diff --git a/docs/ROADMAP.md b/docs/ROADMAP.md index 7e39b4c..b66c0a4 100644 --- a/docs/ROADMAP.md +++ b/docs/ROADMAP.md @@ -417,15 +417,176 @@ channel if its ADR documents the limits and preserves transport-neutral semantic No renderer is needed for server tests; no client may decide energy, mutation, reproduction or death. Steam and final player UI remain later product work. -## Later gates +## M10-M14 delivery gates -| Milestone | Deliverable and acceptance gate | -|---|---| -| M10 Realtime multiplayer | Authenticated intentions, reliable outcomes, prediction/reconciliation/interpolation and interest; 20+ bot clients then measured scaling/fault tests | -| M11 Cellular gameplay | Movement, feeding, predation, mutation selection, reproduction, descendants, total restart; playtest FUN and repeated-life variety | -| M12 Persistent ecology | Days-long unattended worlds, genealogy, extinction evidence, player absence and diverse strategies; analyze collapse/runaway lineages | -| M13 3D R&D | Isolated smooth meshing, body graph, procedural skin/physics/locomotion comparisons with measured exit criteria | -| M14 Multicellular transition | Ratified biological transition and first cell-to-multicellular-to-3D slice with conserved ecological coupling | +Status: **planned**. These gates extend the founding milestones; they do not mark +features implemented or ratify unresolved game rules. Each row uses the same +objective, boundary, dependency, acceptance, evidence and risk contract as M5-M9. +The shared gate evidence above applies, including version/migration review and +updates to STATUS and VALIDATION. Split implementation rows into contract and +implementation reviews when needed. Module paths are proposed, not instructions +to create empty frameworks. New dependencies require scoped ADRs and evidence. + +```mermaid +flowchart LR + M9[M9 native engineering client] --> M10[M10 authoritative multiplayer] + M10 --> M11[M11 playable cellular loop] + M11 --> M12[M12 persistent ecology] + F[M4.6 fidelity follow-up] --> Q[M12 qualified tier policy] + Q --> M12 + M11 --> P[Cellular playability review] + M12 --> P + P --> M13[M13 isolated 3D research] + M13 --> M14[M14 multicellular slice] + M12 --> M14 + W[M8 independent durability and promotion] -. required for World0 launch .-> L[Live World0] + M12 -. product readiness evidence .-> L +``` + +Dev/stage experiments can proceed while World0 launch is blocked by ADR-0075. +Passing a later feature gate does not lift that block or authorize promotion. +Bounded active-only tests can precede aggregate qualification; live automatic +aggregation cannot. M13 starts only after the cellular product is judged compelling, +not merely after its infrastructure tests pass. M14 is a bounded transition slice, +not a commitment to every body plan, planetary solver or locomotion mode. + +## M10 - Real-time multiplayer cell simulation + +**Goal:** responsive, authenticated, server-authoritative cellular sessions with +bounded replication and recovery under loss, overload and region transfer. + +**Entry:** M9 client/session evidence and M7/M8 dev/stage durability. Read +NETWORK_ARCHITECTURE, PERSISTENCE_ARCHITECTURE and ADR-0017. The M9 transport is +not automatically the production selection; proof tick and region sizes remain +scoped experiments until measured here. + +| ID / objective | Files/modules | Dependencies | Completion criteria | Tests / evidence | Risk | +|---|---|---|---|---|---| +| 10.1 Ratify protocol, timing and workload budgets | docs/networking, docs/adr, schemas/network | M9, M7 authority contracts | Define input scheduling window, tick/snapshot cadence, sequence wrap, session/epoch identity, baseline lifetime, queue/byte/work limits and target latency on named hardware | Trace-based timing experiments and explicit exact/tolerant determinism contract | Proof constants silently become game rules | +| 10.2 Select secure transport and implement bounded channels | src/network/transport, dependency manifests, CMake | 10.1, ADR-0017 evaluation | Compare maintained secure datagram/reliability options; pin selected library/crypto with license/maintenance evidence; separate superseding state from reliable actions; define MTU, congestion and timeout behavior | Loss/reorder/duplication/fragmentation and head-of-line comparison; codec fuzz, platform builds, malformed peer tests | Custom cryptography or retransmission amplification | +| 10.3 Bind sessions to accounts and control authority | src/server/session, src/client/session, backend identity adapter | 10.2, M9 dev identity boundary | Short-lived scoped grants, authenticated reconnect and unique organism control; provider-independent account IDs; dev identities cannot authenticate to production | Expiry, replay, foreign world/region, duplicate login, stolen stale grant, rate limiting and credential redaction | Two sessions control one life or test auth escapes dev | +| 10.4 Admit intentions and return durable outcomes | src/network/protocol, src/server admission, schemas/network | 10.1-10.3, M7 batch outcomes | Validate tick/sequence/ranges/targets under current authority; receipt, simulated-input and durable-outcome ACKs are distinct; stable action IDs resolve retries and unknown outcomes | Fabricated position/energy, future inputs, stale epochs, repeated action, lost ACK and crash before/after commit | An input ACK is mistaken for final biological success | +| 10.5 Replicate bounded snapshots and spatial interest | src/network/replication, src/network/interest, client views | 10.4, M6 views | Versioned baselines/deltas, entity removal and fresh-baseline recovery; spatial filter plus perception extension boundary; transmit no unnecessary hidden state or controller labels | Baseline eviction, sequence wrap, dense interest sets, information-leak assertions, bandwidth/memory report | Filtering only in the renderer leaks concealed organisms | +| 10.6 Predict and reconcile local movement | src/client/prediction, shared movement adapter | 10.4-10.5, movement experiment ADR | Replay bounded pending inputs from authoritative ACK state; smooth presentation separately; no predicted reproduction, mutation, energy grant or final death | Latency/jitter/loss traces, correction error distribution, rejected movement and restart resync | Prediction becomes authoritative or an unbounded replay queue | +| 10.7 Interpolate remote views and handle session loss | src/client/interpolation, src/client/session | 10.5-10.6 | Bounded interpolation/extrapolation, stale-view presentation and fresh reconnect baseline; no invented remote outcomes; disconnect stops admission without granting offline progression | Long gaps, reordered snapshots, death/removal, clock drift, reconnect to newer epoch | Extrapolation displays dead or nonexistent organisms as actionable | +| 10.8 Transfer authority between two regions | src/server/handoff, persistence transfer records, src/network/routing | 10.3-10.7, persistence handoff ADR ratified | Non-authoritative ghost, frozen source, durable staged destination and one committed authority switch; route session without a loading screen; in-flight inventories and retries recover once | Crash/partition at each transition, stale owner, duplicate transfer, crossing back, failed destination; unique owner and conservation checks | Region crossing duplicates inventory or leaves two active copies | +| 10.9 Qualify multiplayer load and abuse resistance | tests/network, tests/load, tools/network traces, docs/engineering | 10.1-10.8 | At least 20 simultaneous simulated clients plus native-client observation; then ramp beyond 20 to measured saturation; declare sustainable workload and overload response | Packet fuzz, floods, slow peers, lease loss, disconnect storms; tick/RTT/correction/bandwidth/CPU/memory distributions and sanitizer evidence | Bot throughput hides poor responsiveness or starvation | + +**Exit:** the native client remains responsive against an authoritative server; +20+ clients and a two-region handoff pass reproducible network/fault scenarios. +All queues and per-tick work are bounded, rejected inputs cannot alter biology, +and final outcomes survive the declared dev/stage failure model. This does not +claim Internet-scale capacity, Steam readiness or a complete cellular game. + +## M11 - Actual cellular gameplay + +**Goal:** the first product judged primarily on FUN: repeated primitive-cell lives +with meaningful feeding, predation, mutation, reproduction and absolute death. + +**Entry:** M10 authority and responsiveness; M3 genotype/development/phenotype and +M6 accounting contracts. Read GENOME_MODEL, ENERGY_MODEL and REPRODUCTION_MODEL. +Synthetic fixtures are test oracles, not accepted gameplay balance. Ratify the +offline lifecycle before enabling persistent player lives. + +| ID / objective | Files/modules | Dependencies | Completion criteria | Tests / evidence | Risk | +|---|---|---|---|---|---| +| 11.1 Ratify cellular life and design decisions | docs/biology, docs/networking, docs/adr | M10, founding requirements | Decide bounded first-loop movement/feeding/reproduction rules, mutation timing/cost and heredity; resolve reproduction-cost semantics from ADR-0031/0032; document offline freeze/resume fairness and friend recognition or explicit social-feature deferral | State diagrams, adversarial logout scenarios, worked energy/matter budgets, review against locked vision | Fixture behavior or convenience invents permanent game rules | +| 11.2 Create primitive lives and persistent control | src/server/lifecycle, persistence life records, client session | 11.1, 10.3-10.4 | Every new life uses the same primitive founder rules with funded inventory/provenance; stable account/life separation; life state closes over timers, reserves and retries | Old/new account equivalence, concurrent spawn, failed allocation, restart/retry; no inherited biological power | Historical account data grants a stronger start | +| 11.3 Connect locomotion, feeding and predation | src/biology adapters, simulation interaction, client input | 11.1-11.2, M6 transfers | Server resolves spatial contact, perception, motion and intake; phenotype-derived maintenance/locomotion/defense costs; finite resources and corpse transfers | Range/speed cheats, contention, starvation/predation, closed-box conservation; movement playtest traces | Attractive motion hides free energy or arbitrary RPG modifiers | +| 11.4 Apply earned mutation choices through development | src/biology/mutation, development, client mutation UI | 11.1, 11.3, M3 candidates/archive | Reproducible contextual offers; stale selections rejected; prerequisites and funded development enforced; immutable genome ancestry; no visible linear class tree | Same context/seed offers, stale offer after resource change, failure rollback, restart during choice, multi-system tradeoffs | Mutation UI bypasses costs or implies a guaranteed progression | +| 11.5 Reproduce funded autonomous descendants | src/biology/reproduction, src/server wildlife controllers, archive | 11.1, 11.3-11.4 | One bounded accepted reproductive strategy; parent investment funds descendants and records ancestry durably; descendants use shared biological intentions, are unowned wildlife and survive parent death/logout | Insufficient resources, duplicate request/ACK loss, birth/death races, deterministic bounded AI, genetic vulnerability comparisons | Free offspring, player ownership, or special NPC biological rules | +| 11.6 Commit permadeath and complete restart | src/server/lifecycle, persistence history, client death flow | 11.2-11.5 | Death terminal across reconnect/recovery; corpse inventory preserved; next life is primitive; existing descendants/history remain; no resurrection or corpse-retrieval shortcut | Death versus mutation/birth/logout races, crash at commit, old session replay, repeated-life independence | Recovery or UI restart resurrects a dead life | +| 11.7 Freeze offline organisms and resume uniquely | src/server/lifecycle, persistence suspended state, client session | 11.1-11.6 | Implement ratified detectable-disconnect/logout boundary; no movement, feeding, growth, mutation, reproduction or timer progression after freeze; suspended inventory accounted separately; safe unique resume | Long offline interval, repeated disconnect, suspended worker recovery, death at freeze boundary, duplicate resume and wildlife continuation | Logout becomes an escape exploit or offline farm | +| 11.8 Deliver readable cellular gameplay UI | src/client/ui, organism/environment presentation, client packaging | 11.3-11.7, M9 | Energy/resource needs, mutation tradeoffs and death causes understandable; shared phenotype visuals for players/NPCs; separate authorized friend mechanism only if ratified; debug tools separated | First-session observation, input/accessibility checks, phenotype visual parity, sensory information-leak tests | Engineering UI or controller labels become the player experience | +| 11.9 Run repeated-life playtests and close the loop | tools/playtest telemetry, docs/engineering/playtests, tests/gameplay | 11.2-11.8 | Predeclare playtest questions and participant/session sample; assess movement, food decisions, predation, tradeoffs, descendants and variety over multiple lives; record issues and retest changes | Human feedback plus bounded consent-aware telemetry, death causes/build diversity, full-loop replay and crash regression | Engineering success mistaken for fun or tuning to one seed | + +**Exit:** players can complete repeated lives from primitive spawn through feeding, +reproduction/mutation and permanent death. Descendants remain wildlife and offline +players do not progress. A recorded product review accepts the cellular loop or +keeps iteration open; no invented retention percentage substitutes for that review. +M12 supplies sustained world ecology rather than short-session evidence alone. + +## M12 - Persistent ecology + +**Goal:** player-derived lineages and ecological consequences survive player absence +for days, with defensible aggregation, ancestry and extinction records. + +**Entry:** M11 playable loop and durable wildlife; M4.6's failed fidelity cases must +be resolved before automatic tier selection. Read ECOLOGY_MODEL and the aggregate +budget report. Neither conservation alone nor exact replay proves ecological fidelity. + +| ID / objective | Files/modules | Dependencies | Completion criteria | Tests / evidence | Risk | +|---|---|---|---|---|---| +| 12.1 Ratify ecological identity and time policies | docs/biology, docs/adr, schemas/ecology | M11, M4.6 evidence | Define lineage/species distinction, extinction evidence, viable reserves, outage catch-up and uncertainty; decide NPC mutation/heredity and whether behavior can evolve, including explicit deferral | Worked histories with dormant cells/eggs/spores/transfers; no fixed plant/animal classes or mandatory intelligence | Classification becomes a class tree or absence is called extinction | +| 12.2 Qualify a population model and tier policy | experiments/resolution, src/ecology, docs/engineering | 12.1, M4.6 follow-up, M11 rules | Repair or replace unqualified model; paired active/cohort ensembles meet declared demographic/biomass/diversity error bounds on target host; document unsupported states and active fallback budgets | Reproduce known false extinctions, hold-out seeds/regimes, interval/bin sensitivity, cost/error curves | Faster aggregation silently destroys otherwise viable lineages | +| 12.3 Persist automatic tier transitions | src/ecology/conversion, region scheduling, persistence codecs | 12.2, M7 recovery | Atomic collapse/materialize with deterministic committed samples, one spendable representation, retained genome identity, demographic state and residuals; controlled/suspended organisms excluded | Crashes at conversion boundaries, oscillation/cherry-picking, no healing/age reset, inventory and ancestry audits | LOD switching creates resources or fictitious exact parentage | +| 12.4 Schedule unattended ecology and migration | src/server/ecology_worker, src/ecology/flux, persistence transfers | 12.1-12.3, 10.8 | Owned bounded intervals, backlog/freshness reporting and idempotent cross-region escrow; regional aggregates only where fidelity qualifies; no silent skipped intervals or ecological culling on overload | Player-free advance, worker loss, delayed boundary data, partitioned migration, overload/catch-up; count in-transit stock | Background replay spends an interval twice or timeout duplicates migrants | +| 12.5 Couple ecological and environmental budgets | src/ecology, src/world/resources, regional fields | 12.2-12.4, M6 | Producers/consumers/decay share finite resources; scoped regional atmosphere/water exchanges use conserved inventories, explicit boundaries and versioned coefficients | Closed/open budgets, nutrient exhaustion, oxygen/carbon response, slow/fast interval comparisons | Implicit infinite nutrients or unsupported global climate realism | +| 12.6 Retain genealogy and historical queries | src/biology/archive, backend history, migrations, client codex | 12.1, 12.3-12.4 | Parent/child genomes and lineage divergence retained after extinction; population history queryable; statistical ancestry distinguished from known individual ancestry; account history grants no power | Restore archived extinct genomes, common-ancestor fixtures, idempotent events, bounded large-tree queries | Compaction discards history or invented genealogy appears precise | +| 12.7 Reconcile extinction from complete world evidence | src/ecology/reconciliation, backend history, persistence watermarks | 12.1, 12.4, 12.6 | Durable watermark covers active/cohort, dormant/suspended, viable reproductive and in-flight stores; incomplete evidence reports unknown/absent; terminal extinction follows accepted semantics | Last-member migration, disconnected worker, dormant survivor, residual viable stock, late duplicate event and restored world | One empty region falsely declares a lineage globally extinct | +| 12.8 Soak diverse worlds and close ecological risks | tools/soak, tests/ecosystem, deploy/workflows, docs/engineering | 12.2-12.7, M8 clone/restore | Run reproducible multi-seed worlds unattended for days with real worker restarts; accelerated runs supplement wall-time evidence; measure diversity, monopolies, collapse, cycles, ancestry growth and costs | Seed/regime matrix, no-player versus returning-player scenarios, full conservation/retention audit, clone rule-change comparison | A stable but biologically empty or monopolized world passes | +| 12.9 Review cellular product and expansion readiness | docs/STATUS, docs/engineering/playtests, docs/ROADMAP | 11.9, 12.8 | Returning players encounter changed player-derived ecology; record remaining exploit/fidelity/cost limits; explicit go/no-go for 3D R&D and separate World0 promotion review | Repeated-session playtests and reproducible soak evidence; confirm M8 World0 block remains unless independently lifted | 3D work displaces unresolved core gameplay and ecology | + +**Exit:** unattended ecology, player absence, worker recovery and historical queries +are demonstrated together. Extinction follows reconciled biological evidence; +lineages never disappear merely because their player died. Capacity limits cause +documented admission/backpressure, not fabricated natural selection. World0 launch +still requires the independent durability and promotion evidence from M8. + +## M13 - 3D technology R&D + +**Goal:** choose feasible 3D techniques through isolated, measured prototypes before +integrating them into the persistent cellular product. + +**Entry:** 12.9 accepts cellular playability/ecology and authorizes the research scope. +Read WORLD_ARCHITECTURE and GENOME_MODEL. Every experiment declares workloads, +budgets and go/no-go criteria before comparison. Prototype output is disposable; +accepted formats/algorithms require ADRs. No planet-wide dense allocation. + +| ID / objective | Files/modules | Dependencies | Completion criteria | Tests / evidence | Risk | +|---|---|---|---|---|---| +| 13.1 Bound research and evaluation workloads | docs/adr, experiments/3d, docs/engineering | 12.9, M5/M6 constraints | Define representative terrain edits, body graphs, habitats and locomotion families; target hardware, accuracy/frame/CPU/memory budgets and experiment exit criteria | Review matrix with comparable inputs, reference outputs and headless/client boundaries | Attractive demos set architecture without representative measurements | +| 13.2 Compare planet coordinates and sparse LOD | experiments/planet, src/world contracts | 13.1, M5 coordinates | Compare spherical indexing strategies; region-local precision, seam/deep-edit addressing and sparse refinement with inventory partition rules | Pole/seam/address round trips, deep-to-core sampled edits, refine/coarsen conservation, bounded residency | Global float error, seam cracks or detailed resources created by refinement | +| 13.3 Compare smooth surface extraction | experiments/meshing, docs/adr | 13.1-13.2, M5 codecs | Compare Marching Cubes, Surface Nets and Dual Contouring on shared datasets; assess topology, edits, sharp/smooth features, LOD seams and collision derivation before selection | Thin surfaces/caves, neighbor revisions, dirty rebuild latency, mesh memory and visual/collision checks | Selecting a mesher from appearance alone | +| 13.4 Prototype 3D rendering and streamed caches | experiments/render3d, src/render boundary | 13.2-13.3, M9 renderer | Bounded immutable-view streaming, origin shifts and stale mesh discard; meshes/collision caches regenerable from versioned source; headless simulation remains independent | Camera travel, cache pressure, device failure, terrain edits and frame-time distributions | Renderer owns world state or stale geometry becomes authoritative | +| 13.5 Define developmental body graphs and constraints | experiments/body_graph, docs/biology, schemas proposals | 13.1, M3 development | Typed tissues/organs/attachments/joints/vascular/neural dependencies; genome limits legal structures; funded construction derives mass, maintenance and vulnerable structures | Invalid/disconnected/cyclic graphs as appropriate, symmetry/range tests, material budgets, canonical graph round trips | Arbitrary mesh editor bypasses evolved developmental capability | +| 13.6 Derive meshes and physics from phenotype | experiments/organism_mesh, experiments/physics, docs/adr | 13.5, 13.4 for visuals | Compare specialized physics/library options for the scoped bodies; derive skin, collisions, mass/inertia and damage mapping from one phenotype; document determinism tolerances | Diverse/sessile/mobile graphs, degenerate bodies, joint stability, damage consistency and CPU/memory comparison | Rendered anatomy and simulated mass disagree | +| 13.7 Prototype procedural locomotion under damage | experiments/locomotion, body/physics adapters | 13.5-13.6 | Compare feasible crawling/swimming/walking techniques on evolved bodies; costed propulsion and center-of-mass control; scoped limb-loss compensation; record unsupported flight/other modes | Shape/limb-count sweep, terrain contact, missing limb, solver stability and energy/force bounds | Hand-authored animation or free motion becomes the foundational solution | +| 13.8 Prototype conservative micro/macro exchange | experiments/ecological_coupling, docs/world, schemas proposals | 13.2, 12.3-12.5 | Explicit habitat depth/units and disjoint owned stocks; microbial biomass funds macro food density and corpse transfers fund micro nutrients; durable idempotent transfer design | Closed-box bidirectional cycle, concurrent consumption, delayed exchange, restart/retry and LOD oscillation | Same biomass is edible independently in both domains | +| 13.9 Select the bounded integration path | docs/adr, docs/engineering, dependency manifests | 13.2-13.8 | Publish comparison evidence and accepted/deferred choices; pin only adopted dependencies; document migrations and M14 body/habitat limits; reject failed experiments explicitly | Reproducible benchmark corpus, headless regression, review against predeclared budgets | Research prototypes silently become production systems | + +**Exit:** measured choices support one constrained multicellular slice. Unsupported +locomotion, planet solvers and morphology remain documented R&D, not placeholders +in the live simulation. Full hydrology, atmosphere, flight and arbitrary creatures +are not prerequisites unless the selected slice explicitly needs them. + +## M14 - Multicellular transition + +**Goal:** one persistent life can develop from a primitive cell into a funded +multicellular organism and enter a bounded 3D habitat without losing authority, +history or conservation. Remaining unicellular must remain a viable strategy. + +**Entry:** M12 persistent cellular ecology and M13 accepted prototypes. Ratify the +transition criteria; no automatic stage ladder, account-age shortcut or intelligence +requirement. Existing worlds require clone/migration/ecological comparison before +the new rules can be promoted. + +| ID / objective | Files/modules | Dependencies | Completion criteria | Tests / evidence | Risk | +|---|---|---|---|---|---| +| 14.1 Ratify transition biology and life identity | docs/biology, docs/world, docs/adr | 12.9, 13.9 | Decide physiological/developmental/environmental prerequisites, investment, timing, failure behavior and continuity of organism identity; distinguish growth from reproduction; define bounded supported body/habitat scope | Worked conservation/ancestry examples, cellular alternatives, failure/death/logout scenarios | Multicellularity becomes a mandatory level checkpoint | +| 14.2 Implement funded multicellular development | src/biology/development, phenotype/body graph, persistence codecs | 14.1, 13.5-13.6 | Validated genome develops supported tissues/body graph; construction, maintenance and growth consume real budgets; developmental intermediates persist and remain vulnerable by ratified rules | Insufficient material/oxygen/energy, invalid anatomy, deterministic plan/replay, partial growth restore, mutation incompatibility | Transition grants a free mature body | +| 14.3 Implement atomic domain transfer | src/server/domain_transition, persistence transfer records, schemas | 14.1-14.2, 10.8, 13.8 | Source freezes and escrows complete life state; destination stages valid 3D state; one durable authority transfer preserves inventory, genome, ancestry, timers and action history; no literal cross-scale physics | Crash/partition at every phase, repeated transition, capacity rejection, stale epoch, death/logout race, rollback only before committed switch | Two living copies, lost life or new attempt duplicates resources | +| 14.4 Integrate a bounded authoritative 3D habitat | src/world, src/server region adapter, src/network views | 14.3, 13.2-13.4, M10 | Stream sparse terrain and accepted collision/physics; validate movement and interactions on server; maintain epoch/perception/baseline semantics with bounded interest and physics cost | Seam crossing, stale collision revision, terrain edit, speed/range abuse, resource pressure and restart | 3D client or disposable mesh becomes authoritative | +| 14.5 Activate conservative cross-domain ecology | src/ecology coupling, src/world/resources, persistence journals | 14.3-14.4, 13.8, 12.4-12.5 | Disjoint microbial/population/macro stocks with explicit transfer intervals and residuals; producers fund food density, consumption debits it, death/decomposition returns nutrients | Repeated closed-cycle budgets, simultaneous consumers, stale fields, delayed transfers, recovery with escrow | Ecological bridge creates biomass or couples incompatible timesteps | +| 14.6 Expose genome-constrained morphology and motion | src/client morphology UI, render organism, procedural locomotion | 14.2, 14.4, 13.5-13.7 | Player edits only developable structures; server validates/funds changes; renderer and locomotion derive from phenotype; preview is non-authoritative; no mandatory humanoid/intelligent form | Forged edit, unavailable tissue, unaffordable design, limb loss, alternate valid morphology, visual/physics parity | Cosmetic editing changes biology for free | +| 14.7 Preserve life, offspring and offline invariants in 3D | src/server/lifecycle, biology/reproduction, history | 14.3-14.6, M11/M12 | Ratified reproduction yields autonomous funded descendants; player death is permanent and next life primitive; suspension freezes every 3D biological timer; genealogy spans domains without invented parentage | Death during transfer/growth, reconnect after death, offspring persist, offline gestation/growth freeze, same-account fresh start | 3D arrival becomes a checkpoint or inherited power | +| 14.8 Qualify save formats and ecosystem upgrades | migrations, tools/world-clone, tests/recovery, docs/engineering | 14.2-14.7, M8 upgrade gates | Version complete terrain/body/coupling/life state; clone old world, migrate, accelerate and compare; retain required generators/builds; failed migration leaves source intact; no mixed adjacent physical laws | Old/unsupported format refusal, corruption, transfer-in-flight restore, full history audit, ecological pre/post distributions | Valid schema migration catastrophically changes the living world | +| 14.9 Playtest and accept the first transition slice | tests/end_to_end, tools/playtest, docs/STATUS, docs/engineering | 14.4-14.8 | Native client demonstrates primitive cell → funded multicellular body → 3D habitat → death → primitive restart; cellular players/descendants continue in same ecological world; measured fun, clarity and costs | Multiple seeds/lives/body variants, uninterrupted versus recovered runs, human playtests and scoped load profile | One scripted transformation is mistaken for an evolvable persistent system | + +**Exit:** the integrated transition is playable, conserves resources, survives +recovery and preserves absolute permadeath. The result supports only its declared +morphologies, habitats and load envelope. Further expansion requires measured +gates; neither civilization nor a predetermined species ladder is implied. No date or production capacity is promised without staffing and benchmark data. Every release records simulation/world/network versions; biological upgrades use