dota_factory/docs/content_design.md

Content Design — Ships & Modules

First real-content iterations (June 2026). Pass 1 defined ship hull grids and module surface masks; pass 2 defined the production tree (recipes). Stats and threat costs in the config files are still placeholders for the balancing pass.

Design principle: footprint gating

Which module fits on which hull is controlled purely by geometry — no explicit allow-lists. Each hull grid is shaped so that it physically cannot contain the footprint of modules from a larger size class. This keeps the rules transparent to the player ("it doesn't fit because there is no room") and makes them trivially moddable through the config files alone.

Module footprint ladder

Footprint	Modules	Smallest hull that fits it
1x1	laser_cannon_s, salvager, repair_tool	drone
1x2	maneuvering_thrusters, sensor_booster, armor_plates	frigate
1x3	afterburner	frigate (eats most of it)
L-shape (3 cells)	weapon_stabilizer, weapon_primer, weapon_upgrade	frigate
2x2	laser_cannon_m, drone_bay	cruiser
3x3	laser_cannon_l	battleship
2x6	drone_hangar	carrier (only)

Hull grids

O = buildable cell, X = hull structure (not buildable).

drone (xs, 1 cell) — exactly one 1x1 module: a small gun, a salvager, or a repair tool. This is what makes drone roles swappable.

O

frigate (s, 5 cells) — plus shape. Every 1x2 placement crosses the center cell, so at most ONE 1x2 support fits; alternatively one L-shaped weapon modifier or one afterburner through the center line. Gun-boat with one or two support modules, as intended.

XOX
OOO
XOX

destroyer (s, 8 cells) — gun deck with three turret bumps. More cells than the frigate (more small guns), but still no 2x2 area anywhere, so medium hardware can never be mounted.

OXOXO
OOOOO

cruiser (m, 12 cells) — notched corners. Fits at most two 2x2 m guns (stacked through the middle), leaving the side cells for supports. No 3x3 area.

XOOX
OOOO
OOOO
XOOX

battlecruiser (m, 16 cells) — split bow with two gun cheeks, tapered stern. Fits three 2x2 m guns — one more than the cruiser — with small support slots left over. The bow split and stern taper prevent any 3x3 area (no l gun) and any 2x6 area (no drone hangar).

OOXXOO
OOOOOO
XOOOOX
XXOOXX

battleship (l, 24 cells) — broadside hull with notched flanks on every other row. Fits four 2x2 m guns (two per gun deck) — one more than the battlecruiser — with bow, stern, and flank cells for supports. All 3x3 placements crowd the center columns, so at most ONE l gun fits: mounted center it blocks every m gun mount (pure support strips remain), mounted offset it still allows two m guns. The notched rows are never adjacent-and- full, so no 2x6 drone hangar fits.

XOOOOX
OOOOOO
XOOOOX
OOOOOO
XOOOOX

dreadnought (xl, 36 cells) — the main battery deck is split into three 3x3 gun slots by structural spacer columns, so exactly three l guns fit side by side (or m guns / supports in unused slots), plus bow/stern strips for supports. The spacers cap every horizontal run at 5 cells, so the 2x6 drone hangar can never fit — the carrier stays the only hangar hull.

XXXOOOOOXXX
OOOXOOOXOOO
OOOXOOOXOOO
OOOXOOOXOOO
XXOOXXXOOXX

carrier (xl, 37 cells) — the top flight deck (rows 0–1) is the only region wide enough for the 2x6 drone hangar, and exactly one fits. The middle deck row is broken up by elevator shafts (X cells placed so every 3-column window hits one), which is what prevents any 3x3 l gun from ever fitting. Lower decks hold supports and 2x2 point-defense m guns.

XOOOOOOOOX
OOOOOOOOOO
OOXOOXOOXO
XOOOOOOOOX
XXXOOOOXXX

Verified gating matrix

Checked programmatically against the configs (all four mask rotations, all placements) with tools/verify_layouts.py — re-run it after editing layout grids or surface masks:

python dota_factory/tools/verify_layouts.py

Footprint	drone	frigate	destroyer	cruiser	battlecruiser	battleship	dreadnought	carrier
1x1	x	x	x	x	x	x	x	x
1x2		x	x	x	x	x	x	x
1x3		x	x	x	x	x	x	x
L-shape		x	x	x	x	x	x	x
2x2				x	x	x	x	x
3x3						x	x
2x6								x

Maximum simultaneous (disjoint) placements: m guns — cruiser 2, battlecruiser 3, battleship 4; l guns — battleship 1, dreadnought 3; drone hangar — carrier 1.

Production tree (first pass — superseded)

Superseded (July 2026): this first-pass tree predates the rules in docs/progression_design.md and will be replaced. The decisions for the redesign are recorded in "Production tree v2 — decisions" below; the text of this section is kept for reference until the new tree lands.

Design principle: each game phase adds exactly one new base input chain, so factory complexity ramps alongside ship size.

Phase	New input	How acquired	Unlocks
early	iron_ore, copper_ore	mined	drone, frigate, destroyer; small guns and basic supports
mid	titanium_ore	mined (3x slower than iron)	cruiser, battlecruiser; m guns, drone bay, weapon modifiers
late	advanced_alloy	ONLY from reprocessing salvaged scrap	battleship, dreadnought, carrier; l guns, drone hangar

The advanced_alloy gate is the core loop hook: capital ship production requires fighting (salvaging scrap from kills and reprocessing it), not just mining. The reprocessing plant turns 5 scrap into iron/copper/titanium ingots or advanced_alloy probabilistically.

Intermediate components, by tier:

• Tier 2 (early): copper_wire (copper), steel_plate (iron), circuit_board (iron + wire), building_block (iron).
• Tier 3 (mid): mechanical_parts (steel + iron), targeting_unit (circuits
  • wire), drive_unit (steel + mechanical_parts + circuit), titanium_frame (titanium + steel).
• Tier 4 (late): reinforced_plating (steel + advanced_alloy), capital_core (targeting_unit + drive_unit + 2 advanced_alloy).

Hulls and modules consume intermediates of their tier: early items are built from tier-2 parts, midgame items require tier-3 parts (deeper chains, more assemblers), capital items require tier-4 parts (and therefore combat). Hull items are named <ship>_hull; module items <module>_module. Every item has an [items.*] entry in visuals.toml; hull item outlines match the ship's fleet color from [ships.*].

Consistency is checked by tools/verify_recipes.py — re-run it after editing recipes, ship/module materials, or visuals:

python dota_factory/tools/verify_recipes.py

It verifies every consumed item has a producer, every item has a visuals entry, flags orphaned items, and prints which items are reprocessing-only (currently exactly advanced_alloy).

Production tree v2 — decisions (July 2026)

The old tree is discarded; the new one is designed against the rules in docs/progression_design.md (ratio curve, cost ladder, cost archetypes, refactorability). Decisions fixed so far:

Base inputs (4) and fiction

• iron_ore, copper_ore — from the start, minable on every asteroid tile. Fiction: the asteroid is an M-type (metal) body — its bulk rock is ore, which is why the shipyard operation was built here at all.
• quartz — mid-game, minable only on geode deposit patches in expansion territory (see the Resource deposits rules in progression_design.md). Fiction: ordinary silicate dust is everywhere and worthless; chips and optics need rare, pocket-bound optical-grade crystal.
• voidsteel — late-game, obtained only by reprocessing scrap. Fiction: battle-forged — formed when weapon plasma anneals hull metal in the violence of ship destruction. Any wreck yields it, including the player's own; no foundry can replicate it.
• titanium is dropped. Its hull-gating role moves to quartz-era control systems ("you can smelt all the steel you want, but you cannot steer a battlecruiser without electronics") and possibly a quality steel step — e.g. a long-running hardened-steel recipe (time-heavy archetype). Open question for the tree draft: quality step vs. electronics-only gating; explicitly not sheer steel quantity alone.

Material palette (fingerprints per family)

• iron/steel — structure.
• copper — conduction and heat: wiring, coils, heat sinks.
• silicon family (all derived from quartz): silicon (logic, sensors), glass/optics (lenses, focusing crystals), ceramics (heat shielding, insulators). Carries the non-metal variety without extra base inputs.
• voidsteel — capital-tier structure and exotics.
• Deliberately skipped: carbon (mostly redundant with copper/ceramics), plastics (drags in Factorio-style chemical chains; ceramics read more sci-fi anyway), volatiles/ice (materials are build costs only — no consumption mechanic to justify fuel).

Weapons

• All current lasers are renamed to railguns (laser_cannon_s/m/l → railgun_s/m/l); footprints and the gating matrix are unchanged. Implementation stays as-is (instant damage application, no projectile, no ammunition) — the beam visual reads as a tracer round. Materials: iron slugs, copper coils, steel rails — the starting-metal fingerprint.
• Lasers are reserved for later as a genuinely distinct weapon type (e.g. once projectile/ammunition mechanics exist for other families), arriving with quartz optics. More weapon types are planned; railguns are simply the baseline tech that ships with v1.

Tree structure — draft

Recipes are sketched as input lists only; quantities and durations come in the numbers pass, tuned so every chain sums to its threat-ladder value and follows the ratio curve (t1 nice → t4 strange). Glass/optics are cut from v1 — their only consumers would be lasers, which are deferred; the silicon family ships as silicon + ceramics.

Mined (miner): iron_ore, copper_ore (every tile), quartz (geode deposits in expansion territory).

Smelted (smelter — exactly one recipe per input item):

output	input	ratio class
iron_ingot	iron_ore	nice (1:1 or 1:2)
copper_ingot	copper_ore	nice
silicon	quartz	mid entry
iron_ingot	scrap	the safe, boring scrap sink

Reprocessing pool (scrap): iron_ingot, copper_ingot, silicon, voidsteel — the only source of voidsteel. Weights authored for the fully unlocked pool state.

Tier 2 — early intermediates (clean ratios, ~2:3):

item	inputs	role
steel_plate	iron_ingot	structure backbone, highest volume
copper_wire	copper_ingot	conductors
copper_coil	copper_wire	electromagnets: railguns, thrusters
building_block	steel_plate	depth-3 chain = the doubling-time knob

Tier 3 — mid intermediates (strange ratios begin, need quartz):

item	inputs	role
control_chip	silicon + copper_wire	electronics gate for m+ hulls
capacitor_bank	copper_coil + silicon	power for railgun m/l
hardened_steel	steel_plate (long cycle)	quality gate for m+ hulls; time-heavy
ceramic_plate	quartz	heat shielding: drives, l guns, capitals
drive_unit	steel_plate + copper_coil + control_chip	propulsion for m+ hulls

Tier 4 — late intermediates (need voidsteel):

item	inputs	role
voidsteel_plate	voidsteel + hardened_steel	capital structure
capital_core	voidsteel + capacitor_bank + control_chip	capital heart

Hull items (<ship>_hull, assembler-made; the shipyard consumes the hull item plus module materials). The m+ hull gate is resolved as both of the open-question options: hardened_steel (quality steel, a deliberately long-running recipe — the time-heavy step) and control_chip (electronics):

hull	inputs
drone_hull	steel_plate
frigate_hull	steel_plate + copper_wire
destroyer_hull	steel_plate + copper_coil
cruiser_hull	hardened_steel + control_chip
battlecruiser_hull	hardened_steel + control_chip + drive_unit
battleship_hull	voidsteel_plate + drive_unit + control_chip
dreadnought_hull	voidsteel_plate + capital_core + drive_unit
carrier_hull	voidsteel_plate + capital_core + drive_unit

Module items (<module>_module, assembler-made prefabs — kept as items so shipyard belt inputs stay simple and module production can be stockpiled):

module	inputs	archetype
railgun_s	steel_plate + copper_coil	balanced
salvager	steel_plate + copper_wire	balanced
repair_tool	steel_plate + copper_wire	balanced
armor_plates	steel_plate (many)	material-heavy, fast
maneuvering_thrusters	steel_plate + copper_coil	balanced
sensor_booster	copper_wire + copper_coil	lean (an antenna, no chip)
afterburner	copper_coil + steel_plate	balanced
weapon_stabilizer	steel_plate + copper_coil	balanced
weapon_primer	capacitor_bank + copper_coil	mid; time-heavy
weapon_upgrade	control_chip + copper_coil	mid; time-heavy
railgun_m	capacitor_bank + steel_plate + copper_coil	mid
drone_bay	control_chip + steel_plate + copper_coil	mid
railgun_l	capacitor_bank + hardened_steel + ceramic_plate	late
drone_hangar	voidsteel_plate + control_chip + drive_unit	late (carrier only)

Refactorability check (the default technique holds): railgun_s → m introduces capacitor_bank, built from a subset of the small gun's inputs (copper_coil) plus the new base resource (silicon); the m gun otherwise reuses the small gun's inputs. Hulls likewise: cruiser adds hardening (fed by the existing steel line) and chips (fed by the new quartz territory) without touching the iron/copper core.

Shortcut recipe candidates (drop-only assembler schematics; not every strange chain gets one):

• iron_ore → steel_plate — skips the ingot step on the highest-volume chain in the game.
• quartz → control_chip — skips silicon on the electronics chain.
• iron_ingot → hardened_steel — a nicer-ratio route past the deliberately awkward hardening step.

Item count: 3 mined + scrap + 3 smelted + 4 t2 + 5 t3 + 3 t4 (incl. voidsteel) + 8 hulls + 14 modules ≈ 41 item types — same scale as the first-pass tree. verify_recipes.py must be re-run once this lands in recipes.toml; the numbers pass should add a threat-report tool that prints per-item threat values and producer:consumer ratio tables.

Numbers — first pass

These numbers are live in the config files and verified by tools/threat_report.py; quantities and durations are tuned so fitted ships land on the threat-cost ladder and the ratio curve is realized. The typical loadouts are geometry-validated against the hull grids and configured as each ship's default_modules: the L-shaped weapon modifiers turned out not to fit alongside the full gun complement on the battlecruiser and dreadnought, so their loadouts use armor and small-railgun fillers instead — which moved all three heavy ships closer to their ladder targets.

Economy constants: scrap_per_threat = 0.25 (1 scrap per 4 threat destroyed — a cruiser kill drops ~59 scrap). Reprocessing: 4 scrap per cycle, 4 s, full-pool weights iron_ingot 30 / copper_ingot 30 / silicon 20 / voidsteel 20 → threat(voidsteel) = (4·4 + 4)/0.2 = 100. Scrap smelting: 1 scrap → 1 iron_ingot, 1 s — deliberately value-losing (4 threat of scrap becomes a 2-threat ingot); reprocessing is the value-preserving path.

Recipes (dur in seconds; threat is per output item):

item	recipe	dur	out	threat
iron_ore / copper_ore	miner	1	1	1
quartz	miner (deposit)	2	1	2
iron_ingot	1 iron_ore	1	1	2
copper_ingot	1 copper_ore	1	1	2
silicon	1 quartz	2	1	4
steel_plate	2 iron_ingot	3	1	7
copper_wire	1 copper_ingot	1	2	1.5
copper_coil	2 copper_wire	1.5	1	4.5
building_block	2 steel_plate	2	4	4
control_chip	1 silicon + 2 copper_wire	5	1	12
capacitor_bank	2 copper_coil + 1 silicon	5	1	18
hardened_steel	3 steel_plate	12	1	33
ceramic_plate	2 quartz	4	1	8
drive_unit	2 steel_plate + 2 copper_coil + 1 control_chip	8	1	43
voidsteel_plate	1 voidsteel + 1 hardened_steel	8	1	141
capital_core	2 voidsteel + 1 capacitor_bank + 1 control_chip	10	1	240

Module prefabs (contribution = item threat + module production time):

module	recipe	dur	mod. time	contribution
railgun_s	1 copper_coil	1	1	6.5
salvager	1 steel_plate + 2 copper_wire	2	1	13
repair_tool	1 steel_plate + 2 copper_wire	2	1	13
armor_plates	4 steel_plate	3	1	32
maneuvering_thrusters	1 steel_plate + 1 copper_coil	2	1	14.5
sensor_booster	2 copper_wire + 1 copper_coil	2	1	10.5
afterburner	2 copper_coil + 1 steel_plate	3	1	20
weapon_stabilizer	1 steel_plate + 1 copper_coil	2	1	14.5
weapon_primer	1 capacitor_bank + 1 copper_coil	4	2	28.5
weapon_upgrade	1 control_chip + 1 copper_coil	4	2	22.5
railgun_m	1 capacitor_bank + 2 steel_plate + 1 copper_coil	4	3	43.5
drone_bay	1 control_chip + 2 steel_plate + 1 copper_coil	4	3	37.5
railgun_l	1 capacitor_bank + 2 hardened_steel + 1 ceramic_plate	6	4	102
drone_hangar	1 voidsteel_plate + 2 control_chip + 1 drive_unit	10	6	224

Ships (fitted = hull item + ship base time + typical loadout; the typical loadouts double as the default_modules for enemy waves):

ship	hull recipe	dur	base	typical loadout	fitted (target)
drone	1 iron_ingot	1	1	railgun_s	10 (10)
frigate	2 steel_plate + 1 copper_wire	2	2	2× railgun_s, maneuvering_thrusters	47 (40)
destroyer	3 steel_plate + 2 copper_coil	4	3	3× railgun_s, armor_plates, sensor_booster	99 (80)
cruiser	2 hardened_steel + 2 control_chip	6	4	2× railgun_m, armor_plates, maneuvering_thrusters	234 (200)
battlecruiser	3 hardened_steel + 2 control_chip + 1 drive_unit	8	5	3× railgun_m, armor_plates, 2× railgun_s	355 (350)
battleship	3 voidsteel_plate + 1 drive_unit + 2 control_chip	10	6	railgun_l, 2× railgun_m, weapon_stabilizer, 2× railgun_s	723 (700)
dreadnought	5 voidsteel_plate + 1 capital_core + 2 drive_unit	12	8	3× railgun_l, 4× armor_plates, railgun_s	1492 (1500)
carrier	5 voidsteel_plate + 1 capital_core + 2 drive_unit	12	8	drone_hangar, 2× railgun_m, 2× armor_plates, sensor_booster	1436 (1500)

Checks:

• Ratio curve realized: t1 all 1:1 (miner:smelter); t2 clean 2:3 (ingot→plate, wire→coil); t3 strange — 2:5 (silicon→chip), 3:5 (coil→capacitor), 3:4 (plate→hardened, plate→drive); t4 inverted 3:2 (hardened→voidsteel_plate).
• Belt feasibility: worst input demand is 1.33 items/s (wire→coil, plate→armor) — under the ~2/s single-belt cap everywhere; no accidental multi-belt recipes.
• Block economy: building_block = 4 threat ⇒ for the 4-minute doubling at 30% capacity, the average building must cost ≈ 18 blocks (0.3 × 240 / 4). buildings.toml costs should be set around that mean (belts cheap, producers ~20–30).
• Small-end deviation: frigate–cruiser land 6–24% hot because the fixed chain overhead dominates small hulls. Recommendation: accept and adjust the ladder targets to the achieved values (the ~×2-per- class curve shape is preserved) rather than thinning the early chains below readability.
• Rule bug discovered: the scrap→iron_ingot smelter recipe combined with REQ-THREAT-ITEM's max-across-recipes rule would set threat(iron_ingot) to the scrap path (1 + 4 = 5, or more at other scrap values) instead of 2, inflating every downstream item. Fix: scrap-consuming recipes count toward an item's threat only when no scrap-free recipe produces that item, mirroring the reprocessing-path rule (see progression_design.md action items).

Combat stat anchors — first pass

All combat stats derive from a few anchors so that power-per-threat holds by construction; the arena suite in bin/balancing/data/balancing.toml verifies it empirically. Change the anchors first, re-derive stats second.

• Weapon DPS per threat pays a concentration tax that grows with gun size: railgun_s 2 dmg × 2.0 Hz = 4.0 DPS at 6.5 threat (0.62), range 50; railgun_m 14 × 1.5 = 21 at 43.5 (0.48), range 70; railgun_l 52 × 0.8 = 41.6 at 102 (0.41), range 100. Arena round 1 (see below) showed that at equal threat, concentrated fleets win flawlessly with a flat-ish DPS curve — concentration itself (focus survivability, range, no losses-to-attrition of own DPS) is worth paying for, so bigger guns get less raw DPS per threat.
• Hull HP = 15 per threat of hull contribution (incl. ship base time) as the prior; per-hull HP is the empirical trim knob for arena results (guns are shared across many hulls, so gun changes move many matchups at once — hull HP moves exactly one). Current values after round-5 trims: drone 60, frigate 300, destroyer 550, cruiser 1500, battlecruiser 2400, battleship 6300, dreadnought/carrier 24000. (The capital hulls sit well above the 15/threat prior while their guns sit below the DPS prior — the arena consistently prices big ships as tanks with taxed guns.)
• Armor HP ≈ 37 per threat — a strong premium over hull HP (15) because armor is pure HP with no capability, and fights snowball: killing enemies removes their DPS, while surviving merely delays — so a point of HP must be cheaper than a point of DPS. armor_plates adds 1200 HP at 32 threat (raised from 640 → 1000 → 1200: the all-guns destroyer beat the armored one in three consecutive arena rounds).
• TTK check: a mirror cruiser duel (51 DPS vs 2140 EHP) lasts ~42 s — inside the 30–60 s parity-fight band implied by the 25-ship fleet target. Capital mirrors run longer (~90 s), deliberately.
• Repair ≈ 0.7 HP/s per threat: repair_tool heals 9 HP × 1 Hz at 13 threat (nerfed 25 → 12 → 9: in-combat sustain effectively removes enemy DPS and must be priced like DPS, not like HP — the escorted team held a persistent +24% at 12). Salvager: collection range 60 (was a nonsensical 500, beyond any sensor), cargo 20, 0.5 collections/s. scrap_despawn_seconds raised 30 → 120: a capital kill drops hundreds of scrap collected one per cycle, so 30 s despawned nearly everything.
• Mobility ladder is monotone in size — small = fast and nimble (drone 45 m/s, 60 accel) down to capitals (10 m/s, 8 accel); the old placeholder had the drone as the least agile ship. Sensor ranges keep the existing 150→350 ladder; all weapon ranges stay below sensor ranges.
• Weapon modifiers are capital economy: ×1.2 damage at 22.5 threat beats adding a gun once a ship carries more than ~68 threat of weapons — so modifiers pay off on gun-heavy big hulls and are a waste on small ones. Damage/rate kept at ×1.2; the stabilizer's range multiplier was cut ×1.5 → ×1.3 after round 2 — range is the strongest stat in the orbit-AI's hands (free approach fire), and the stabilized battleship was the one loadout still overperforming.
• Stations: a fresh player station (3000 HP, 25 dmg × 1 Hz, range 120) holds one early parity wave unaided; the enemy station at level 0 matches it exactly and scales per push level (+1500 HP, +12 dmg, +0.1 Hz per level). HQ 5000 HP. Range reduced from 200 in round 1: at 200 (4× a small gun's range) two stations annihilated a 3× threat swarm through approach fire alone — range, not HP, is the station dominance lever.
• Known imbalance, accepted: the carrier loses its equal-threat arena until the drone-launching capability exists — the hangar is 224 threat of dead weight. Do not compensate with stats; fix by implementing drones.

Arena round 1 results (2026-07-04): mirrors snowball as expected (40–50% of the winner survives) but team 1 won all three mirrors — possible sim-side bias, under investigation (progression_design.md action items 6–7). All four equal-threat cross-tier matchups were flawless wins for the concentrated side; glass destroyers beat armored; repair escort beat raw numbers flawlessly; two stations shrugged off a 3× threat swarm. Round-2 knob changes: concentration tax on m/l gun DPS (17→14, 70→52 damage), armor 640→1000, repair 25→12, station range 200→120. Gun ranges deliberately untouched — if concentrated fleets still win flawlessly in round 2, the range ladder (kiting) is the next suspect, ahead of further DPS cuts.

Arena round 5 results (2026-07-04, with fight durations) — combat pass converged: durations validate the TTK anchor: drone mirror 22.7 s, cruiser mirror 71.2 s, battleship mirror 94.6 s, most fights in the 22–72 s band; the destroyers-vs-dreadnought slugfest is the outlier at 214 s (extreme tank vs. chip damage — acceptable for the extreme case). Converged: mirrors 5–10%, swarm vs cruisers +15%, frigates vs battleship dropped to +13% on its own (confirming the noise-floor read — now inside band anyway), dreadnought at +3% knife edge, glass vs armored +10%, two-to-one 78%, station assault stable at 44%. Two two-round signals actioned in round 6: mixed beat battlecruisers again (+20%, +18%) → battlecruiser hull 2200→2400; repair escort held +24% twice with all frigates surviving → repair_amount 12→9. Everything else is frozen; further refinement moves to real-game playtests.

Arena round 4 results (2026-07-04) and convergence policy: mirrors at 0–3%; swarm vs cruisers +7%; glass vs armored resolved at +3% for armored (armor 1200 confirmed after four rounds of signal). Noise floor established: the battleship held +23% straight through a −10% EHP cut, and repair drifted 14→24% with no repair changes — single-run margins move ~±10% on a re-roll. Policy from here: only act on signals that persist two rounds; treat anything inside ±20% as converged for v1 (real-game playtests refine further). Armor-coupling overshoots corrected in round 5: battlecruiser 2000→2200, dreadnought/carrier 22500→24000 (two-round persistent deficit vs destroyers). Accepted, not chased: the stabilized battleship beats a pure frigate fleet by ~23% — the edge survived a stat cut, so it is mechanical (range), and "the extreme smallest-ship fleet modestly loses to a range-fitted capital" is desirable doctrine texture; the fair anti-capital answer is the mixed fleet, whose arena is balanced.

Arena round 3 results (2026-07-04, narrow lanes): arenas were narrowed (height 10) so fleets can no longer pass each other — arena geometry is part of the fixture from here on; margins are only comparable within the same geometry. Results: mirrors healthy (1–7%); dreadnought closed to −11%; battleship improved to +22% (still over); drone swarm now beats cruisers by 14% (full engagement in the narrow lane favors numbers); mixed vs battlecruisers +18%; glass beat armored for the third consecutive round (+12%); repair 14%, two-to-one 78%, station set cracked at 51% by the 3× swarm (geometry changed under the stations — watch before touching stats). Round-4 knob changes: armor_plates 1000→1200 (the persistent glass signal; also lifts the armor-carrying cruiser/BC/destroyer/DN loadouts), battlecruiser hull 2500→2000 (net −300 EHP after its armor gain), battleship 7000→6300 (no armor in its loadout, clean −10%), dreadnought/carrier 19000→22500 (the opposing destroyers gain armor too, so the DN needs the larger step).

Arena round 2 results (2026-07-04, with EHP-margin logging): the mirror "bias" was noise — repeated runs split randomly, and winners finish at 5–7% EHP (near-mutual annihilation, healthy). Drones vs cruisers 10% and mixed vs battlecruisers 14% — near parity. The two capital fights diverged: battleship beat frigates at +33% while the dreadnought lost to destroyers at −37% — same guns, so the difference is the destroyers carrying the buffed armor plus the battleship's stabilizer putting all its guns at 75–150 m against 50 m opponents. Glass vs armored narrowed to 11% (watch; if it persists, armor 1000→1200). Repair escort 11% — fair. Two-to-one decisive at 80%; station assault cracked by the 3× swarm at 13% (stations strong but breakable). Round-3 knob changes: stabilizer range ×1.5→×1.3, and per-hull HP trims — battlecruiser 2700→2500, battleship 7500→7000, dreadnought/carrier 15500→19000.

Pacing pass — first values

The mapping between station levels and boss cycles: at the intended push cadence (first set around cycle 2–3, roughly one per cycle from mid on), the destroyed set's level ℓ is reached around cycle ℓ+2. So level 2 ≈ the mid boundary, level 6 ≈ mid-late, levels 8–9 ≈ capitals with enough cycles left to ramp voidsteel production before late.

Starting set (unlock_at_station_level = -1), per the starting-set rule (exactly enough to reach the first push): drone, frigate, railgun_s, salvager. Note: the building_block recipe must carry an explicit -1 — building blocks appear in no schematic's materials, so implicit unlocking can never reach it (fixed in this pass; it was silently locked).

Unlock ladder (level → unlocks; ← marks unlock_requires):

level	ships	modules	recipes
0	destroyer	repair_tool, armor_plates
1		maneuvering_thrusters, sensor_booster	shortcut_steel_plate
2	cruiser	railgun_m, afterburner	shortcut_control_chip, shortcut_hardened_steel
3		weapon_stabilizer
4	battlecruiser ← cruiser	weapon_primer, weapon_upgrade
5		drone_bay
6	battleship ← battlecruiser	railgun_l ← railgun_m
8	dreadnought ← battleship
9	carrier ← battleship	drone_hangar

Level 0's pool has exactly three entries, so the first push offers a full dialog with no artifacts competing. Level 2 is the quartz gate: cruiser and railgun_m are the first schematics whose chains reach quartz (the shortcut recipes' outputs only become implicitly unlocked alongside them, so they cannot drop early).

Threat rate 2*x + 0.15*x*x, tuned against the factory-size curve with roughly half the player's output assumed military:

cycle x	rate (threat/s)	player military (≈ curve/2)
2	4.6	~12
6	17.4	~30
15	63.8	~60
20	100	~75
24	134	—

Below the player early, crossing at the late boundary, overwhelming by ~24 — the endless-race shape from the progression rules. Typical wave at x=6 (30 s gap): ~520 threat ≈ two cruisers plus escorts; boss at x=15: ~3800 threat ≈ two and a half dreadnoughts.

Economy: starting_building_blocks 1000 → 200 (minimal block loop

• first shipyard ≈ 150, plus beginner slack); expansion cost 200 → 400 flat (placeholder until the escalating formula, action item 4); artifact_win_count 3 → 5 with chance 0.05*x: ~15–18 pushes in a winning run offer ~7 artifact options cumulatively, so winning takes choosing the artifact over a schematic five times — a real decision each time, and the win lands in the cycle 20–24 target window.

Balancing targets (first pass, July 2026)

The six root numbers for the balancing pass. Every derived value (threat rate, recipe quantities, block costs, scrap rates, unlock ladder) is tuned to hit these; when rebalancing later, change these first and re-derive, never the other way around. The rules they follow live in docs/progression_design.md.

All targets are in game time. The player can pause and accelerate, so real session length differs; playtests should measure both. The time unit is the boss cycle (world.toml boss_countdown_seconds, 300 s). Destroying a station set advances the boss countdown by boss_advance_seconds (60 s), so cycles run shorter than nominal when pushing actively — these targets deliberately ignore that; playtesting will show real run lengths.

1. Run length — a winning run takes up to 2 hours of game time: win around boss cycle 20–24. Losing runs end earlier.
2. Phase boundaries — early = cycles 1–5 (iron/copper, small hulls), mid = cycles 6–14 (quartz, medium hulls), late = cycles 15+ (voidsteel, capitals). Push cadence roughly one station set per cycle from mid onward. This fixes the unlock_at_station_level ladder and, with the artifact win count, the artifact_chance_formula pacing.
3. Factory size curve — producing buildings over time; when saturated, output threat/s equals this count, so this curve IS the player power curve. Targets: ~25 when the starting asteroid is full (end of cycle 2), ~60 at the start of mid (cycle 6), ~120 at the start of late (cycle 15), ~150 near the win. threat_rate_formula must remain a fraction of this curve, and buildings plus belts must physically fit the asteroid plus affordable expansions.
4. Threat-cost ladder — total production-seconds per fitted hull (including a typical module loadout): drone ~10, frigate ~40, destroyer ~80, cruiser ~200, battlecruiser ~350, battleship ~700, dreadnought/carrier ~1500. Every production chain must sum to its ladder value. Cross-check against (3): fitted-ship cadence = ladder value / (factory threat/s devoted to military) — e.g. a mid factory spending half of 60 threat/s on ships fields a fitted cruiser roughly every 7 s.
5. Fleet size — swarm-leaning: ~25 player combat ships as the standing mid-game fleet. Standing fleet = build cadence (4) × average ship lifetime, so this target drives time-to-kill and therefore the combat stat magnitudes tuned in the arena pass.
6. Block economy roots — bootstrap complete (starting asteroid full) by the end of cycle 2; a factory spending ~30% of its capacity on blocks doubles in ~4 minutes early game; one expansion affordable per cycle at ~1/3 of block income, with escalating costs that eventually outrun any income (see the growth curve rules in progression_design.md).

Deliberate placeholders / open questions for later passes

• All new hulls have threat.cost_formula = "0" so enemy waves do not spawn them yet (WaveSystem treats any ship with positive threat cost as wave- eligible, regardless of unlock level). The balancing pass should set real threat costs together with default_modules loadouts so waves spawn them armed.
• All new hulls and all assembler recipes are unlock_at_station_level = -1 (available from the start) to make testing easy; the balancing pass should stagger these so mid/lategame recipes drop as schematics from enemy defence stations.
• Recipe quantities and durations are a first guess, deliberately roughly tiered (capital hulls ~60 s, drones 4 s); the balancing pass tunes them.
• drone_bay and drone_hangar are footprint-only placeholders: the drone launching capability does not exist in the simulation yet, so they define no capability section.
• Renames in this pass: laser_cannon_xs → laser_cannon_s (the old 2x2 laser_cannon_s became laser_cannon_m), armor_plate → armor_plates, manuvering_thrusters → maneuvering_thrusters (typo fix). Test data under bin/test/data/config intentionally still uses the old ids — it is an independent fixture set.