Decision layers and clocks

Section titled “Decision layers and clocks”

The NPR strategic planner runs the decision layers of a non-player race, from strategic clocks and system values to fleet missions, colony seeds, production requests and research progression. The colony, terraforming, installation, production queue and ground deployment mechanics are in NPR colony and economic AI. The examples on this page are synthetic. Comparisons and orderings describe what the game does, not an intended optimal strategy.

The main subject is standard NPRs. Special-race rules are described where they share these entry points. Only NPR races run this planning. It is not an automatic planner for player races. The time increment runs it for every NPR , and the construction cycle runs it again for the same races.

This page covers planning and its immediate consumers. Weapon-by-weapon targeting and complete special-race generation are outside it. Design-side inputs belong to NPR race generation, which owns operational groups and automated ground forces. NPR fleet readiness and recovery owns readiness, fuel, repair and reload destinations. Movement and logistics owns route execution and transfers. These planners often queue those mechanisms rather than deliver ships, installations or colonists at once.

There is no single unified optimiser. Four layers share the work. The race planner allocates roles across systems. The system planner chooses local responses in each system. Each fleet carries its own role, readiness and order state. The population planner produces local projects from the racial design philosophy and the capital’s minerals. Research follows an ordered design philosophy progression. It does not select the technology that best counters the current contact.

Trigger or clock Literal gate Consequence
Pre-increment strategic call NPR AI processing enabled, race is an NPR . Not forced, elapsed time equal to the increment length Rebuild threat classifications and run eligible strategic and local decisions
Construction-cycle strategic call NPR, NPR AI processing enabled, . Forced, elapsed time 0 Enter the main assignment block regardless of contact presence
Threat-response clock Forced, or (systems with recent hostile ships exist and now minus the threat-response timestamp is above 10,000 seconds) Set the timestamp to now and run the main assignment block
Monthly clock Inside the main block only. Now minus the monthly timestamp is above 2,592,000 seconds (30 days) Set the timestamp. Industry requests, support allocation, ground transport planning
Six-month clock Inside the main block only. Now minus the six-month expansion timestamp is above 15,552,000 seconds (180 days) Standard colony seeding, eligible harvester assignment
Local system pass Systems with recent hostile ships exist, or elapsed time is above 10,000 seconds Fleet-occupied systems with recent hostile ships, or all fleet-occupied systems when elapsed time is above 10,000 seconds.

A timer fires only when its interval is exceeded, so an exact match does not fire it. A long elapsed increment alone enters the later local pass, not the main assignment block. A forced call does not bypass the nested monthly and six-month comparisons. Only the standard NPR rule sets the six-month timestamp.

A race with neither ships nor populations returns. Empty fleets are not removed from the race’s fleets when planning starts.

Invocation order and snapshots

Section titled “Invocation order and snapshots”

At entry the planner takes snapshots of the race’s fleets, ships and populations. It builds a one-year hostile contact list, the hostile contacts this race detected that were updated less than 31,536,000 seconds (365 days) ago. From that list it derives recent hostile ship systems (ship contacts under five days old, 432,000 seconds), recent hostile STO populations, known hostile ship systems, other hostile presence systems, the explored neighbours of known hostile ship systems and the safe valued systems. These sets are not rebuilt after each queued order.

The main block runs in this order. Later phases can change orders selected earlier.

Order Action Race qualification or purpose
Before main gate Reset contact counts and threat levels, fleet reservations, each fleet’s avoid danger setting (from DIM_OperationalGroup.AvoidDanger) and redeploy target All participating races
Before main gate Refresh readiness and order retreats from known hostile ship systems Recent hostile ships exist.
Before main gate Exit the disconnected valued network Runs when at least one system has value None
1 Refresh readiness if needed. Zero each population’s reserved fuel cache Main block
2 Standard valued-system deployments, Race-specific.
3 Offensive reinforcement step, run twice the list of hostile presence systems without ships is not empty
4 Scouts, standard diplomacy, threatened jump point coverage Role-specific
5 Unexplored transits and gravitational and geological survey coverage
6 Tanker, local orbital reload, refuel, standard repairs, reload
7 Monthly economic, support and ground deployment work Nested monthly clock
8 Colony seeding and harvester deployment Nested six-month clock
After main gate Rejoin-target rendezvous orders Can clear both rejoining and target fleet orders
Last Occupied-system tactical redeployment, Separate local-pass gate

A fleet that has received a redeployment order carries the redeploy flag (FCT_Fleet.RedeployOrderGiven). The rendezvous step gives each fleet in the same system as its rejoin target a join fleet order and the redeploy flag, after clearing its orders. Each target fleet then clears its own orders and moves to the nearest rejoining fleet.

Each fleet’s reservation flag is cleared on entry. The reservation flag marks a fleet an earlier step has claimed, so later steps of the same pass do not take it. Its redeploy flag (FCT_Fleet.RedeployOrderGiven) is cleared there only for a fleet with no move orders. For a fleet with orders, the redeploy target is refreshed to the destination system of its last move order. Successful routing and local reservation therefore matter to later steps in the same pass. The disconnected-network step can refresh fleet readiness, but the main block can then refresh readiness a second time.

In the construction cycle, population production runs before system value recalculation, mineral transfers and the forced strategic call. Monthly requests chosen by the strategic call were not available to the production decision earlier in that same cycle. Standard research target initialisation runs earlier still. It sets the first research target cost of a standard NPR whose stored target is 0, while NPR AI processing is enabled and the race did not start as pre-industrial or as a conventional NPR.

System valuation and connectivity

Section titled “System valuation and connectivity”

Each NPR keeps a value for every system it knows, stored in FCT_RaceSysSurvey.SystemValue. The values are ordered None 0, Insignificant 1, Low 2, Medium 3, High 4, Extreme 5 and the capital tier 6. Recalculation assigns the capital tier to the race’s own capital system, and that tier gets the largest standard fleet deployment. A stored 6 does not mean the system belongs to an enemy or another race.

The system value recalculation first resets every value other than None to Insignificant, sets the capital system to the capital tier, and scores each presence system (one with an own population or fleet) whose value is not None. The per-system score leaves None and the capital tier unchanged. None therefore survives ordinary valuation, and the capital tier survives per-system scoring. The construction cycle runs this recalculation for standard NPRs .

For standard NPR systems, define P as the summed total population of the race’s populations in the system, D as the development score and L as the logistics installation score. The first matching row wins:

Condition Value
P above 100 Extreme
D above 500 Extreme
P above 25 OR D above 125 OR L above 4 High
P above 5 OR D above 25 OR L above 1 Medium
A suitable existing population exists and the lowest maximum colony cost among them is 0 Medium
P above 0.5 OR D above 0 OR L above 0 OR the lowest suitable maximum colony cost is from 0 up to but not including 1 Low
Otherwise Insignificant

D begins with the summed population mining score, mines plus automated mines plus 10 times civilian mining complexes. When the system holds a race population, D adds the undamaged orbital mining modules of asteroid miners whose fleet is assigned to a population in the system. It also adds 4 times the undamaged terraforming modules of terraformers assigned the same way, rounded. Whether or not the system holds a race population, D also adds the undamaged sorium harvester modules of the harvesters orbiting a gas giant or superjovian in the system that has mineral deposits. The total of those modules is divided by 4 and rounded once, and an exact half goes to the even neighbour. L sums 5 times spaceports, plus refuelling stations, plus ordnance transfer stations, plus maintenance facilities divided by 10. These are remembered population scores plus selected live ship modules, not the summed cost of every local asset. The remembered scores come from the population value refresh.

The colony-cost test uses existing populations whose maximum colony cost is 0 or more, on a body whose gravity is at least the species minimum. This test does not impose the maximum gravity. The P, D and L thresholds are strict. A system at P = 25, D = 125, L = 4 is Medium unless an earlier Extreme condition applies.

After the direct scores, Extreme and capital-tier systems raise explored adjacent systems that are not None to at least Medium. High systems raise them to at least Low. Unstabilised explored links count. The list of significant presence systems (those above Insignificant) is fixed before propagation. The High test then uses the values those systems hold after the Extreme and capital raises. The recalculation then gathers every system above Low and evaluates their adjacent Insignificant neighbours as colonisation candidates.

The adjacent colonisation test raises a candidate to Low when the mineral values of its populations that reach the seed threshold of 6 sum to 25 or more, or when any population’s mineral value is 10 or more. It also raises it to Low when the smallest positive terraforming work score among its populations is below 0.1. It does not survey every planet afresh. It reads the remembered population scores.

Connectivity is a breadth-first walk from the capital through explored adjacent systems whose value is not None, including unstabilised links. Insignificant counts as connected. A valuable enclave beyond a None bridge is disconnected even when its own value is High. Disconnected fleets other than diplomatic fleets (main function Diplomacy) search for a redeployment exit system with valueless destinations allowed, and get the redeploy flag when a route is found. The step marks the disconnected systems only after it has found affected fleets. Explored adjacency itself is described in the jump point network.

Threat assessment and reinforcement

Section titled “Threat assessment and reinforcement”

Threat classification is a precedence chain over contact-derived system sets, not a confidence probability. Each later step overwrites the earlier one:

Applied in order Input Result
Reset All of the race’s systems Clear. Recent and known ship counts zero
Adjacent Explored neighbours of known hostile ship systems Adjacent to hostile system
Other hostile presence One-year population, ground unit, STO, shipyard and ship contacts, excluding systems with ship contacts Hostile presence
Known ships Hostile ship contacts under one year old Hostile ships known. Count of distinct FCT_Contacts.ContactID
Recent ships Hostile ship contacts under five days old Recent hostile ships. Count of distinct ContactID

Contact age is measured from FCT_Contacts.LastUpdate. A ship contact exactly five days old is known but not recent. A contact exactly one year old is excluded. The counts count contact IDs, while the later strength steps count each ship once.

The safe valued systems exclude None systems, known hostile ship systems and their explored neighbours. The name does not guarantee the absence of hostile population or ground presence, because that set is not subtracted.

The planners compare three strength figures. A hostile rating sums FCT_ShipClass.ClassCrossSection x MaxSpeed over hostile noncommercial ships. A fleet’s raw rating sums class Size x MaxSpeed over all its ships. A fleet’s adjusted rating sums the same size x speed product over its ships that are not carried by a mothership. Each ship’s share is divided by 100 for a commercial class, by 2 for a class below size 21, and by 5 when the ship transited less than 300 seconds ago and has not fired since that transit.

The offensive reinforcement step targets the recent hostile ship systems in descending value. When there are none, it uses the hostile presence systems. For each target it sums the hostile rating of recent hostile noncommercial ships. Its contact query filters by detecting race and age but not by target system. The same race-wide recent hostile total is therefore used for each target.

A single local offensive fleet (DIM_OperationalGroup.OffensiveForce) with an adjusted rating strictly above 2 times the hostile total satisfies the target. Otherwise the redeployment step first checks fleets already sent there. A mission capable offensive fleet with the redeploy flag and this target whose raw rating is above the requirement satisfies it. Failing that, it selects an unreserved, mission capable offensive fleet outside the target, without the redeploy flag, whose raw rating is above the requirement of 2 times the hostile total. Ordering is jump distance, then ascending raw rating. The first successful route wins. Adjusted local and raw remote ratings are distinct measures.

When no qualifying reinforcement can route for at least one target, compatible colocated offensive fleets may merge. The candidates are mission capable offensive fleets that are not missile fighter or beam fighter groups, not redeployed and not reserved. The merge step sorts them by descending adjusted rating. It moves the ships of one compatible source fleet into a receiver and stops after one merge. The source must sit at exactly the same coordinates as the receiver, and the operational group compatibility rules of fleet merging decide whether the two fleets can combine. Running the reinforcement step twice can enable a second merge, not a full global consolidation.

Strategic fleet and exploration allocation

Section titled “Strategic fleet and exploration allocation”

Standard valued-system allocation requests one mission capable fleet per listed role, in this order. Fleet roles are the main function of the fleet’s operational group (DIM_OperationalGroup.MainFunction). The functions used are Attack 1, FAC Hunt 6 and FAC 30. Mission capable is the highest fleet mission availability rating, described in NPR fleet readiness.

Tier Requested main function
Capital tier (6) Attack, FAC Hunt, FAC
Extreme Attack, FAC Hunt. The FAC request targets the capital tier again
High Attack, FAC Hunt. The FAC request targets the capital tier again
Medium FAC Hunt. The FAC request targets the capital tier again

The repeated FAC requests are literal. These tiers do not allocate FAC fleets to Extreme, High or Medium targets. Most tier lists are ordered by system ID, while the capital-tier list has no set order.

One shared fleet allocator serves every role. For each target it sets the remaining requirement to the requested count, minus fleets of that role already sent there (redeploy flag set, redeploy target this system, mission availability at least the minimum). It reserves local candidates only when their count can cover the entire remaining requirement. A reserved local fleet gets its standing orders reset from its operational group, and the reservation keeps later steps of the pass from taking it. Otherwise the allocator searches remote, unreserved candidates.

Candidate ordering is jump distance, then descending mission availability, then descending DIM_OperationalGroup.MainFunctionPriority. Source fleets qualify when they sit in the list of systems the role may be drawn from or when their current system’s remaining requirement is below 1. On a successful route the planner sets the redeploy flag, resets standing orders from the operational group, stores the target and decrements the target’s requirement. A failed route falls through to the next fleet. Nothing guarantees that every requirement can be met. The reset standing orders are the operational-group standing orders.

Other mission Target and priority Mechanism
Scout POI Urgent POI waypoints first, then POI waypoints. Descending value, then system ID One Scout fleet per target system
Diplomacy (standard) Nonhostile contacts with communication possible, updated less than three months ago (7,776,000 seconds). Descending value Diplomatic Ship groups only. One per system
Grav survey Gravitational survey not done (FCT_RaceSysSurvey.SurveyDone) and value not None. Descending value Two mission capable fleets per target
Geo survey Default geological survey not done (FCT_RaceSysSurvey.GeoSurveyDefaultDone) and value not None. Descending value Two mission capable fleets per target
Unexplored transit Charted but unexplored jump points, excluding jump points that already have a standard transit order Local eligible Scout or Gravitational Survey fleet first, then the nearest fleet that can route there
Threatened jump point Links toward known hostile ship systems, or explicit transit POI waypoints Existing position or order coverage first, local fleet next, remote route last

The POI, urgent POI and transit POI markers are the waypoint types the NPR creates. A transit requires an existing jump gate or a fleet capable of unstabilised transit. The planner queues travel and then the transit. Successful planning is not discovery already completed. The survey source rules accept a fleet in a system that is surveyed or not None (and the geological equivalent), not one that must be both. Jump Point Defence fleets cover threatened jump points, . Jump point coverage ends the whole pass when an examined target system has no uncovered jump points, so later target systems in that pass may not be examined. The surveys themselves run as described on their own page.

Local combat and target selection

Section titled “Local combat and target selection”

The per-system pass takes a new local subset of the original fleet snapshot. In a system with recent hostile ships it clears move orders and redeploy flags for the mobile military fleets (DIM_OperationalGroup.MobileMilitary) whose mission availability is above unavailable. Otherwise it excludes fleets already redeployed. It sorts local populations by value tier (FCT_Population.AIValue), then mining score, then logistics score. Current local contacts require LastUpdate equal to the current game time. Recent local ships require LastUpdate later than five days ago. The full response order is in NPR tactical behaviour.

Before normal engagement, the pass can order missile-escape transits, using a 60-second window for both reach and threat. A fleet that can reach a charted jump point within 60 seconds, facing salvos that can reach it within 60 seconds, transits the nearest charted jump point when the missiles of the furthest threatening salvo outnumber its point defence capacity against that missile speed. Local reload orders follow, then damage-specific ramming, surrender, counterattack or detachment responses for fleets damaged in the last 120 seconds. These steps consume candidates before the later general target allocator. The mobile defence rating sums the adjusted ratings of the available offensive fleets. The hostile threat rating sums the hostile rating of recent local hostile noncommercial ships.

With exactly one currently detected hostile alien ship, an earlier rule passes the offensive fleets, sorted by ascending adjusted rating, to the overmatch step. That step orders candidates by distance, so the ascending adjusted rating only breaks distance ties. It picks the first fleet whose adjusted rating is above 2 times the target ship’s hostile rating, where a commercial target counts as 0. The nearest scout is also assigned to shadow the ship. The ordinary current-hostile-fleet table below applies only when that earlier rule did not apply, and after the special-race exceptions.

Relative local ratings General decision
Hostile above 2 times mobile Defensive rally or retreat
Hostile above mobile and exactly one recent hostile fleet Defensive rally or retreat
Hostile above mobile and multiple recent hostile fleets Try individual engagements
Hostile at or below mobile Try individual engagements

If no interceptor was assigned, the ordinary rule can request a defensive rally. When offensive fleets remain after some assignments, they instead move toward valued populations they can reach on a clear path, and colocated compatible offensive fleets can merge. Scouts can shadow current hostiles, the nearest free scout for each hostile fleet.

The defensive rally considers populations above the Developing tier (Industrial, Core and Capital) in value order. It requires local static defence fleets (DIM_OperationalGroup.StaticForce) or friendly STO units at the population, and it skips bodies with hostile STOs. Total defence = mobile defence rating + 5,000 x summed static ship cross-section + 100 x summed STO element size, counting each ship in the static defence fleets at the population and each friendly STO element there. A hostile rating at or below 1.5 times that total lets offensive and jump point defence fleets rally there, while the other nonstatic fleets retreat. With no qualifying rally, nonstatic fleets retreat. A retreat targets jump points first and then the system’s populations on bodies with no dust, and each path must stay 100,000,000 km clear of hostile forces. The retreat and population searches keep going after a match, so the last qualifying jump point or population wins. Rally and retreat details are in NPR tactical behaviour.

The fleet-versus-fleet allocator sorts interceptors by descending adjusted rating. For each interceptor it calculates the path distance via Lagrange points to every hostile fleet and a hostile priority = max(0.1, (500,000,000 - distance) / 500,000,000) x hostile fleet adjusted rating, with the distance in km. A nearby weaker hostile within 30,000,000 km (half the 60,000,000 km engagement clearance distance) takes precedence, nearest first. Otherwise unassigned targets below 0.5 times the interceptor’s rating are tried in descending priority. Each must pass the intervening-threat path check over the 60,000,000 km clearance.

When escalation is allowed, the ratio advances 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, with strict comparisons of target below ratio times interceptor. The ordinary rule disables escalation. A final reinforcement search can select an already-assigned target below 2 times the interceptor’s rating, subject to distance and path tests. The chosen target ship is the largest ship in the target fleet. Engagement paths are described in NPR tactical behaviour.

These allocator rules set the redeploy flag and count an assignment after requesting an engagement order, without checking whether the order was accepted. An assignment count therefore does not prove the engagement order was accepted. Other rules, such as the damage responses, set their state only after a successful engagement order.

After combat, local work includes jump point coverage, lifepod rescue, , , carrier following, valued-population patrol and POI routes. Idle fleets can fall back to the best-valued population, , a local harvester or a later patrol. These are ordered heuristics, not simultaneous objective optimisation. The patrol routes are described in NPR patrol routes.

Expansion and economic targets

Section titled “Expansion and economic targets”

The target systems of the six-month standard expansion are the systems above Insignificant whose default geological survey is done (FCT_RaceSysSurvey.GeoSurveyDefaultDone), excluding the explored neighbours of known hostile ship systems. The mineral seed step instead scans all of the race’s surveyed bodies without an own population, minus banned bodies. It does not apply the list of target systems. Even an empty list of target systems is checked only after mineral seeding.

Bodies with a mineral value of 6 or more qualify, the default seed threshold. The planner creates a population record and calculates the colony cost with orbital extremes. When the current cost is not negative, it requests one Deep Space Tracking Station. A worst-case cost from 0 up to but not including 2.5, with gravity inside the species range, requests 50 infrastructure and sets the population to 0.0001. Gravity below the species minimum with a worst-case cost from 0 up to but not including 2 requests 100 infrastructure and the same population. Creating the population record does not itself deliver the requested installations.

Terraforming seeding uses the target systems of the six-month expansion and requires FCT_Race.ColonyDensity above 0. It counts the race’s population records with total population above 0, counting separate populations on the same body separately. It proceeds when that count is at most 1, or when candidate system count / (that count - 1) is above ColonyDensity. Candidate bodies are surveyed bodies without an own population in those target systems, with gravity inside the species range and within the 40 AU apoapsis limit. All candidates with a worst-case cost from 0 up to but not including 1.5 are seeded. Only when none exist does the step choose one fallback body with a cost from 0 up to but not including 2.5. The fallback is ordered by worst-case colony cost, then by descending jump distance from the capital.

The terraforming seed reuses or creates a population, sets it to 0.0001, requests one Deep Space Tracking Station and, when the worst-case cost is above 0, requests 50 infrastructure. It does not queue a colony ship voyage. Separately, the population value refresh can seed suitable empty standard colonies and write colonist source and destination roles. Expansion is therefore not confined to the six-month step. The refresh, survey-time seeds and both removal rules are in Colony seeding and removal.

The mineral value of a body sums a value per deposit. A body with no deposits is worth 0, and gas giants and superjovians are worth 0 here. For every other race, each deposit below 2,000 is ignored. A deposit’s value starts at its accessibility and is multiplied by 0.5 when the amount is below 10,000. When accessibility is above 0.4, amounts above 1,000,000, above 250,000 and above 100,000 get x2, x1.5 and x1.25. Only the highest matching size bonus applies. In the stockpile planning mode, each deposit is then weighted by the capital’s current stockpile of that mineral, testing the rows in this order:

Capital stockpile of the deposit’s mineral Multiplier
Below 1,000 x3
Below 3,000 x2
Below 5,000 x1.5
Above 400,000 x0.1
Above 200,000 x0.25
Above 100,000 x0.5
Otherwise (5,000 to 100,000) x1

A race without a capital counts as holding 0 of every mineral, so every deposit gets x3. The stockpile planning mode is the one the colony seeding and the population value refresh use. The other mode replaces the stockpile weighting with material bonuses, x2 for Duranium and Gallicite and x1.5 for Corundium and Neutronium.

Monthly standard planning removes empty valueless colonies, assigns terraformers and refreshes mining, Deep Space Tracking Station and refuelling station requests. It also assigns orbital miners, recalculates workforce and refreshes construction factory, research lab and ground force construction complex requests. Every race then assigns salvagers. Terraformer targets are populations with a positive terraforming work score, smallest first, excluding None systems. Travel remains subject to the standing order distance limit and routing, as in terraformer fleet assignment. The six-month harvester selection ranks surveyed gas giants and superjovians with Sorium above 50,000 by descending Sorium accessibility.

Ground transport planning runs monthly only when the race has idle (no move orders), mission capable troop transport fleets without the redeploy flag. Systems with recent hostile population contacts (under five days old) exclude the systems with recent hostile ships. The known alien populations in the remaining systems become candidate invasion staging targets. A staging population record is created on the body when the race has none there. The step rebuilds deployment balances, accounts for pending load and unload orders and embarked formations, and uses the design philosophy deployment requirements by population value.

An invasion staging population estimates its defenders as the alien population’s current population (FCT_Population.Population) times 1,000 when no hostile ground unit contact exists. Otherwise the estimate is 100 times the dominant terrain’s fortification modifier (DIM_PlanetaryTerrain.FortificationModifier) times 3 times the contact strength (FCT_Contacts.ContactStrength). This estimate derives template demand. It does not resolve combat. Template counts and pickup and drop routing are in deployment balances. Ground combat execution belongs to ground combat.

Force production and research

Section titled “Force production and research”

The population production tick runs the NPR planners only for NPR races while NPR AI processing is enabled. Standard and Minor starting levels use the trans-Newtonian planner. Conventional NPRs use their own planner. Both planners need a capital population. The trans-Newtonian planner sets fuel production status, refreshes population state, and queues ordnance and fighters when the design philosophy enables ordnance or fighter factories (FCT_DesignPhilosophy.OrdnanceFactories, FighterFactories). All of this runs before its wealth and efficiency gate. Other projects require racial wealth of 2,500 or more and population efficiency of 1 or more. Conventional industry and training require wealth of 1,000 or more and efficiency of 1 or more.

Planning consumer Immediate constraint or priority
Ship class selection Classes that are not obsolete (FCT_ShipClass.Obsolete), not tied to a shipping line and not already a yard’s build class. Needed classes by descending size. Naval and commercial yards are matched separately and the class must fit the yard capacity
Specialist hull demand MaxGeo, MaxGrav, MaxScout, MaxDiplomatic, MaxStabilisation, MaxSalvage and MaxTanker of FCT_DesignPhilosophy, minus existing ships of that automated design and queued builds
Other automated hull demand Target three ships of that automated design in the Reinforcement Group. This fallback does not subtract queued builds
Repair and build task creation A repair candidate comes before new construction. Available slipways, the capital’s mineral stock and the shortfall limit the builds
Industry demand Essential imports minus exports across the race’s populations. Choose the maximum net demand times cargo points. The budget is 2,400 cost. The whole net demand is queued when it costs below 2,400, one unit when a unit costs 2,400 or more, otherwise 2,400 divided by the unit cost, rounded down

Strategic fleet demand is therefore not translated into “build enough adjusted strength for each hostile system.” Deployment uses existing operational roles. Construction uses design philosophy limits and reinforcement stocks. Monthly installation requests connect colony plans to production and civilian delivery. Factories do not appear the moment a request is written. The yard side is described in build class assignment.

Without an essential deficit, trans-Newtonian industry tries fallback targets in this order. They are Deep Space Tracking Stations, construction factories below 100, philosophy-enabled ordnance and fighter ratios, research labs, ground force construction complexes, logistics, automated mines, shipyards above 300 industry, terraforming, academy, more factories and financial centres. The first target that queues a project ends the step. Each ratio, amount and mineral clamp is listed in the trans-Newtonian installation queue.

Research lab redistribution skips None systems and clears the previous lab request. A negative manufacturing worker balance requests the export of one whole lab when the population has at least one, plus its fractional part. Otherwise, with labs present, it requests an import of the fractional part, plus one when the worker balance is above 5, marked nonessential. A population with zero labs gets no import request. This is workforce-driven redistribution, separate from research technology selection.

NPR research takes the first uncompleted design philosophy progression by ProgressionOrder (DIM_DesignPhilosophyTechProgression). The cost is the next technology’s FCT_TechSystem.DevelopCost or a group-specific summed cost. The propulsion group adds the next power plant, engine and fuel consumption technologies, plus the maximum engine thrust modifier when the design philosophy’s missile strategy (FCT_DesignPhilosophy.NPRMissileStrategy) is none (0). These are scripted progression goals, not combat-target scoring. The progression list is described in ordered technology progression.

Research points per population = research output per facility x whole research capacity x multiplier x elapsed years. The multiplier is the larger of two figures. The first is the research bonus of the race’s best scientist specialised in the progression’s research field (ResearchField) times 4 minus 3, or 1 without such a scientist. The second is the best research bonus among all the race’s scientists, or 1 without one. Points accumulate in FCT_Race.CurrentResearchTotal. Once the total reaches the target cost (FCT_Race.ResearchTargetCost), one progression completes, the cost is subtracted and the next target cost is computed. There is no catch-up. One population’s research completes at most one step, and later populations can complete further steps. Propulsion completion refreshes the automated classes. Armour technology completion rebuilds the ground units. Research runs only for populations with total population above 0 and a whole research capacity above 0. Scientist bonuses are described in research and scientific personnel.

Special races and diplomatic boundary

Section titled “Special races and diplomatic boundary”
Race or layer Distinct behaviour on these paths
The common population and development tests then run and replace that result
Hives with FCT_Ship.BioEnergy above 250. A full hive other than Very Large refits, Small to Large and Medium or Large to Medium. Local composition and bioenergy choose new hulls
The ordinary scout, survey and recovery steps differ or are skipped
Skip the early ordinary retreat. Combat in a system with no own population allows strength-ratio escalation. Idle fleets can guard abandoned factories
Summed EM signature of all populations in the system above 20,000 Extreme, above 5,000 High, above 1,000 Medium. A second test for above 1,000 giving Low can never be reached, so everything else is Insignificant. Returns before the common scoring
Strategic planning and population production return at once

The common tests that follow give Extreme when P is above 100 or D is above 500. The asset score tier therefore does not survive the recalculation.

The construction cycle recalculates values only for standard NPRs . Special generation and the special-race technology catalogues are in Special race lifecycles.

Standard territorial diplomacy runs before the forced strategic call in the construction cycle. It reads the current territorial presence contacts, sovereignty and protection status, system value, xenophobia and connectivity loss. After conceding a system it can set that system’s value to None, change its controlling race (FCT_RaceSysSurvey.ControlRaceID) and transfer populations. The connected strategic value sums the squared values of connected systems above Insignificant, Low 4, Medium 9, High 16, Extreme 25 and the capital tier 36. Temporarily setting a system to None measures the lost connected value, and the loss divided by 4 feeds the decision. A system below Low whose removal loses nothing is conceded at once. This feeds the future availability of network paths and missions. Claims and warnings are described in territorial claims, and the transferred populations take the Occupied status of the political status catalogue.

Its relative power comparison uses remembered alien mobile military tonnage, the class sizes of known alien ships of that race that are not destroyed, have a speed above 1, are not commercial and were seen within five years. A race with no such tonnage counts as 0. Alien tonnage is multiplied by 1.5 unless the NPR’s own capital system is also the alien race’s capital system, then divided by the race’s own nominal military tonnage (own noncommercial ships with a class speed above 1). It is then multiplied by the ratio of the highest weapon technology DevelopCost and of the highest engine technology DevelopCost, alien over own, each only when the alien figure is above 0. This is a separate diplomatic comparison from the fleet cross-section times speed ratings. The alien weapon technology scan covers known alien classes of every race, whereas the engine scan covers only the contacted race. The full diplomacy thresholds and communication outcomes are not reproduced on this page. They are in intelligence mechanics.

Persistence and transient decisions

Section titled “Persistence and transient decisions”
State What the game stores
System value FCT_RaceSysSurvey.SystemValue, restored to each system’s planning state when a game loads
Research target cost and research total FCT_Race.ResearchTargetCost and FCT_Race.CurrentResearchTotal, saved and loaded
Completed design philosophy progressions FCT_TechProgressionRace rows, loaded back into progression membership
Fleet operational group and redeploy flag FCT_Fleet.NPROperationalGroupID and FCT_Fleet.RedeployOrderGiven, restored with each fleet’s planning state
Monthly, threat-response and six-month timestamps Not saved by the game. A loaded game gives each NPR a new race planner with all three at zero
Threat levels and counts, mobile and hostile ratings Rebuilt or reset by the strategic and local passes
Reservations, remaining fleet requirements, remembered redeploy targets Runtime planning state refreshed or rewritten by the assignment steps
Colony seeds, installation requests, production and movement tasks Changes to the normal game records. These durable outcomes differ from the planners’ temporary lists

A reload therefore preserves system values, research progress and queued work, but it does not restore every planner cooldown and remembered reservation. Rebuilt state can cause an eligible periodic pass to run earlier after loading. The saved colony, installation request, shipyard and ground formation fields are listed under NPR colony and economic AI persistence.

Worked examples

Section titled “Worked examples”
Synthetic inputs Literal outcome
Not forced, elapsed 15,000 seconds, no recent hostile ships, monthly clock overdue The main gate stays closed. The later local pass can run in occupied systems. Monthly work waits for a main-gate entry
Standard P = 25, D = 125, L = 4, no zero-cost colony Medium. The strict High thresholds all fail. P = 25.01 gives High
Capital, then an Insignificant bridge, then a High enclave, explored links All connected. Set the bridge to None and the enclave falls outside the capital’s valued network despite its High value
Two survey fleets needed, the target holds one free local fleet The local count rule does not reserve that lone fleet. The general search handles the remaining demand with its own rules for where a fleet may come from and which fleets are free
Two threatened systems, recent hostile warships rate 40 and 60 in separate systems The reinforcement step sees a total of 100 for each target because it does not filter by system. The threshold for one sufficient fleet is above 200
Interceptor 100, hostile 50, no nearby shortcut, ordinary ratio 0.5 The strict below-50 test rejects exactly 50. Escalation to 0.6 can admit it only where escalation is enabled and the path is safe
No target systems for the six-month expansion, surveyed unowned mineral body with value 6 Mineral population creation can still occur before that check. Terraforming seeding does not proceed
Deposit amount 300,000, accessibility 0.8, capital holds 2,000 of its mineral Base 0.8 x 1.5 = 1.2. The planning-mode stock multiplier x2 gives 2.4. Three such qualifying deposits total 7.2, above the default seed threshold of 6
Terraforming costs 1.0 in two eligible systems Both are seeded. The single-body fallback is bypassed
Research generation 120, target 100, next target 10 One progression completes and 20 remains. The research step does not complete the next progression with it at once

Concept map

Section titled “Concept map”

Resources

Section titled “Resources”

Related community documentation and developer posts. These may describe a different game version.