Regulation Structures in Persistent Selection Systems:
RG-0 to RG-9 and the Life-Regulation-Proto Kernel in D-Architecture

1. Introduction

D-Architecture Core and Structural Consequences establish why a persistent selection system cannot be understood through optimization, total knowledge, or single-goal fixation. The published OC note then states what implementation-level disciplines must be respected if such a system is not to undermine its own openness in use. Applied Regulation occupies the next position in the chain. Its question is narrower and later: once Life is already present, what structure becomes necessary so that repeated selection does not harden the boundary into self-collapse?

The answer is not a new purpose layer. Regulation is the structure by which Life imposes selection restraint upon itself. It slows, buffers, postpones, and filters selection so that repeated fixation does not irreversibly outpace restoration. Yet the canonical source goes further than a simple four-item list. Regulation is only one part of a larger operational kernel, because restraint without reopening would itself become another collapse mechanism. For that reason the canonical document presents a coupled kernel: Life-Regulation-Proto, or LRP.

This note restates that kernel in paper form. It preserves the architectural claim that Regulation is the first necessary structure after Life, and that Restoration Channel and Proto-Model follow not as luxuries but as the minimum additional structures required to stop regulation from becoming rigid, blind, or self-suffocating.

2. Status, Invariant, Notational Commitments, and Structural Position

The invariant preserved throughout the note remains the minimal maintenance condition:

The condition does not demand success, stability, improvement, or optimization. It demands only that at some later time, reachable options remain non-empty. Regulation enters precisely because Life, once it repeatedly selects under constraint, produces pressures that tend to shrink O(x) over time.

The Life definition used here follows LIFE-P0 in the canonical source: Life is a structure that forms and maintains a boundary in order to continue selection under constraint. Regulation becomes necessary because that same continuity of constrained selection generates accumulated pressure against future openness unless selection is restrained.

The formulas used in this note are observational or isomorphic notation. They do not function here as optimization laws, predictive engines, or executable control rules. Their role is to state structural constraints and relations in a compact paper form. In that sense they remain companion notation rather than implementation mandates.

The following notation is sufficient for the present note.

• O_hat(x): the carried-forward approximate local summary of option-space status; this ASCII form corresponds to the canonical O-hat notation.
• O_hat(x) := (N_hat(x), R_hat(x), T_hat(x)): approximate multiplicity trend, restorability estimate, and temporal slack estimate under locality limits.
• o_hat_t: the current local summary used by Delay judgment at time t.
• Theta_near: the signal for accelerating approach toward option-space collapse; this ASCII form corresponds to the canonical Theta_near.
• P_t: the restoration candidate pool.
• Pi_t: the finite slot set of the Proto-Model; this ASCII form corresponds to the canonical Pi.

From the canonical source, four status claims follow immediately.

• Regulation is not a purpose layer, a consciousness layer, or a higher ontological stage.
• Regulation does not make selection better in a teleological sense; it prevents selection from ruining its own continuation conditions.
• The structural position is Matter -> Field -> Life -> Regulation -> [OPEN].
• Later branches such as Consciousness, Intelligence, and Civilization remain open rather than necessary at this level.

The canonical document further sharpens the status declaration by treating Life-Regulation-Proto as a single operational kernel. Regulation cannot be isolated from restoration and internal summary while still preserving I_min in practice. The role of the present note is therefore not merely to translate RG-0 through RG-9, but to show why the LRP grouping is structurally coherent.

3. Four Necessary Modes of Regulation

RG-1 to RG-3 derive the pressure that makes Regulation unavoidable. If Life repeats selection under accumulating constraint, boundary rigidity rises, internal freedom decreases, reachable options contract, and I_min approaches violation. The architecture therefore requires at least one of four restraint modes, and in practice all four must coexist.

3.1 Rate and Delay

Rate limits how often actual selection may pass. Delay licenses a held non-selection state rather than forcing immediate reaction.

Rate is necessary because repeated fixation can outrun restoration under cost and delay. Delay is necessary because reflexive reaction hardens local misjudgment into a single path. Together they create temporal room in which observation, threshold detection, and restoration can remain meaningful.

3.2 Buffer and Suppress

Buffer reduces the impact of incoming events on internal state. Suppress refuses to promote some incoming signals into full events.

Buffer is necessary because unattenuated impact drives abrupt state change, which in turn accelerates constraint accumulation. Suppress is necessary because not every signal can be treated as an actionable event without triggering input explosion. Yet neither may become absolute closure: Buffer may not turn into information sealing, and Suppress may not become permanent failure-erasure or danger-removal.

3.3 Set necessity and common prohibitions

The canonical claim is stronger than separate usefulness. Rate controls frequency, Delay controls timing, Buffer controls impact, and Suppress controls the number of event candidates. If any one is absent, the others are progressively neutralized. Regulation is therefore a set-necessity rather than a menu.

Several prohibitions apply to the whole set. Regulation must not be optimized as a single number, must not rely on global-state control, must not harden into static policy, and must not redescribe itself in terms of normal versus abnormal states. Each of those moves reintroduces teleology, omniscience, or closure through the back door.

4. Gate Composition and Shared Signals

If the four regulatory modes exist but are not compositionally fixed, they can interfere with one another and produce the very narrowing they were meant to prevent. RG-4 therefore treats Regulation not as a monolithic rulebook but as a gate composition.

In this notation, e_t^eff is the output of Suppress, o_hat_t is the current local summary used by Delay judgment, and G_rate is the Rate gate. When G = 0, the system does not pass through to actual selection; it remains in HOLD or NOP.

The preferred order is fixed as Suppress -> Buffer -> Delay -> Rate. Suppress acts first because it reduces the number of candidate events before they flood the system. Buffer acts next because it attenuates the impact of what remains. Delay comes after impact control because it preserves a non-reactive state in which local summary can update. Rate comes last because early rate blockage would hide internal overload behind a blanket refusal to act.

The shared signal layer is intentionally minimal. The canonical source requires only two common signals: overload for input-volatility surge and Theta_near for approach toward option-space collapse.

Both signals are pattern-based rather than numerically final. overload compares recent input against ordinary local variation, while Theta_near requires co-occurrence among restoration stagnation, HOLD saturation, and return-without-gain patterns. The release rule is equally necessary: if both signals relax, regulatory pressure must also relax gradually. Otherwise Regulation itself hardens the boundary and suffocates Life.

5. Restoration Channel

RG-6 identifies a central paradox: Regulation preserves the system by constraining selection, but prolonged constraint also blocks the emergence of new paths. If no counter-structure exists, Regulation eventually preserves only a rigid shell. Restoration is therefore reintroduced here in a narrower operational form: not rollback, not return to a prior safe state, but the reopening of conditions under which options may arise again.

The canonical source solves the paradox by separating candidate generation from execution. Candidate generation must remain partially outside the immediate Regulation gate, while actual admission into action must still respect conditions.

Three minimum requirements follow. First, candidate generation is not fully dependent on Rate or Delay gates and may partially bypass Suppress. Second, generation is low-cost and non-decisional: it is production and storage rather than success/failure choice. Third, admission is delayed rather than immediate; candidates are stored, retained, and only later incorporated when regulatory pressure relaxes. Immediate execution and endless storage are both collapse paths. The channel exists to reopen possibility, not to guarantee improvement.

6. Proto-Model and Forgetting

Once Regulation and Restoration Channel coexist, a further problem appears. Without any internal summary, the system cannot tell what has repeatedly been held, ignored, or proposed for restoration. It will regenerate the same failed candidates, repeat blocked patterns, and overtighten Regulation through blind repetition. Proto-Model is introduced as the minimum answer to that problem.

Proto-Model is not a world model, not a predictive system, not self-awareness, and not a value calculator. It is a minimal internal summary that allows Regulation and Restoration Channel to operate without continuously sabotaging one another.

Each slot carries only a non-semantic label, a rough recurrence trace, a recent timestamp, and a coarse source such as HOLD, SUPPRESS, or RESTORE. The decay term is not optional. Without forgetting, Proto-Model becomes permanent global memory and violates locality by another route. With forgetting, the summary remains finite, reusable, and recyclable.

The most important restriction is non-intervention. Proto-Model does not select, does not directly permit or forbid, and does not control Rate or Delay. It may only help detect rough repetition or duplication. The moment semantic meaning, goal significance, or direct decision power is assigned to it, Proto-Model is contaminated into an evaluative model and reopens the SC-1 closure path.

This is also where the OC-8 prohibition of self-closure becomes relevant at the regulation level. Proto-Model may support coordination, but it may not treat its own summary as a complete internal proof that the system is fully understood, fully safe, or self-validating.

7. The LRP Kernel and Open Branches

RG-8 gathers the preceding layers into a single operational module. The point is not to produce a full engineering design, but to state the minimum coupled form under which Life, Regulation, restoration, and summary remain jointly coherent.

Here O_hat denotes the carried-forward approximate local summary, while Pi denotes the finite slot set of the Proto-Model. This module preserves the same invariant I_min while closing only the internal dynamics that must be fixed. Interfaces such as A(x) and O(x) remain open sets rather than completed lists. The kernel is therefore partially closed for operational discipline and partially open for future-selectable continuation.

The canonical sequence is: Life, then Regulation, then the signal layer, then Restoration Channel, then Proto-Model. This is a derivational order rather than a temporal stage theory. From that point onward, further branching becomes structurally open rather than necessary. Meta-Model may appear if relations among summaries become necessary. Distributed Regulation may appear when multiple Life systems interfere. Consciousness may appear if D11' related conditions become relevant. None of these branches is licensed as a guaranteed ascent or a teleological completion.

8. Minimum Implementation Conditions and Audit View

RG-9 condenses the kernel into a minimal checklist, but the canonical source also supplies named minimum conditions for auditability. The practical question is not whether a system looks sophisticated, but whether the required channels, states, and release paths actually exist in a non-collapsing form.

8.1 Four regulatory modes

• R1-R3: Rate requires a realized selection counter, a time window, and a gate that actually changes passage frequency. A pure stop-only mode is prohibited because it blocks restoration search and reproduces shrinkage under another name.
• D1-D3: Delay requires an explicit HOLD state, an anti-collapse loop during HOLD, and a pattern-based release condition. While held, the system must keep updating local summary, detecting threshold-approach patterns, and allowing restoration-candidate search.
• B1-B3: Buffer requires internal state storage, an input accumulator, and an attenuation or clamp rule so that outside input cannot overwrite internal state one-to-one. Buffer may damp, but it may not become total information closure, because that would also suppress restoration signals.
• S1-S3: Suppress requires an event classifier, an ignored-log, and a re-promotion path for repeated signals. It must reduce event count rather than erase danger, eliminate failure, or freeze omission into permanent exclusion.

8.2 Restoration and Proto-Model conditions

• Restoration R1-R3: candidate generation may bypass immediate Rate and Delay gates and may partially bypass Suppress; generation remains low-cost and non-decisional; binding remains delayed through storage and later admission only under release conditions.
• P1: Proto-Model requires a finite slot system rather than limitless global memory.
• P2: labels must remain non-semantic. Good, bad, dangerous, and goal-bearing labels contaminate the summary into an evaluative model.
• P3: Proto-Model must not directly intervene in D9, permission or prohibition, or Rate or Delay control.
• P4: forgetting is necessary. Recurrence traces decay and slots become reusable; otherwise the summary hardens into another form of closure.

8.3 Condensed checklist

• Rate must alter realized selection frequency rather than exist only as a label.
• Delay must preserve an explicit HOLD state in which observation and pattern detection remain active.
• Buffer must prevent external input from overwriting internal state one-to-one.
• Suppress must retain an ignored-log or re-promotion path so that omission does not become permanent prohibition.
• Theta_near must require co-occurring patterns rather than a single absolute threshold.
• overload must read local deviation rather than a universal global number.
• Restoration candidates must be able to form outside the immediate gate path and enter action only under release conditions.
• Proto-Model must remain finite, forgetful, non-semantic, and non-decisional.

8.4 Processing boundary declaration

This structure handles internal summary and, at most, the emergence of relations among such summaries. Demands for expression, evaluation, cost comparison, or directional assignment are not judged, designed, or called from within the present kernel. Those demands depend on conditions outside this structure, and the branch is therefore kept open rather than defined here.

9. Appendix: Core Correspondence

The canonical source concludes with an explicit correspondence table showing how Applied Regulation remains downstream of Core and OC rather than introducing a new autonomous doctrine. The table is reproduced here in compact form.

10. Conclusion

Applied Regulation in D-Architecture is not a theory of governance, optimization, or self-improvement. It is the argument that Life, once it persists under repeated constrained selection, must acquire a negative operational kernel that prevents self-destruction by over-selection. That kernel begins with Rate, Delay, Buffer, and Suppress, but it does not end there. Without Restoration Channel, Regulation hardens into rigidity. Without Proto-Model, restoration and regulation repeatedly sabotage one another. The resulting LRP kernel is therefore the first necessary operational structure after Life rather than an optional extension.

SC-9 applies reflexively here as well: this document cannot fully describe the Regulation structure it formalizes.

References

1. Jung, J.H., 2026a. Structural Necessity in Selection Systems: A Reductio-Based Derivation of D-Architecture from a Minimal Maintenance Invariant. Zenodo preprint, Version 1.0. https://doi.org/10.5281/zenodo.19342655.
2. Jung, J.H., 2026b. Operational Conditions for Selection Systems: OC-1 to OC-10 as Implementation-Level Necessities in D-Architecture. Zenodo preprint, Version 1.0. https://doi.org/10.5281/zenodo.19365452.
3. Jung, J.H., 2026c. D-Architecture Core v1.2.2. GitHub repository file. Available at https://github.com/voidafter/D-architecture/blob/main/D-Arch_Core.txt (accessed April 1, 2026).
4. Jung, J.H., 2026d. D-Architecture Index v1.4.5. GitHub repository file. Available at https://github.com/voidafter/D-architecture/blob/main/D-Arch_Index.txt (accessed April 1, 2026).
5. Jung, J.H., 2026e. D-Architecture Compressed Reference (Layer alpha). GitHub repository file. Available at https://github.com/voidafter/D-architecture/blob/main/d-arch-reference.txt (accessed April 1, 2026).
6. Jung, J.H., 2026f. D-Architecture Applied Regulation v1.0. GitHub repository file. Available at https://github.com/voidafter/D-architecture/blob/main/D-Arch_Applied_Regulation.txt (accessed April 1, 2026).
7. Jung, J.H., 2026g. Regulation Structure Argument. Framework companion page for D-Architecture. Available at https://voidafter.com/app/framework/regulation.html (accessed April 1, 2026).