads as "starved." |
| 6 | VRIL cost curve (5 cost-classes → fraction of pool) | Map the vril_cost_class enum to fractions of the early-game baseline max pool: low ~5%, medium ~12%, high ~22%, very_high ~35%, extreme ~55%. Fractions of baseline → as max grows via progression+exploration the effective cost shrinks. Cost is DERIVED by the balance formula, never author-set (spellcraft firewall). | DERIVABLE | T1_Ability_Tree owns the enum consumed by the damage formula (Combat Spec lines 19/71); geometric-ish spacing gives "extreme" the pool-draining-ultimate feel (e.g. Crown-seat Collective Surge). | Combat Spec line 102 explicitly locks absolute numbers to the Phase-5M audit ("numbers plug in when balance data exists"); 5/12/22/35/55 pre-empts a deferred ruling. Read as illustrative shape (correct ordinal spacing), not locked values. |
| 7 | HEALTH model + integrity-band coupling | HEALTH is the enumerated 4th canonical type (always-on) and does NOT passively self-regenerate (no out-of-combat auto-heal — this is what distinguishes it from Stamina). Restored only via vril-healing (Heart-seat tree), Alchemy potions (Trade 5, Herbalism-fed), recharge/rest sites, safe-camp rest. Incoming healing modulated by integrity bands (L1 +10%, L2 +5%; L3 no bonus +5% dmg-received; L4/L5 no bonus). | DERIVABLE | C2's "confirm HEALTH is the 4th bar" is RESOLVED by the constitutional landing (Combat Spec line 41 / HL_0074 enumerate HEALTH); no-passive-regen is the only model consistent with canon (spec gives only incoming-healing modifiers, never an auto-regen line) + Pillar-13 consumable-crafting. | No-passive-regen loads the Alchemy/Herbalism + Heart-seat economy heavily; if those are thin early (Trade 5 is chapter-gated) early chapters feel attrition-punishing. A minimal safe-zone-only regen floor may be needed — a tuning valve, not a change to the in-combat no-regen rule. |
| 8 | Max-growth "fifths" scheme + level/exploration split | 5 collectible pieces (fifths) = 1 full additional unit, uniform across all four types: 5 Health pieces = +1 stacked bar; 5 Vril pieces = +1 bubble LAYER; 5 Stamina/Rage pieces = +1 swap-bar unit. Two sources COMPOSE: level/character progression + found exploration/puzzle/achievement fifths. Recommended split ~60% progression / ~40% exploration (completionist meaningfully stronger; non-explorer never progression-gated). | DERIVABLE | Josh authored "fifths / 5 pieces = one full unit" himself (C6, Zelda heart-container convention); the once-flagged "exploration-grows-max for non-vril bars" is RESOLVED — Combat Spec line 41 landed it constitutionally. Only the 60/40 split is tunable. | C6 still tags the total number of fifths distributed across the locked 77 chapters (the real exploration ceiling) as a Fable-5 distribution job — so "5-per-unit" is safe, but the aggregate exploration-max is not yet a concrete number (feeds item 9). |
| 9 | Stack / layer CEILINGS | VRIL bubble layers cap at 7 (mirrors 7 devices / vril-seats), HEALTH stacked bars cap at 5 (mirrors 5 integrity bands, keeps the MK stack readable), STAMINA/RAGE = a single bar scaling numerically with NO visual stacking (the swap-slot must stay legible). Ceilings only; fifths+level growth fills toward them. | JOSH-CALL | A stack ceiling is a genuine numerical maximum-power quantum; anchoring 7/5 to existing locked counts makes them feel canonical + keeps the HUD legible (an unbounded MK stack becomes unreadable). | This is exactly the numerical ceiling the do-not-invent rule reserves — 7/5/1 pre-empts a balance decision that depends on real power-per-unit and on how the difficulty tiers + NG+/prestige want the ceiling to sit. The 7/5 anchoring is aesthetic, not balance-derived. |
---
Rides Trade-9 Inscription + the 185 abilities as custom spell-merges; three canon rune lanes (worn / spell-creation catalyst / enchantment) per T1_Combat_System_Spec "Resource and Rune-Enhancement Note" (landed) + RULINGS J8. No ordinary-rune grind (hard mandate). J8 fixed the lanes + gave the operator band (~12-16) + "4 rarity tiers"; the ceilings/curves route to Fable-5.
| # | Item | Recommended value | Class | Reasoning (1-line) | Strongest objection |
|---|---|---|---|---|---|
| 10 | Worn rune-operator library size | 16 worn operators, distributed across the six §2.3 grammar families as 4 / 3 / 4 / 4 / 1: SHAPE 4 (wheel, wave, orb, field) · DELIVERY 3 (charge-hold, sustained-channel, beam) · MOTION-LOCK 4 (split, direction-lock, hold-to-home, orbit) · TRANSFORM 4 (extract, thermal[melt/freeze], forge-shape, magnetise-conduct) · ELEMENT-COMBINE 1 (fuse). MOBILITY operators ride the existing 8+2 mobility ability rows — 0 new worn runes. | DERIVABLE | Sits at the top of Josh's stated band (J8: "~12-16"); 16 over the 185 closed vocab × 7 seat-elements × knob-ranges × transform-ordering yields combinatorially unbounded personal spells with zero new abilities (185 lock intact). Maps 1:1 onto the six authored operator families. | 16 worn operators is a real UX surface to teach/manage; if too many read as noise, a curated 12 (J8's floor) may play cleaner. Within the Josh band either way, so a confirm-the-endpoint, not a fork. |
| 11 | Rarity-tier count + names | 4 tiers: Common / Rare / Epic / Legendary. Common = basic shapes/deliveries; Rare = motion-locks (split, direction-lock); Epic = multi-step transform chains + hold-to-home; Legendary = dual-element fusion + lattice/EAM operators, tied to the 72 Legendary-weapon questlines. Rarity gates *access to an operator you then wear* — never a per-cast burn or a re-stock grind. | DERIVABLE | J8 gave "4 rarity tiers"; the names are the genre-standard ladder already used in combat-and-spellcraft §2.4, and the Legendary→72-Legendary-weapon-quest tie is already canon. Changes no locked count. | combat-and-spellcraft §6 originally reserved "rarity-tier count/names" for Josh, and Legendary's tie to the 72-weapon questlines gives it a canon-adjacent flavor worth a quick confirm. But the count is J8-set and the sub-taxonomy touches no locked number. |
| 12 | Split-N (live) + merge arity (authoring) caps | Live modulation split_N: unlocks at the Portal wisdom-test gate (the level 5→6 crossing) at N=2; +1 at each inner-grade milestone (tiers 8/10/12) → cap N=5 on an EAM device-bonded seat, N=3 on the standard unbonded seventh seat (the 6+1 asymmetry expressed as a ceiling). Authoring merge arity: 2 of the 185 per custom spell at Adeptus-Minor Inscription; +1 per major Inscription gate → cap 4 (the transform-chain-length-per-tier dial). | JOSH-CALL | §6 explicitly reserves "split_N caps per vril-seat tier" and "transform-chain length per Inscription tier" as needs-Josh numerical; the schema fixes only the ordering (split_N ≤ vril-seat tier, bonded > standard), not the magnitude. | 5/3 and 2→4 are genuine power ceilings whose magnitude depends on real add-count / target density and on how bonded-vs-standard should feel — pure balance data the repo does not yet hold. The gate-anchoring is structurally motivated, not balance-derived. |
| 13 | Enchantment curve (source · slots · re-roll) | SOURCE = BOTH: base item crafted via Engineering (T8) / Alchemy (T5, Herbalism-fed); enchant-runes both Inscription-adjacent-crafted AND Sigma-bounded milestone/boss drops — never a farm faucet. SLOTS per item tier: 1 / 2 / 3 / 4 for Common / Rare / Epic / Legendary (mirrors the 4 tiers). RE-ROLL: cheap deterministic cost that RISES per consecutive re-roll on the same item, with a hard anti-luck FLOOR — after ~5 re-rolls the pool narrows and one affix becomes lockable (Borderlands-4 targeted re-roll); re-roll currency is a common craft output, never a rare farm. | JOSH-CALL | combat-and-spellcraft C5 explicitly flags enchantment SOURCE / slot-counts / affix-pool / re-roll cost-curve as [DECISION NEEDED - JOSH] (numerical/system canon); the direction (both-source, slots=rarity, cheap-rising-with-lock-floor) is the grounded "innocent RNG" anti-grind rec, but the numbers are Josh/Fable-5. | Slots=rarity + a lockable floor is a clean anti-grind shape, but the affix POOL size and re-roll cost-curve are where "optimization" tips into "attrition" — precisely the fun/grind boundary Josh reserved. Needs the affix-pool population (a Fable-5 job) before the curve can be real. |
---
| Cluster | Items | DERIVABLE | JOSH-CALL |
|---|---|---|---|
| 1 — Combat 4-bar reconciliation | 9 | 8 | 1 (#9 stack ceilings) |
| 2 — Rune-system numbers | 4 | 2 | 2 (#12 split-N/merge, #13 enchantment) |
| Total | 13 | 10 | 3 |
The 3 JOSH-CALL items are the genuine numerical ceilings/curves whose magnitude needs real balance data: (#9) HUD stack/layer caps, (#12) split-N + merge-arity caps, (#13) the enchantment source/slots/re-roll curve. Everything else is either constitutionally landed (Combat Spec §3 / HL_0074, commit b27da74), a value inside a Josh-stated band (operator count in J8's ~12-16; rarity count = J8's 4), or a creative-within-vision concretization that changes no locked count.
RN_ rune registry + the Ability_Tree spellcraft-grammar columns + the Combat Spec regen/cost curves (WS7 anchor layer is already in place), refreshes the fidelity baseline, and lands it gated.