Number
Equation 14
Canon 14 · Update Rule
Equation 14
A living system changes through evidence, not impulse alone.
Equation 14 defines the Update Rule for moving from the current state to the next one. Witness identifies pressure for change. Coherence controls the size of the step. Memory preserves provenance. The Causal Lattice identifies where intervention belongs. Ethical Law permits, modifies, holds, or halts the update.
Canon Identity
Equation record.
Canonical title
Update Rule
Primary form
\( \eta(\Xi)\nabla_{\mathfrak{R}}\mathcal{J} \)
Status
Codex-origin revision framework
Function
Produce the next state from witness, coherence, memory, causality, and ethics.
Current version
Canonical web edition · 2026
Equation 14
The Update Rule
$$
\boxed{
\mathfrak{R}_{t+1}
=
\mathcal{E}_{9}
\left[
\mathfrak{R}_{t}
+
\eta\left(\Xi_t\right)
\nabla_{\mathfrak{R}}
\mathcal{J}
\left(
\mathfrak{R}_{t};
\mathcal{W}_{t},
\mathcal{L}_{c},
\mathbb{M}_{\tau}
\right)
\right]
}
$$
The next reality-state \( \mathfrak{R}_{t+1} \) begins with the current state \( \mathfrak{R}_{t} \), receives a coherence-weighted correction \( \eta(\Xi_t)\nabla_{\mathfrak{R}}\mathcal{J} \), and remains contained by Ethical Law \( \mathcal{E}_{9} \).
Bounded update form
$$
\Delta\mathfrak{R}_{t}
=
\operatorname{clip}
\left(
\eta(\Xi_t)\mathbf{g}_t,
-\Delta_{\max},
\Delta_{\max}
\right)
$$
The correction vector \( \mathbf{g}_t \) is bounded by a maximum permitted change \( \Delta_{\max} \). This prevents a single observation, emotional reaction, or noisy result from rewriting the whole system at once.
Plain Language
What Equation 14 means.
The system should change when the evidence changes, but not every new signal deserves the same force. Strong, repeated, coherent witness may justify a larger correction. Weak, contradictory, or incomplete witness may justify only a note, a hold, or a small test. Every update should preserve what changed, why it changed, and what prior state remains recoverable.
Witness supplies pressure
A failure, contradiction, result, report, or new source creates the reason to review the current state.
Coherence controls step size
Stable, well-supported correction can move farther than uncertain or conflicting input.
Memory protects provenance
The earlier version remains traceable rather than disappearing under the new one.
Ethics controls permission
A technically effective update may still be modified or stopped when it creates disproportionate harm.
Terms
Every symbol in Equation 14.
\( \mathfrak{R}_{t} \)
Current state
The present configuration of the claim, system, artifact, page, protocol, or field.
\( \mathfrak{R}_{t+1} \)
Next state
The revised configuration after the update is applied and reviewed.
\( \mathcal{J} \)
Correction objective
The declared target the update is trying to improve, repair, reduce, align, or resolve.
\( \nabla_{\mathfrak{R}}\mathcal{J} \)
Correction direction
The direction in which the current state should move to improve the declared objective.
\( \eta(\Xi_t) \)
Coherence-weighted step
The size of the update as a function of current coherence and confidence.
\( \mathcal{W}_{t} \)
Current witness
The observations, records, conditions, interpretations, and corrections available at time \(t\).
\( \mathcal{L}_{c} \)
Causal Lattice
The network identifying which nodes, paths, gates, and feedback loops the update may affect.
\( \mathbb{M}_{\tau} \)
Memory Window
The relevant prior states, revisions, failures, outcomes, and retained influence.
\( \mathcal{E}_{9} \)
Ethical Mask
The containment rule deciding whether the update proceeds, narrows, delays, archives, or halts.
\( \Delta_{\max} \)
Maximum update
The largest permitted change in one revision cycle.
\( \mathbf{g}_{t} \)
Raw update vector
The proposed changes before coherence weighting, clipping, and ethical review.
Update Layers
Seven decisions inside every revision.
1
Trigger
What new witness, contradiction, failure, request, or opportunity opened the update?
2
Objective
What exactly should improve, stop, align, repair, or become clearer?
3
Direction
Which variables, words, components, nodes, rules, or paths need to change?
4
Magnitude
How large should the update be given coherence, confidence, reversibility, and risk?
5
Containment
What ethical, technical, legal, or practical boundary limits the change?
6
Version
How is the old state preserved, named, dated, and recoverable?
7
Review
What result, date, threshold, or consequence will determine the next revision?
Step Size
Not every update deserves the same magnitude.
TRACE
Record only
Preserve the new witness without changing the active system yet.
PATCH
Small correction
Repair a typo, link, label, selector, parameter, status, or narrow factual error.
REVISION
Moderate change
Update a section, method, component, policy, protocol, or claim while preserving the larger architecture.
REBUILD
Structural change
Replace major architecture when local repair cannot resolve global incoherence.
HOLD
Delay action
Preserve the proposed change while waiting for evidence, consent, review, materials, or capacity.
ROLLBACK
Return to prior state
Restore an earlier version when the current update creates instability, error, or harm.
Coherence-Weighted Learning Rate
Confidence changes how far the system moves.
$$
\eta(\Xi)
=
\eta_{\min}
+
\left(
\eta_{\max}-\eta_{\min}
\right)
\Xi^{p}
$$
A normalized coherence value \(0\leq\Xi\leq1\) can map to a step between \( \eta_{\min} \) and \( \eta_{\max} \). Exponent \(p\) controls how quickly the update grows. This is a configurable implementation, not a universal law.
Low coherence
Use small, reversible tests and preserve uncertainty.
Medium coherence
Apply bounded changes with scheduled review and rollback.
High coherence
Permit broader revision when evidence, consequence, and ethics align.
Ethical halt
Even high coherence cannot override a zeroed ethical condition.
Versioning
The previous state remains part of the truth.
$$
\mathbb{M}_{\tau}
\leftarrow
\mathbb{M}_{\tau}
\cup
\left\{
\mathfrak{R}_{t},
\Delta\mathfrak{R}_{t},
\rho_t,
t
\right\}
$$
Before activation, the memory window receives the prior state, the update, the reason \( \rho_t \), and the time. Revision without provenance becomes replacement without accountability.
State before
Preserve the exact file, object condition, claim, setting, or policy before change.
Reason
Record the witness, contradiction, failure, request, or objective that triggered revision.
Diff
Identify what changed, what stayed, and what was retired.
Result
Record expected and actual consequences after activation.
Rollback path
Keep a known method for returning to the earlier state when practical.
Remnant
Carry forward the strongest surviving structure and correction.
Update Protocol
From witness to the next version.
Save the exact file, record, artifact condition, claim, configuration, or rule before editing.
Identify the new witness, contradiction, failure, request, risk, or opportunity.
State what should improve, align, stop, become safer, or become more accurate.
Use the Causal Lattice to identify direct changes, dependencies, gates, and possible side effects.
Compare evidence, intention, language, structure, action, timing, consequence, memory, and ethics.
Select Trace, Patch, Revision, Rebuild, Hold, or Rollback.
Check consent, dignity, truth, proportion, responsibility, correction, stewardship, and halt conditions.
Use the smallest reversible implementation capable of testing the correction.
Compare expected and actual results, including failures, side effects, and affected voices.
Keep the update, revise it again, return to the prior state, or carry forward only the strongest surviving structure.
Worked Example 1
Repairing the shared mobile navigation.
Every theory page includes a visible mobile menu button and shared navigation markup.
The button appears but does not open the menu across multiple pages.
The likely shared nodes are theory.js, theory.css, selector names, open-state classes, and responsive rules.
Patch the shared assets rather than rewriting every page.
Add toggle, close, Escape, resize, link-click, and accessibility behavior; add matching mobile open-state CSS.
Test touch, keyboard, desktop resize, page navigation, focus, and aria-expanded on several pages.
Keep the existing header structure while repairing only the shared behavior layer.
Worked Example 2
Updating an artifact after structural failure.
A wearable artifact is registered as complete and active.
A joint loosens during normal use.
Status changes to Repair, Hold, Archived, or Retired before further release.
Joint geometry, material choice, tension, cure time, wear, and inspection frequency are reviewed.
Repair one prototype and test the revised structure before applying it to the whole series.
Preserve photographs, failure condition, repair method, date, and later result.
The successful form, name, materials, and symbolism continue while the failed connection changes.
Worked Example 3
Updating an overstated field claim.
“The artifact generated a unique field that caused the compass deflection.”
Known magnets inside the artifact provide a conventional mechanism.
The observation remains strong, but the uniqueness and causal certainty are weak.
Reduce certainty, preserve direct witness, and identify the leading known explanation.
“The compass deflected near the magnet-containing artifact under the recorded conditions.”
Isolate or remove magnetic components and compare any residual response.
The witnessed deflection survives; the unsupported mechanism does not retain authority.
Worked Example 4
Updating a frequency preset.
A four-deck preset has fixed gains, timing, routing, and symbolic labels.
Listeners report that one carrier is harsh or the master level is too low.
Compare device, room, volume, carrier balance, listener reports, and actual output.
Change one gain range or master default at a time rather than altering the entire field.
Preserve obvious mute and stop controls, safe level guidance, and listener consent.
Keep the earlier preset and label the revised one with date, purpose, and changes.
Preserve the strongest carrier relationships while correcting usability and level problems.
Scope
What Equation 14 does and does not do.
It does
Organize bounded, coherence-weighted, versioned revision across claims, systems, artifacts, pages, and protocols.
It does not
Guarantee that the chosen objective, gradient, weight, or next state is correct.
It supports reversibility
Updates should be testable, reviewable, and recoverable where practical.
It supports continuous learning
The next state remains provisional and open to later witness rather than becoming permanently protected.
Failure Modes
How an update becomes instability or erasure.
Impulse update
A single emotional reaction or unverified report rewrites the system immediately.
Objective drift
The system optimizes a convenient metric while abandoning the original purpose.
Runaway step size
A large update creates more damage than the original problem.
No version history
The prior state, reason, and correction path disappear.
Update lock
A high coherence score is treated as proof that the system should never change again.
Ethics bypass
Efficiency, engagement, profit, speed, or internal consistency overrides consent and consequence.
Ethical Boundary
Not every improvement is ethically permitted.
$$
\Delta\mathfrak{R}_{t}^{\,\mathrm{allowed}}
=
\mathcal{E}_{9}
\left[
\eta(\Xi_t)
\nabla_{\mathfrak{R}}\mathcal{J}_t
\right]
$$
An update may improve a chosen objective while damaging people, privacy, dignity, truth, access, or the larger field. Ethical Law evaluates the correction itself, not merely the result it promises.
Consent before deployment
Affected people should understand significant changes when their participation or data is involved.
Proportionate change
Use the smallest update capable of meeting the legitimate need.
Responsibility after deployment
Name who monitors outcomes, receives reports, repairs harm, and triggers rollback.
Visible correction
When a public system or claim changes materially, the update should not conceal the earlier error.
Practice
Equation 14 update card.
What is the current state?
Preserve the exact version before editing.
What triggered revision?
Name the witness, contradiction, failure, risk, request, or opportunity.
What is the correction objective?
State what should improve, stop, align, or become safer.
How large should the update be?
Choose Trace, Patch, Revision, Rebuild, Hold, or Rollback.
What might the update affect?
Map dependencies, gates, side effects, people, and rollback conditions.
What survives into the next state?
Preserve provenance, useful structure, correction, and Remnant.
Canon Connections
The Update Rule turns evaluation into revision.
Equation 2
Witness
New witness provides the trigger and evidence for change.
Open Eq. 2 →Equation 5
Remnant
The strongest surviving structure moves into the next state.
Open Eq. 5 →Equation 9
Ethical Law
The ethical mask permits, modifies, holds, archives, or halts the update.
Open Eq. 9 →Equation 12
Causal Lattice
The lattice identifies leverage points, dependencies, and side effects.
Open Eq. 12 →Equation 13
Coherence Ξ
Coherence controls the confidence and magnitude of revision.
Open Eq. 13 →Equation 15
Learning Kernel
Repeated updates become an accumulated learning process.
Open Eq. 15 →Revision Record
Canonical web edition.
This edition formalizes Equation 14 as a coherence-weighted, ethically contained, version-preserving update rule. It defines bounded step size, six update states, seven update layers, a ten-step protocol, four worked examples, rollback logic, and the requirement that correction preserve provenance.
Continue Reading
Next, turn repeated revision into a learning process.
Next: Equation 15
The Learning Kernel and accumulated adaptation across repeated updates.
Previous: Equation 13
Return to Coherence Ξ.
Memory Theory
Study version history, weighting, revision, and return.
Canon 0–17
Return to the full equation index.
Notation Guide
Review gradients, learning rates, clipping, state transitions, and deltas.
Glossary
Review the working definitions used on this page.