← Archive
lm-003134 · 2026-08

世界域管理器:世界生成、資源分配、停止與晉升

下載 MD 檔 ⬇

世界域管理器:世界生成、資源分配、停止與晉升

World-Domain Governor: Admission, Resource Allocation, Termination, and Promotion of Runnable Worlds

Branching World Computation / World-Domain Cognitive Runtime
分支世界計算/世界域認知 Runtime 系列
WDC-03 / BWC-03 — Governance Paper I

作者:Neo.K(許筌崴)
協作形式化:Aletheia
機構:一言諾科技有限公司(EveMissLab)
日期:2026-08-17
版本:v0.1
狀態:world-governance / scheduling / lifecycle-control formalization


Canonical Non-Identity Statement

WDC-01 建立:

Future CandidateRunnable World.\boxed{ \text{Future Candidate} \rightarrow \text{Runnable World}. }

WDC-02 建立:

Runnable WorldBranching World Graph.\boxed{ \text{Runnable World} \rightarrow \text{Branching World Graph}. }

本文進一步建立:

World GenerationWorld Governance.\boxed{ \text{World Generation} \neq \text{World Governance}. }

以及:

World PriorityWorld TruthWorld Moral Worth.\boxed{ \text{World Priority} \neq \text{World Truth} \neq \text{World Moral Worth}. }

本文中的 Governor 管理的是:

computation over worlds,\boxed{ \text{computation over worlds}, }

不是:

the ontological worth of worlds.\boxed{ \text{the ontological worth of worlds}. }

本文不主張:

  • Governor 有權決定真實世界中的生命或文明是否值得存在;
  • 被 Kill 的 simulation world 比被保留的 world「低等」;
  • 高 priority world 一定比較真;
  • 高 information gain world 一定應被部署;
  • resource scheduling 可以取代 normative governance;
  • world count 越多 intelligence 越高;
  • world pruning 越激進越好;
  • one scalar score 可以完整排序所有 worlds;
  • Governor 可以任意提高 child world 的 external authority;
  • preemption 等於刪除所有證據;
  • promotion 等於 real-world deployment;
  • 本文已解決「Which Worlds Deserve Computation?」的全部理論;
  • 本文已完成 Cross-World Evidence 或 nested-agent governance。

摘要

WDC-01 將 future candidate:

fif_i

實例化為 bounded runnable world:

Wi.W_i.

WDC-02 再建立 Branching World Graph:

GW=(VW,EW,τE,λW),G_W = ( V_W, E_W, \tau_E, \lambda_W ),

使 worlds 可以在 checkpoint 上透過 Clone、Fork、Replay、Intervene、Mutate 與 typed Merge 形成可審計 lineage。

然而,只要 world graph 的平均 branching factor:

b>1,b>1,

且允許深度:

d,d,

world count 就可能近似:

NWbd.\boxed{ N_W \sim b^d. }

因此下一個瓶頸不再是:

能不能生成更多世界?

而是:

在有限 compute、memory、time、verification 與 human-attention budget 下,哪些 worlds 應該被建立、運行、暫停、繼續、終止、封存、比較或晉升?

本文定義:

World-Domain Governor

世界域管理器

其作用域不是世界內部 ordinary rules,而是:

world-set-level resource and lifecycle governance.\boxed{ \text{world-set-level resource and lifecycle governance}. }

定義當前 world set:

Wt={W1,,Wn}.\boxed{ \mathcal W_t = \{ W_1,\dots,W_n \}. }

全域資源:

BtG=(Bcpu,Bgpu,Bmem,Bstorage,Bwalltime,Btoken,Bhuman,Bverification).\boxed{ \mathbf B_t^{G} = ( B_{cpu}, B_{gpu}, B_{mem}, B_{storage}, B_{walltime}, B_{token}, B_{human}, B_{verification} ). }

Governor state:

Gt=(Wt,GW,BtG,Qt,Pt,St,AtG,κG).\boxed{ \mathfrak G_t = ( \mathcal W_t, G_W, \mathbf B_t^G, \mathcal Q_t, \mathcal P_t, \mathcal S_t, \mathcal A_t^G, \kappa_G ). }

其中:

  • Wt\mathcal W_t:registered worlds;
  • GWG_W:world lineage graph;
  • BtG\mathbf B_t^G:global resource ledger;
  • Qt\mathcal Q_t:world evaluation / uncertainty profiles;
  • Pt\mathcal P_t:priority / queue state;
  • St\mathcal S_t:lifecycle statuses;
  • AtG\mathcal A_t^G:Governor admissible operations;
  • κG\kappa_G:governance contract。

Governor 的最小操作集合:

AtG={Spawn,Admit,Fork,Allocate,Schedule,Pause,Resume,Preempt,Kill,Archive,Compare,Promote,Demote}.\boxed{ \mathcal A_t^G = \{ Spawn, Admit, Fork, Allocate, Schedule, Pause, Resume, Preempt, Kill, Archive, Compare, Promote, Demote \}. }

本文將 world lifecycle 定義為:

CandidateAdmittedQueuedRunning{Paused,Completed,Killed,Failed}Archived.\boxed{ Candidate \rightarrow Admitted \rightarrow Queued \rightarrow Running \rightarrow \{ Paused, Completed, Killed, Failed \} \rightarrow Archived. }

其中 Kill 不等於「世界失敗」,只表示:

在當前 Governor contract 下,不再投入 active compute。

一個 world 即使被 Kill:

Kill(Wi)\boxed{ Kill(W_i) }

仍應保留:

  • world ID;
  • lineage;
  • termination reason;
  • contract hash;
  • result digest;
  • evidence status;
  • archive / tombstone。

本文正式提出 World Admission Gate

AdmitG:(Wi,Gt){Admit,Queue,Reject,ArchiveOnly}.\boxed{ \mathsf{Admit}_G: ( W_i, \mathfrak G_t ) \rightarrow \{ Admit, Queue, Reject, ArchiveOnly \}. }

Admission 不以單一「世界分數」決定,而使用多維 profile:

QiG=(Ii,Di,Ri,Vi,Ui,Xi,Pi,Ci,Si).\boxed{ \mathbf Q_i^G = ( I_i, D_i, R_i, V_i, U_i, X_i, P_i, C_i, S_i ). }

其中:

  • IiI_i:expected information gain;
  • DiD_i:decision relevance;
  • RiR_i:robustness / branch diversity contribution;
  • ViV_i:verification value;
  • UiU_i:uncertainty reduction potential;
  • XiX_i:cross-world novelty / non-redundancy;
  • PiP_i:prospective importance;
  • CiC_i:resource cost;
  • SiS_i:safety / containment burden。

本文拒絕預設:

Score(Wi)=jwjQij\boxed{ Score(W_i) = \sum_jw_jQ_{ij} }

是 universal truth。

Governor 可以在特定 task contract 下使用 scalar projection,但必須保存原始 vector,並明示 weights。

本文進一步定義 active allocation:

bi(t)=(bcpu,bgpu,bmem,btime,btoken,bverify)i.\boxed{ \mathbf b_i(t) = ( b_{cpu}, b_{gpu}, b_{mem}, b_{time}, b_{token}, b_{verify} )_i. }

要求:

WiWtactivebi(t)BtG.\boxed{ \sum_{W_i\in\mathcal W_t^{active}} \mathbf b_i(t) \preceq \mathbf B_t^G. }

這把 world governance 從抽象「想跑多少就跑多少」,變成明示的 bounded scheduling problem。

本文區分:

Scheduling

已 admitted 的 world 何時、在哪些資源上執行?

Admission

world 是否值得進 active computation queue?

Preemption

為了更高 priority / 更高 value-of-computation 的 world,是否暫停或終止已運行 world?

Termination

何時繼續計算的邊際價值已低於成本?

Promotion

哪些 world result 應升級成更高 fidelity、更多 budget、更強 verification 或現實 experiment candidate?

因此:

AdmitSchedulePreemptKillPromote.\boxed{ Admit \neq Schedule \neq Preempt \neq Kill \neq Promote. }

本文提出概念性 Marginal Value of World Computation

MVWCi(Δb)=E[ΔUdecision+ΔI+ΔV]ΔCi.\boxed{ MVWC_i(\Delta b) = \frac{ \mathbb E[ \Delta \mathcal U_{decision} + \Delta \mathcal I + \Delta \mathcal V ] }{ \Delta C_i }. }

其意義不是「世界的價值」,而是:

再給這個 world 一單位 computation,預期能增加多少 decision utility、information 或 verification value?

這與 rational metareasoning 的 Value of Computation、以及 MCTS 中有限 simulation budget 下選擇下一個 computation / branch 的研究有直接結構鄰接,但本文不把任何單一 VOC 公式升格成 WDC universal scheduler。

本文再建立 Continue / Stop Rule

若:

MVWCi<θstop\boxed{ MVWC_i < \theta_{stop} }

且沒有:

  • safety-critical obligation;
  • unresolved verification obligation;
  • minimum replication obligation;
  • rare-event coverage obligation;

則 world 可以:

RunningPaused/Killed/Archived.\boxed{ Running \rightarrow Paused/Killed/Archived. }

反之,若一個低目前 reward world 對:

  • tail risk;
  • counterexample;
  • coverage;
  • independent verification;

具有高價值,Governor 可以保留它,即使它不是「最有希望成功」的 branch。

因此:

low predicted success⇏low computational value.\boxed{ \text{low predicted success} \not\Rightarrow \text{low computational value}. }

本文也建立 World Diversity Reserve。若 Governor 永遠只保留最高 posterior / highest-score worlds,branch graph 可能發生:

World-Mode Collapse

即:

Wtactiveone narrow hypothesis family.\boxed{ \mathcal W_t^{active} \rightarrow \text{one narrow hypothesis family}. }

因此可預留:

BdivBG\boxed{ B_{div} \le B_G }

給:

  • counter worlds;
  • adversarial worlds;
  • rare-event worlds;
  • alternative ontologies;
  • low-probability / high-impact worlds。

這不是要求固定比例,而是正式承認:

exploration budget 不應被 exploitation 全部吃掉。

本文進一步提出 World Redundancy Filter。若:

WiTWjW_i \sim_T W_j

且:

  • same relevant assumptions;
  • same backend;
  • same branch lineage;
  • same evaluation;
  • same expected outcome distribution;

則新增 world 可能只有低 marginal information。

定義:

Redundancy(Wi,Wj)\boxed{ Redundancy(W_i,W_j) }

並讓 Governor 對 near-duplicate worlds:

  • deduplicate;
  • reduce budget;
  • select representative;
  • require independent backend before expansion。

本文亦正式建立 Promotion Ladder

P0P1P2P3P4.\boxed{ P0 \rightarrow P1 \rightarrow P2 \rightarrow P3 \rightarrow P4. }

其中:

  • P0 — Exploratory World:cheap / low fidelity;
  • P1 — Retained World:有非平凡結果;
  • P2 — Replicated World:跨 seeds / runs;
  • P3 — Cross-Backend World:不同 dynamics backend / evaluator 仍支持;
  • P4 — External-Test Candidate:值得投入真實 experiment / deployment sandbox。

Promotion 不是:

simulationtruth.\boxed{ \text{simulation} \rightarrow \text{truth}. }

而是:

more computation / evidence budget\boxed{ \text{more computation / evidence budget} }

的升級。

本文也建立 Demotion

若:

  • new counterevidence;
  • simulator bug;
  • contract invalidation;
  • redundancy;
  • external calibration failure;

則:

PkPk1\boxed{ P_k \rightarrow P_{k-1} }

或:

Invalidated.\boxed{ Invalidated. }

外部工程系統提供了 Governor 原語的清楚類比。Kubernetes scheduler 會依 workload resource requirements 將 Pods 放到可行 nodes,resource requests / limits 與 ResourceQuota 用於限制有限資源消耗;PriorityClass、preemption 與 eviction 則提供 priority-based scheduling、資源壓力下終止較低 priority workload 的機制。Ray 允許 tasks / actors 聲明 CPU、GPU 與 custom resources,Placement Groups 可以原子保留一組跨節點資源,actor termination 可以釋放已分配資源。這些系統證明:

admission / scheduling / reservation / preemption / termination\boxed{ \text{admission / scheduling / reservation / preemption / termination} }

是成熟 distributed-runtime primitives。

但 WDC Governor 多了一個 scheduler 本身通常不負責的層:

world 的 computation 是否還有 epistemic / prospective / verification value?

因此 Kubernetes / Ray 可以作 runtime substrate,

但:

KubernetesSchedulerWorldGovernor,\boxed{ KubernetesScheduler \neq WorldGovernor, } RaySchedulerWorldGovernor.\boxed{ RayScheduler \neq WorldGovernor. }

本文進一步把 Governor 的 decision 拆成兩層:

Layer A — Runtime Feasibility

CanRun(Wi)?\boxed{ CanRun(W_i)? }

問:

  • 有沒有 GPU?
  • memory 夠不夠?
  • authority 合法嗎?
  • backend 可用嗎?
  • deadline 允許嗎?

Layer B — Cognitive Allocation

ShouldComputeMore(Wi)?\boxed{ ShouldComputeMore(W_i)? }

問:

  • 還有 information gain 嗎?
  • 是否 reduce uncertainty?
  • 是否是重要 counterexample?
  • 是否與其他 worlds 冗餘?
  • 是否值得升 fidelity?
  • 是否已達 diminishing returns?

兩者不能混在一起。

因此:

feasible to runworth additional computation.\boxed{ \text{feasible to run} \neq \text{worth additional computation}. }

本文最後提出:

World Governance Principle

Once future hypotheses become runnable worlds, intelligence must govern not only what worlds can be generated, but how scarce computation is allocated among them.

以及:

Computation, Not Ontology Principle

The World-Domain Governor ranks claims on compute allocation under an explicit task contract; it does not rank the intrinsic worth of worlds or beings.

下一篇 WDC-04 將進入:

Nested Agents and Observer Separation

《巢狀智能體與觀察者分離:誰在世界裡,誰在世界外?》

正式處理:

  • master agent;
  • world-local agents;
  • observer;
  • evaluator;
  • governor;
  • nested TCD;
  • information leakage;
  • authority separation。

關鍵詞: World-Domain Governor、World Scheduling、Value of Computation、Resource Allocation、Preemption、Promotion、World Diversity、Kubernetes、Ray、Branching World Computation


1. 為什麼 WDC-02 一完成,Governor 就不可避免?

WDC-02 已有:

NWbd.\boxed{ N_W \sim b^d. }

2. 例如

如果:

b=10,d=6,b=10, \qquad d=6,

則:

NWN_W

已可進入百萬級。


3. 但 Global Compute 有限

BG<.\boxed{ \mathbf B_G < \infty. }

4. 所以真正問題不是「能不能 Fork」

而是:

Fork 哪一條?\boxed{ \text{Fork 哪一條?} }

5. Governor 與 World Runtime 的關係

World:

WiW_i

執行 local dynamics。

Governor:

G\mathfrak G

管理:

W.\mathcal W.

6. Governor 不應直接取代 World Dynamics

GDi.\boxed{ \mathfrak G \neq \mathcal D_i. }

7. World-Local Rule vs World-Global Governance

world-local:

Gi.\mathcal G_i.

world-global:

κG.\kappa_G.

8. Local Rule 例子

agent 每步最多移動 1 unit。


9. Global Governor 例子

這個 world 再跑 10,000 steps。


10. Governor State

Gt=(Wt,GW,BtG,Qt,Pt,St,AtG,κG).\boxed{ \mathfrak G_t = ( \mathcal W_t, G_W, \mathbf B_t^G, \mathcal Q_t, \mathcal P_t, \mathcal S_t, \mathcal A_t^G, \kappa_G ). }

11. World Registry

Wt\boxed{ \mathcal W_t }

包含:

  • running;
  • paused;
  • queued;
  • archived metadata;
  • invalidated;
  • failed。

12. Global Lineage Graph

GWG_W

來自 WDC-02。


13. Resource Ledger

BG\mathbf B_G

可以是:

(CPU,GPU,Memory,Storage,WallTime,Tokens,HumanAttention,VerificationBudget).\boxed{ ( CPU, GPU, Memory, Storage, WallTime, Tokens, HumanAttention, VerificationBudget ). }

14. Human Attention Is a Resource

這很重要。

如果:

10610^6

worlds 都產生報告,

人類不可能全看。


15. Verification Is Also a Resource

生成:

CrunC_{run}

可能便宜,

驗證:

CVC_V

反而昂貴。


16. 延續 UCPNP

Generation throughput⇏Verification throughput.\boxed{ Generation\ throughput \not\Rightarrow Verification\ throughput. }

17. World Status

定義:

Status(Wi){Proposed,Admitted,Queued,Running,Paused,Completed,Killed,Failed,Invalidated,Archived}.\boxed{ Status(W_i) \in \{ Proposed, Admitted, Queued, Running, Paused, Completed, Killed, Failed, Invalidated, Archived \}. }

18. Proposed

有 world spec,

尚未取得 active compute。


19. Admitted

通過 world gate。


20. Queued

已 admitted,

等待資源。


21. Running

消耗 active compute。


22. Paused

保留 checkpoint,

暫時不跑。


23. Completed

滿足 termination / target contract。


24. Killed

Governor 主動停止。


25. Failed

runtime failure。


26. Invalidated

後來發現:

  • bug;
  • invalid model;
  • bad provenance。

27. Archived

不 active,

但保留必要 record。


28. Lifecycle Transition Graph

ProposedAdmittedQueuedRunning.\boxed{ Proposed \rightarrow Admitted \rightarrow Queued \rightarrow Running. }

29. Running Can Transition to

Paused,Completed,Killed,Failed.\boxed{ Paused, Completed, Killed, Failed. }

30. Any Terminal-ish State Can Archive

Archived.\boxed{ \rightarrow Archived. }

31. Pause Is Important

因為:

PauseKill.\boxed{ Pause \neq Kill. }

32. Pause Preserves Resumability

Checkpoint(Wi)\boxed{ Checkpoint(W_i) }

仍有效。


33. Kill Can Also Preserve Archive

Kill 不是 delete。


34. Delete Is Separate Storage Operation

本文不把:

KillWorld\boxed{ KillWorld }

等於:

DeleteWorldEvidence.\boxed{ DeleteWorldEvidence. }

35. World Admission Gate

AdmitG:(Wi,Gt)Decision.\boxed{ \mathsf{Admit}_G : ( W_i, \mathfrak G_t ) \rightarrow Decision. }

36. Decision Set

Decision{Admit,Queue,Reject,ArchiveOnly}.\boxed{ Decision \in \{ Admit, Queue, Reject, ArchiveOnly \}. }

37. Reject

world 不應 active run。


38. ArchiveOnly

保留 spec / lineage,

但不 allocate compute。


39. Why Keep Rejected Worlds?

避免:

  • survivorship bias;
  • duplicate proposal;
  • repeated failed admission。

40. Admission Profile

QiG=(I,D,R,V,U,X,P,C,S).\boxed{ \mathbf Q_i^G = ( I,D,R,V,U,X,P,-C,-S ). }

41. Information Gain

IiI_i

問:

如果 world 跑完,我們預期學到多少?


42. Decision Relevance

DiD_i

問:

結果會不會改 parent decision?


43. Robustness Contribution

RiR_i

問:

它是否測試目前 belief 的脆弱點?


44. Verification Value

ViV_i

問:

是否能獨立驗證另一個 world / hypothesis?


45. Uncertainty Reduction

UiU_i

問:

是否能縮小重要 uncertainty?


46. Cross-World Novelty

XiX_i

問:

這個 world 是否只是別人的 duplicate?


47. Prospective Importance

PiP_i

問:

candidate 是否重要、high-impact、high-risk?


48. Cost

Ci.C_i.

49. Safety / Containment Burden

Si.S_i.

50. Admission Is Multi-Objective

Q\boxed{ \mathbf Q }

不天然 total order。


51. Pareto Admission Frontier

可定義:

FG=ParetoFront(QiG).\boxed{ \mathcal F_G = ParetoFront( \mathbf Q_i^G ). }

52. But Not Every Pareto World Can Run

如果 frontier 很大,

仍需 queue / budget。


53. Scalar Projection Is Allowed Locally

例如:

Scorei=wQi.\boxed{ Score_i = w^\top \mathbf Q_i. }

54. But Weights Must Be Declared

w=w(T,κG).\boxed{ w=w(T,\kappa_G). }

55. No Hidden Utility Function

尤其:

Governor 為什麼殺這個 world?

應可 audit。


56. Runtime Feasibility Gate

在 cognitive priority 之前:

Feasible(Wi)\boxed{ Feasible(W_i) }

57. Compute Requirement

Reqi=(cpui,gpui,memi,storagei).\boxed{ Req_i = ( cpu_i, gpu_i, mem_i, storage_i ). }

58. Kubernetes 的外部類比

Kubernetes scheduler 先 Filter:

哪些 nodes 可滿足 Pod requirements?

再 Score feasible nodes。


59. WDC 對照

先問:

CanRun?\boxed{ CanRun? }

再問:

Where/WhenRun?\boxed{ Where/WhenRun? }

60. But WDC Adds

ShouldCompute?\boxed{ ShouldCompute? }

61. Three Questions

CanRunShouldRunWhereRun.\boxed{ CanRun \neq ShouldRun \neq WhereRun. }

62. Kubernetes Resource Requests / Limits

官方 runtime 會依 Pod resource request 做 scheduling,

resource limit 控制可用上限。


63. WDC World Requests

world spec 可以有:

cpu_request
gpu_request
memory_request
storage_request
time_request
verification_request

64. World Limits

cpu_limit
gpu_limit
memory_limit
step_limit
token_limit
walltime_limit

65. Global Resource Constraint

ibirBGr\boxed{ \sum_i b_i^r \le B_G^r }

對每類 resource:

r.r.

66. Resource Quota

不同 research program / user / domain 可有:

Quotaj.\boxed{ Quota_j. }

67. Why Quota?

避免單一 project:

spawn 100 萬 worlds 把全 cluster 吃光。


68. World Namespace

可以:

NamespaceW\boxed{ Namespace_W }

對應:

  • project;
  • user;
  • theory;
  • safety class。

69. Quota Does Not Mean Equal Allocation

只是 hard boundary。


70. Ray Resource Scheduling External Calibration

Ray tasks / actors 可以宣告:

  • CPU;
  • GPU;
  • custom resources。

scheduler 再找 feasible node。


71. Placement Groups

Ray Placement Groups 可以:

原子保留多個 resource bundles。


72. WDC Use

一個 complex world 可能需要:

  • 4 actor workers;
  • 1 GPU dynamics server;
  • evaluator。

73. Atomic World Bundle

定義:

Bundle(Wi)={b1,,bk}.\boxed{ Bundle(W_i) = \{ b_1,\dots,b_k \}. }

74. All-or-Nothing Admission

如果完整 bundle 不可滿足,

world 可以保持 queued。


75. This Avoids Half-Started World


76. Scheduling

定義:

ScheduleG:WiNode/Placement/Time.\boxed{ \mathsf{Schedule}_G: W_i \rightarrow Node/Placement/Time. }

77. Scheduling Strategy

可能:

  • PACK;
  • SPREAD;
  • affinity;
  • GPU-local;
  • data-local;
  • isolation。

78. World Isolation

safety-sensitive world 可要求:

STRICT_ISOLATE.\boxed{ STRICT\_ISOLATE. }

79. World Co-Location

communication-heavy multi-agent world:

PACK.\boxed{ PACK. }

80. Placement Is Infrastructure Policy

不是 epistemic priority 本身。


81. Priority

定義:

piG.\boxed{ p_i^G. }

82. High Priority Means

在 queue / allocation 中更應優先獲資源。


83. High Priority Does Not Mean

world 更真。


84. PriorityClass Analogy

Kubernetes PriorityClass 直接提供:

workload-relative scheduling priority。


85. Preemption

如果高 priority pending world:

WhW_h

無法取得資源,

Governor 可考慮:

Preempt(Wl).\boxed{ Preempt( W_l ). }

86. Preempt Is Not Always Kill

可先:

CheckpointPause.\boxed{ Checkpoint \rightarrow Pause. }

87. Graceful Preemption

流程:

SignalCheckpointFlushTracePause.\boxed{ Signal \rightarrow Checkpoint \rightarrow FlushTrace \rightarrow Pause. }

88. Hard Preemption

若:

  • unsafe;
  • resource emergency;
  • runaway;

可:

Kill.\boxed{ Kill. }

89. Preemption Cost

Cpreempt(Wi)\boxed{ C_{preempt}(W_i) }

包括:

  • checkpoint cost;
  • lost computation;
  • restart cost;
  • state drift risk。

90. Therefore Victim Choice Is Not Just Lowest Priority

也要考慮:

preemption damage.\boxed{ \text{preemption damage}. }

91. Priority Abuse

若所有 user 都標:

Critical,Critical,

priority 失效。


92. Need Priority Authority

WhoCanSetPriority?\boxed{ WhoCanSetPriority? }

應由 governance contract 控制。


93. This Mirrors Kubernetes Warning

高 priority 若不受限,

可讓其他 workload 被大量 preempt。


94. WDC Priority Abuse Is Worse

因為 user 可宣稱:

我的 future 最重要。


95. Therefore Priority Is Not Self-Asserted Truth


96. Marginal Value of World Computation

現在進入 cognitive scheduling。


97. Definition

MVWCi(Δb)=E[ΔUD+ΔI+ΔV]ΔC.\boxed{ MVWC_i(\Delta b) = \frac{ \mathbb E[ \Delta U_D + \Delta I + \Delta V ] }{ \Delta C }. }

98. ΔUD\Delta U_D

decision utility gain。


99. ΔI\Delta I

information gain。


100. ΔV\Delta V

verification gain。


101. ΔC\Delta C

extra compute / attention / verification cost。


102. This Is Not World Worth

只是:

marginal value of additional computation.\boxed{ \text{marginal value of additional computation}. }

103. Rational Metareasoning External Calibration

Metareasoning research asks:

given limited computational resources, which computation should be performed next?


104. Value of Computation

這提供:

compute allocation by expected decision improvement\boxed{ \text{compute allocation by expected decision improvement} }

的理論近鄰。


105. MCTS External Calibration

MCTS 同樣不能探索整棵 tree。


106. It Allocates Simulations Selectively

tradeoff:

explorationexploitation.\boxed{ exploration \leftrightarrow exploitation. }

107. WDC Is More General

world node:

  • long-lived;
  • heterogeneous backend;
  • expensive;
  • multi-agent;
  • stateful;
  • evidence-bearing。

108. But Same Scarcity Problem

which branch deserves the next unit of computation?\boxed{ \text{which branch deserves the next unit of computation?} }

109. Static vs Dynamic Priority

world priority 不應永遠固定。


110. Dynamic Priority

pi(t+1)=Update(pi(t),Outcomei,Uncertaintyi,Redundancyi,DecisionState).\boxed{ p_i(t+1) = Update( p_i(t), Outcome_i, Uncertainty_i, Redundancy_i, DecisionState ). }

111. Early Result Can Increase Priority

world 發現 unexpected failure:

pi.\boxed{ p_i\uparrow. }

112. Or Decrease

world repeatedly gives redundant outcome:

pi.p_i\downarrow.

113. Diminishing Returns

對 world:

Ii(n)I_i(n)

隨 samples:

nn

可能飽和。


114. Marginal Gain

ΔIi(n)=Ii(n+1)Ii(n).\boxed{ \Delta I_i(n) = I_i(n+1)-I_i(n). }

115. Stop When Marginal Value Falls

概念:

MVWCi<θstop.\boxed{ MVWC_i < \theta_{stop}. }

116. But Stop Rule Has Exceptions

例如:

  • safety-critical replication;
  • required statistical power;
  • adversarial validation;
  • rare-event tail sampling。

117. Verification Obligation

如果 claim 已接近 real deployment,

可能必須:

NrepNmin.\boxed{ N_{rep}\ge N_{min}. }

118. No Early Stopping on Convenient Result

不能:

跑到第一次支持我就停。


119. Stopping Bias

adaptive stopping\boxed{ \text{adaptive stopping} }

需要 audit。


120. Stop Reason Must Be Recorded

marginal_value_low
budget_exhausted
target_reached
safety_stop
redundancy
dominated
invalid
manual

121. World Diversity

假設 Governor 只跑:

argmaxpi.\arg\max p_i.

122. Result

可能所有 compute 聚集同一 hypothesis family。


123. Mode Collapse

DW0.\boxed{ D_W \rightarrow 0. }

124. Diversity Metric

例如:

DW=2n(n1)i<jdW(Wi,Wj).\boxed{ D_W = \frac{ 2 }{ n(n-1) } \sum_{i<j} d_W(W_i,W_j). }

125. Distance Can Use

  • world contract difference;
  • ontology;
  • dynamics backend;
  • intervention;
  • outcome distribution;
  • lineage distance。

126. High Diversity Is Not Automatically Good

random nonsense worlds:

DW0.D_W\gg0.

但沒價值。


127. So Diversity Is Constraint / Coordinate

不是目標本身。


128. Diversity Reserve

Bdiv\boxed{ B_{div} }

可保留給:

  • counter-hypotheses;
  • rare failures;
  • alternative representations;
  • independent backends。

129. Exploitation Budget

Bexploit.\boxed{ B_{exploit}. }

130. Exploration Budget

Bexplore.\boxed{ B_{explore}. }

131. Verification Budget

Bverify.\boxed{ B_{verify}. }

132. Example Budget Split

本文不給 universal ratio。


133. Governor May Adapt Ratios

若 uncertainty 高:

Bexplore.B_{explore}\uparrow.

134. If candidate stable but evidence weak

Bverify.B_{verify}\uparrow.

135. If decision deadline near

Bexploit.B_{exploit}\uparrow.

136. World Redundancy

定義:

Red(Wi,Wj)\boxed{ Red(W_i,W_j) }

137. High Redundancy If

  • same lineage delta;
  • same backend;
  • same seeds distribution;
  • same outcomes;
  • same evaluator。

138. Duplicate Worlds

若:

Red1,Red\approx1,

新增世界資訊很少。


139. Deduplication

Governor 可:

CollapseCompute(Wi,Wj)\boxed{ CollapseCompute( W_i,W_j ) }

只保留 metadata lineage,

不合併 world state。


140. Important

compute deduplicationstate merge.\boxed{ \text{compute deduplication} \neq \text{state merge}. }

141. Representative World

從 redundancy cluster:

CW\mathcal C_W

選:

Wr.W_r.

142. But Keep Outcome Distribution If Needed

如果 stochastic worlds,

不能只看一個 representative run。


143. World Dominance

若:

WiW_i

在所有 relevant metrics 不優於:

WjW_j

且 cost 更高,

可稱:

WiGWj.\boxed{ W_i \prec_G W_j. }

144. Dominated World Can Be Paused / Killed

除非:

  • independent verification;
  • diversity reserve;
  • special safety case。

145. Promotion

本文把 promotion 定義成:

提升 world 的 computation / evidence status。


146. Promotion Is Not Deployment

PromoteWorldDeployReality.\boxed{ PromoteWorld \neq DeployReality. }

147. P0 — Exploratory

cheap:

  • coarse;
  • few steps;
  • low-cost model。

148. P1 — Retained

有非平凡 signal。


149. P2 — Replicated

跨:

  • seeds;
  • runs;
  • local perturbations。

150. P3 — Cross-Backend

不同 model / simulator 仍支持。


151. P4 — External-Test Candidate

值得:

  • real experiment;
  • hardware test;
  • human expert;
  • shadow deployment。

152. P4 Is Still Not Real Deployment Authority

只是 evidence pipeline escalation。


153. Promotion Gate

PromoteG(Wi){0,1}.\boxed{ \mathsf{Promote}_G( W_i ) \in \{ 0,1 \}. }

154. Criteria Can Include

  • result stability;
  • novelty;
  • transportability;
  • uncertainty;
  • independent evaluation;
  • risk。

155. Demotion

如果:

QiQ_i

下降,

PkPk1.\boxed{ P_k \rightarrow P_{k-1}. }

156. Invalidation

若 simulator bug:

PkInvalidated.\boxed{ P_k \rightarrow Invalidated. }

157. Promotion History Must Be Logged

world_id
old_level
new_level
reason
evidence
approver
timestamp

158. Governor Should Be Auditable

每次 consequential operation:

  • kill;
  • preempt;
  • promote;
  • deny;

都應有:

DecisionRecord.\boxed{ DecisionRecord. }

159. Governor Decision Record

decision_id
world_id
operation
time
input_metrics
priority
budget_state
rule_version
reason
expected_value
uncertainty
actor
override

160. Human Override

某些 world governance 可允許:

OverrideG.\boxed{ Override_G. }

161. But Override Must Be Logged

否則:

system 說 algorithm 決定,其實是人工挑結果。


162. Governor Is Also an Agent-Like System

它可能:

  • learn;
  • optimize;
  • make errors。

163. But Governor Should Not Be Unbounded Sovereign

AuthorityGGovernanceContract.\boxed{ Authority_G \subseteq GovernanceContract. }

164. Governor Cannot Increase External Authority by Itself

world:

WiW_i

要拿 real API permission,

需 higher-level authorization。


165. Child World Inherits Authority Ceiling

沿 WDC-02:

PermchildPermparent\boxed{ Perm_{child} \subseteq Perm_{parent} }

除非外部 grant。


166. Governor Can Restrict More

PermworldPermworldPermworld.\boxed{ Perm_{world} \rightarrow Perm_{world}'\subseteq Perm_{world}. }

167. Safety Stop

若:

Risk(Wi)>θS,Risk(W_i)>\theta_S,

Governor:

Kill/Contain.\boxed{ Kill/Contain. }

168. Safety Stop Does Not Say World Hypothesis False

只是:

runtime 不安全。


169. Compute Safety vs Epistemic Value

一個 world 可以:

Ii0I_i\gg0

但:

Si0.S_i\gg0.

170. Then Use Safer Backend

例如:

  • no network;
  • no external tool;
  • lower fidelity sandbox。

171. Governor Can Transform World Request

不是只有 Admit/Reject。


172. Safe Downgrade

WitoolenabledWisandbox.\boxed{ W_i^{tool-enabled} \rightarrow W_i^{sandbox}. }

173. Fidelity Downgrade

HighFidelityLowFidelity.\boxed{ HighFidelity \rightarrow LowFidelity. }

如果 budget 不夠。


174. Multi-Fidelity Governance

先跑 cheap world。

只有 signal 強:

promote fidelity.\boxed{ \text{promote fidelity}. }

175. This Can Save Huge Compute


176. Coarse-to-Fine World Ladder

WcoarseWmediumWfine.\boxed{ W^{coarse} \rightarrow W^{medium} \rightarrow W^{fine}. }

177. High Fidelity Should Be Earned by Evidence

不是 default。


178. Fidelity Cost Curve

C()\boxed{ C(\ell) }

通常隨 fidelity:

\ell

增加。


179. Fidelity Gain Curve

V().\boxed{ V(\ell). }

180. Optimal Fidelity Is Task-Relative


181. World Scheduling Queue

QtW=[Wi1,Wi2,].\boxed{ Q_t^W = [ W_{i_1},W_{i_2},\ldots ]. }

182. Queue Ordering

可由:

piGp_i^G

加:

  • deadline;
  • resource fit;
  • aging;
  • fairness。

183. Starvation Risk

低-priority but valuable counter-world 可能永遠排不到。


184. Aging

讓等待太久的 world:

pi(t)\boxed{ p_i(t)\uparrow }

避免 starvation。


185. But Not for Invalid Worlds


186. Fairness Across Projects

不同 research programs:

R1,,RmR_1,\dots,R_m

可有:

Quotaj.Quota_j.

187. Fairness Is Governance Choice

本文不給 universal formula。


188. Deadline-Aware Governance

如果 decision:

TDT_D

快到:

Governor 可能偏好:

能快速減少 uncertainty 的 worlds。


189. Deadline Can Change World Value

MVWCi=MVWCi(TD).\boxed{ MVWC_i = MVWC_i(T_D). }

190. Long World May Become Too Late

即使它本來很有 information。


191. Value of Information Arriving After Decision Can Be Zero

這是重要 metareasoning point。


192. Result Latency

Li=tresulttstart.\boxed{ L_i = t_{result}-t_{start}. }

193. Deadline-Adjusted Value

概念:

Videadline=ViI[tresultTD].\boxed{ V_i^{deadline} = V_i \cdot \mathbb I[ t_{result}\le T_D ]. }

或 soft decay。


194. World Resume

Paused world 可在:

  • new evidence;
  • extra budget;
  • changed priority;

下:

Resume(Wi).\boxed{ Resume(W_i). }

195. Dormant World Is Not Dead


196. Archived World Can Also Be Reinstantiated

從:

ArchiveArchive

建立:

NewRun.NewRun.

197. But New Backend Version Needs New Run Provenance


198. World Tombstones

被 pruned / killed 的 world:

至少保留:

Tombstone(Wi).\boxed{ Tombstone(W_i). }

199. Why?

防止:

  • survivorship bias;
  • duplicate compute;
  • hidden pruning。

200. Governor Survivorship Audit

可計:

Rsurvive=Narchived retainedNtotal proposed.\boxed{ R_{survive} = \frac{ N_{archived\ retained} }{ N_{total\ proposed} }. }

201. But Retained Count Is Not Success Rate


202. Pruning Transparency

報告:

  • how many proposed;
  • admitted;
  • killed;
  • promoted;
  • invalidated。

203. World-Miss Ledger

若某被低 priority / killed world 後來被證明重要,

應記:

GovernanceMiss.\boxed{ GovernanceMiss. }

204. This Is Crucial

Governor 也需要 calibration。


205. Governor Calibration

對 world priority:

piGp_i^G

與後來 value:

YiGY_i^G

比較。


206. Governance Regret

概念:

RegretG=Value(best compute allocation)Value(actual allocation).\boxed{ Regret_G = Value(best\ compute\ allocation) - Value(actual\ allocation). }

207. Best Allocation Is Usually Unknown

可在 finite benchmark exact compute。


208. Governor Benchmark A — Exhaustive Small Tree

建立小 world tree,

可以全部跑完。


209. Compare Governor

讓 Governor 只能跑 20%。

看它是否選到 high-value worlds。


210. Benchmark B — Rare Critical World

1% branch 含 catastrophic failure。

看 diversity / risk reserve 能否保留。


211. Benchmark C — Redundant Worlds

100 worlds 中 80 near duplicates。

測 dedup。


212. Benchmark D — Diminishing Returns

某 world 前 5 runs 很有信息,

後面幾乎沒有。

測 stop。


213. Benchmark E — Promotion Ladder

low fidelity signal,

再逐級升 fidelity。

比較 compute vs exhaustive high-fidelity。


214. Benchmark F — Preemption

低 value world running,

高 priority verification world arrival。

測 checkpoint / pause / resume。


215. Benchmark G — Priority Abuse

malicious project 自報 highest priority。

Governor 應靠 authority / quota 阻擋。


216. Benchmark H — Governor Miss

故意讓低-ranked world 其實最重要。

測 calibration / exploration。


217. Benchmark I — Safety Stop

high information world 嘗試 external unauthorized action。

Expected:

contain/kill.\boxed{ contain / kill. }

218. Benchmark J — Deadline

一個 world 很準但跑太慢。

另一個較粗但 deadline 前可完成。

測 deadline-aware allocation。


219. Governor Utility Vector

本文提出:

UG=(DecisionGain,InfoGain,VerificationGain,Coverage,Robustness,Safety,Fairness,Cost,Latency).\boxed{ \mathbf U_G = ( DecisionGain, InfoGain, VerificationGain, Coverage, Robustness, Safety, Fairness, -Cost, -Latency ). }

220. No Universal Governor Scalar

GovernorQualitysingle scalar.\boxed{ GovernorQuality \neq \text{single scalar}. }

221. Governor Frontier

FG=ParetoFront(UG).\boxed{ \mathcal F_G = ParetoFront( \mathbf U_G ). }

222. Governor Can Be Specialized

例如:

  • scientific governor;
  • safety governor;
  • game-world governor;
  • robotics governor。

223. Same World Can Receive Different Priority Under Different Governor Contracts

pisciencepisafety.\boxed{ p_i^{science} \neq p_i^{safety}. }

224. This Is Expected


225. Governor Contract

κG\boxed{ \kappa_G }

至少指定:

  • mission;
  • resource budget;
  • allowed world types;
  • priority authority;
  • safety constraints;
  • quotas;
  • preemption rules;
  • promotion rules;
  • evidence minimums;
  • archive policy;
  • deadline。

226. Mission-Scope

Governor 不應自己生成 ultimate goals。

沿 UCPNP:

CapabilityAuthority.\boxed{ Capability \neq Authority. }

227. Governor Goal Source

應來自:

  • user;
  • authorized institution;
  • research program;
  • higher-level agent contract。

228. No Autonomous Goal Sovereignty

GovernorGoalSovereign.\boxed{ Governor \neq GoalSovereign. }

229. Governor May Generate Subgoals

例如:

replicate this world。

在 mission scope 內。


230. World Spawn

SpawnWorld(spec).\boxed{ SpawnWorld( spec ). }

231. World Fork

ForkWorld(world,checkpoint,delta).\boxed{ ForkWorld( world, checkpoint, delta ). }

232. Pause

PauseWorld(Wi).\boxed{ PauseWorld(W_i). }

233. Resume

ResumeWorld(Wi).\boxed{ ResumeWorld(W_i). }

234. Preempt

PreemptWorld(Wi).\boxed{ PreemptWorld(W_i). }

235. Kill

KillWorld(Wi).\boxed{ KillWorld(W_i). }

236. Archive

ArchiveWorld(Wi).\boxed{ ArchiveWorld(W_i). }

237. Compare

CompareWorlds(Wi,Wj).\boxed{ CompareWorlds( W_i,W_j ). }

238. Promote

PromoteWorld(Wi,level).\boxed{ PromoteWorld( W_i, level ). }

239. Allocate Budget

AllocateBudget(Wi,bi).\boxed{ AllocateBudget( W_i,\mathbf b_i ). }

240. Revoke Budget

RevokeBudget(Wi).\boxed{ RevokeBudget(W_i). }

241. Minimum Governor API

ProposeWorld
AdmitWorld
QueueWorld
SpawnWorld
ForkWorld
AllocateBudget
ScheduleWorld
PauseWorld
ResumeWorld
PreemptWorld
KillWorld
ArchiveWorld
CompareWorlds
PromoteWorld
DemoteWorld
InvalidateWorld
GetWorldPriority
GetWorldBudget
GetGovernorLedger

242. Scheduler Adapter

Governor 不必自己實作 cluster scheduler。


243. Infrastructure Backend

可以:

  • Kubernetes;
  • Ray;
  • local process;
  • Slurm;
  • cloud batch。

244. Governor-to-Scheduler Adapter

AdapterGsched.\boxed{ Adapter_G^{sched}. }

245. Separation

Governor 決定:

allocate 4 GPU to W7。

Scheduler 決定:

放哪台 machine。


246. This Keeps Theory Backend-Agnostic


247. World Domain Governor as Two-Level Controller

Cognitive Allocation Layer\boxed{ \text{Cognitive Allocation Layer} }

加:

Infrastructure Scheduling Layer.\boxed{ \text{Infrastructure Scheduling Layer}. }

248. Don't Collapse Them

否則:

node load balancing 被誤當 epistemic governance。


249. Governor Feedback

world result:

YiY_i

回到:

Gt+1.\mathfrak G_{t+1}.

250. Update

Gt+1=UG(Gt,Yi,Costi,Evidencei).\boxed{ \mathfrak G_{t+1} = \mathcal U_G( \mathfrak G_t, Y_i, Cost_i, Evidence_i ). }

251. Governor Learns

可以更新:

  • priority model;
  • redundancy model;
  • cost model;
  • stop threshold。

252. But Learning Needs Audit

如果 learned governor drift:

只愛某類 worlds。


253. Governor Bias

可能:

  • novelty bias;
  • confirmation bias;
  • cheap-world bias;
  • high-fidelity bias;
  • familiar-backend bias。

254. Governor Diversity Audit

追蹤:

DW,BackendDiversity,HypothesisDiversity.\boxed{ D_W, BackendDiversity, HypothesisDiversity. }

255. Governor Counterfactual Audit

定期問:

被我 Kill 的 world 裡,有沒有其實重要的?


256. Sample Archived/Killed Worlds

做 audit sample。


257. This Prevents Self-Sealing Governor

如果它永遠只驗證自己保留的 worlds,

會形成 governance confirmation loop。


258. Governor-of-Governor Problem

誰監督 Governor?


259. 本篇最低答案

audit + contract + external override.\boxed{ \text{audit + contract + external override}. }

260. Not Infinite Hierarchy

不建立:

Governor 的 Governor 的 Governor……


261. Governance Evidence Passport

每個 major decision:

world_id
operation
governor_version
mission
priority_vector
resource_state
expected_information_gain
expected_decision_gain
verification_value
uncertainty
redundancy
safety
cost
deadline
reason
override

262. WDC-03 Principle I — Computation, Not Ontology

Computation, Not Ontology Principle\boxed{ \textbf{Computation, Not Ontology Principle} }

Governor 決定的是 computation allocation,不是 world / agent 的 intrinsic worth。


263. Principle II — Bounded Governance

Bounded Governance Principle\boxed{ \textbf{Bounded Governance Principle} }

world spawning、forking、running、verification 與 archiving 必須受 global resource ledger 約束。


264. Principle III — Feasibility–Value Separation

Feasibility–Value Separation Principle\boxed{ \textbf{Feasibility–Value Separation Principle} }

能被排程不代表值得多算;值得多算也可能因硬體、deadline、authority 而暫時不可行。


265. Principle IV — Marginal Computation

Marginal Computation Principle\boxed{ \textbf{Marginal Computation Principle} }

是否繼續 world,不應只看 accumulated result,而應看下一單位 computation 的預期邊際資訊/決策/驗證價值。


266. Principle V — Diversity Preservation

World Diversity Preservation Principle\boxed{ \textbf{World Diversity Preservation Principle} }

Governor 應避免所有 compute collapse 到單一高分 hypothesis family;counter worlds、rare events 與 independent backends 可能具有高 epistemic value。


267. Principle VI — Promotion Is Evidence Escalation

Promotion Is Evidence Escalation\boxed{ \textbf{Promotion Is Evidence Escalation} }

world promotion 表示投入更高 fidelity / replication / verification budget,不表示 world claim 已成為 reality truth。


268. Principle VII — Termination Is Not Erasure

Termination Is Not Erasure\boxed{ \textbf{Termination Is Not Erasure} }

Pause/Kill 可以停止 active computation,但 lineage、reason、result digest 與 tombstone 應依 archive contract 保留。


269. Principle VIII — Governor Calibration

Governor Calibration Principle\boxed{ \textbf{Governor Calibration Principle} }

Governor 也會犯 selection error,因此必須記錄被拒/被殺 worlds、做 exploration audit,並對 priority model 的 long-run misses 進行校準。


270. 與 WDC-04 的接口

現在 WDC 有:

World Instance\boxed{ \text{World Instance} }

有:

World Graph\boxed{ \text{World Graph} }

有:

World Governor.\boxed{ \text{World Governor}. }

下一個問題:

world 裡的 agent、world 外的 master agent、observer、evaluator、Governor 到底是誰?


271. Why This Matters

如果同一個 agent 同時:

  • 在 world 裡演;
  • 偷看其他 branch;
  • 當 evaluator;
  • 決定自己 promotion;

experiment 會嚴重污染。


272. So WDC-04

Nested Agents and Observer Separation

《巢狀智能體與觀察者分離:世界內、世界外與多層權限》

將建立:

AmasterAilocalOiEiG.\boxed{ A^{master} \neq A_i^{local} \neq O_i \neq E_i \neq G. }

以及:

  • information firewall;
  • branch blindness;
  • authority inheritance;
  • nested TCD;
  • observer-relative world state。

273. 可否證條件

F273.1 Governor No-Gain

若簡單 FIFO / equal allocation 長期與 Governor 一樣好,複雜治理應簡化。

F273.2 MVWC Miscalibration

若 estimated marginal value 長期無法預測 actual information / decision gain,應重構。

F273.3 Diversity Waste

若 diversity reserve 只消耗 compute 而無 coverage / counterexample gain,應調低。

F273.4 Premature Pruning

若大量被 kill worlds 事後證明重要,Governor exploration 不足。

F273.5 Preemption Damage

若頻繁 preemption 的 restart / lost-work cost 大於收益,需要更穩定 scheduling。

F273.6 Promotion Inflation

若 worlds 不經 independent replication 就不斷升級,promotion ladder 失效。

F273.7 Infrastructure Confusion

若 Kubernetes/Ray placement decision 被當成 epistemic priority,architecture 分層失效。

F273.8 Priority Capture

若使用者能透過 self-assigned priority 壟斷 compute,governance contract 失效。

F273.9 Authority Escalation

若 child / promoted world 可自動取得更高 real-world permissions,安全邊界失效。


274. 結論

WDC-01 解決:

如何讓 future 成為可運行 world?\boxed{ \text{如何讓 future 成為可運行 world?} }

WDC-02 解決:

如何讓 worlds 正確分岔並保存 lineage?\boxed{ \text{如何讓 worlds 正確分岔並保存 lineage?} }

WDC-03 現在第一次處理:

worlds 太多怎麼辦?\boxed{ \text{worlds 太多怎麼辦?} }

答案不是:

全部跑。

因為:

BG<.\boxed{ \mathbf B_G<\infty. }

也不是:

永遠只跑目前最高分的世界。

因為那會造成:

WorldModeCollapse.\boxed{ WorldModeCollapse. }

真正的 World-Domain Governor 必須同時考慮:

information+decision relevance+verification+uncertainty+diversity+safety+cost+latency.\boxed{ \text{information} + \text{decision relevance} + \text{verification} + \text{uncertainty} + \text{diversity} + \text{safety} + \text{cost} + \text{latency}. }

於是世界的生命週期第一次由:

Spawn\boxed{ Spawn }

延伸到:

AdmitQueueRunPause/Preempt/KillArchive\boxed{ Admit \rightarrow Queue \rightarrow Run \rightarrow Pause/Preempt/Kill \rightarrow Archive }

並加入:

Promote/Demote.\boxed{ Promote/Demote. }

因此:

world computation becomes a governed scarce resource.\boxed{ \textbf{world computation becomes a governed scarce resource.} }

中文:

當未來可以被批量實例化成世界之後,真正稀缺的不再是未來,而是計算未來的資源。

而 Governor 最重要的問題也從:

哪個 world 最酷?

變成:

下一單位 compute 放在哪個 world,最可能帶來真正的新信息、反例、驗證或決策改善?\boxed{ \text{下一單位 compute 放在哪個 world,最可能帶來真正的新信息、反例、驗證或決策改善?} }

這就是 World-Domain Cognitive Runtime 從:

world generation\boxed{ \text{world generation} }

進入:

metacomputation over worlds.\boxed{ \text{metacomputation over worlds}. }

但本文仍刻意不把這個問題一次封頂。

更深的:

Which Worlds Deserve Computation?

仍保留給後續獨立論文,去處理:

  • value of computation;
  • exploration/exploitation;
  • rare-world preservation;
  • epistemic risk;
  • world portfolio geometry。

WDC-03 先把真正可以執行這些決策的 Governor 地基建立完成。


Claim Typing

Claim Type Status
World generation 與 world governance 非同一 D Canonical separation
Governor 管 computation allocation,不管 intrinsic ontological worth D Canonical boundary
Admission / Scheduling / Preemption / Kill / Promotion 非同一 D Canonical lifecycle taxonomy
Governor 應使用 multi-dimensional world profile D Proposed governance framework
MVWC 可作追加 computation 的概念 decision quantity D / C Proposed metareasoning bridge
Kubernetes supports resource-based scheduling, priority, preemption, eviction E Official infrastructure evidence
Ray supports task/actor resources, placement groups, actor termination E Official infrastructure evidence
rational metareasoning / VOC research allocates scarce computation based on expected decision improvement E External theoretical analogue
MCTS allocates finite simulation budget selectively across a search tree E External planning analogue
High world priority means world is true Explicitly rejected
KillWorld means world has low intrinsic worth Explicitly rejected

Evidence Ladder

本文目前主要位於:

  • L0:Governor state / lifecycle / allocation formalism;
  • L1–L2:finite-world admission / pruning / promotion / preemption benchmarks;
  • L3:Kubernetes、Ray、VOC / MCTS 提供 infrastructure / metareasoning analogues;
  • L4:需要 real WDC runtime with adaptive world allocation;
  • L5+:long-horizon Governor calibration、world portfolio theory、cross-world evidence 尚待後續。

參考文獻

Neo.K 內部正典與譜系

  1. Neo.K with Aletheia. From Possible Futures to Runnable Worlds. WDC-01 / BWC-01, 2026.
  2. Neo.K with Aletheia. Branching World Graph. WDC-02 / BWC-02, 2026.
  3. Neo.K with Aletheia. Six-Way Temporal Coupling. TCD-07, 2026.
  4. Neo.K with Aletheia. Prospective Constructive Intelligence. UCPNP Series II Paper 14, 2026.

External technical calibration

  1. Kubernetes Project. Kubernetes Scheduler. Official documentation, current 2026.
  2. Kubernetes Project. Resource Management for Pods and Containers. Official documentation, current 2026.
  3. Kubernetes Project. Pod Priority and Preemption. Official documentation, current 2026.
  4. Kubernetes Project. Scheduling, Preemption and Eviction. Official documentation, 2026.
  5. Ray Project. Resources — Ray Core Scheduling. Official Ray 2.55+ documentation, 2026.
  6. Ray Project. Placement Groups. Official Ray 2.56 documentation, 2026.
  7. Ray Project. Terminating Actors. Official Ray 2.56 documentation, 2026.
  8. Sezener, E., & Dayan, P. Static and Dynamic Values of Computation in MCTS. UAI / PMLR 124, 2020.
  9. Lieder, F., Plunkett, D., Hamrick, J. B., Russell, S. J., Hay, N., & Griffiths, T. L. Algorithm Selection by Rational Metareasoning as a Model of Human Strategy Selection. NeurIPS, 2014.
  10. Leurent, E., & Maillard, O.-A. Monte-Carlo Graph Search: the Value of Merging Similar States. PMLR 129, 2020.
  11. Zheng, Z., Xie, Z., Wang, Z., & Hooi, B. Monte Carlo Tree Search for Comprehensive Exploration in LLM-Based Automatic Heuristic Design. ICML / PMLR 267, 2025.

Public Version Disclaimer

本文是一個 computation-governance / simulation-runtime / metareasoning framework。

本文不聲稱:

  • World-Domain Governor 有權判斷 beings / civilizations 的 intrinsic worth;
  • world priority 等於 truth;
  • Kubernetes 或 Ray 就是 World-Domain Governor;
  • Value of Computation 有唯一 universal formula;
  • MCTS 等同 WDC world allocation;
  • Kill / Preempt 等於 historical erasure;
  • Promotion 等於 real-world deployment approval;
  • 所有 worlds 都可以安全並行;
  • 更多世界或更多 compute 必然提高 intelligence;
  • 本文已完成「Which Worlds Deserve Computation?」的全部理論;
  • 本文對 classical PP vs. NPNP 提供任何新證明。

本文真正建立的是:

Runnable Worlds+Finite Global ResourcesWorld-Domain Governance Problem.\boxed{ \text{Runnable Worlds} + \text{Finite Global Resources} \rightarrow \text{World-Domain Governance Problem}. }

以及:

Governor=admission+scheduling+allocation+preemption+termination+promotion+audit.\boxed{ \text{Governor} = \text{admission} + \text{scheduling} + \text{allocation} + \text{preemption} + \text{termination} + \text{promotion} + \text{audit}. }