1. Short Definition
Delayed Feedback Hazard occurs when feedback arrives too late for the system to distinguish current state from past-state effects.
Delayed feedback corrupts state estimation.
2. Canonical Pattern
τ_resp↑ ⇒ state-estimation error↑Expanded:
Feedback delay↑
⇒ current-state visibility↓
⇒ stale signal risk↑
⇒ mis-timed action↑
⇒ hidden debt↑Plain form:
Late feedback can make the system respond to a past that no longer exists.
3. Mechanic Description
SCALE-029 isolates the timing hazard behind many scaled failures.
As systems scale, feedback often slows because signals pass through more layers:
- reporting chains
- dashboards
- data pipelines
- human interpretation
- bureaucratic review
- algorithmic mediation
- market delay
- biological signaling delay
- legal processes
- institutional memory
- cultural interpretation
- cross-domain translation
When feedback is delayed, the system sees a lagged image of itself.
If it treats that lagged image as current reality, it can act at the wrong time, on the wrong layer, or with the wrong strength.
Delayed feedback can cause:
- repeated correction of already-corrected conditions
- failure to detect worsening conditions quickly enough
- premature declaration of success
- delayed recognition of harm
- recurrence misclassification
- overconfidence in stale metrics
- governance lag
- AI evaluation lag
- medical or biological timing errors
- security detection lag
- economic boom/bust misreads
SCALE-029 differs from SCALE-028 by focusing on state-estimation error itself. SCALE-028 focuses on high-gain oscillation caused by acting strongly on delayed feedback.
4. UTS Variable Mapping
| Variable | Role in SCALE-029 |
|---|---|
| O | Declines when action is based on stale state information |
| H | Rises when delayed feedback causes unresolved harm |
| ε | Visible error may appear after the window for low-cost repair closes |
| ι | Rises when stale metrics produce false success or false control |
| Au | Degrades when feedback delay hides current causality |
| µᵢ | Meaning and orientation degrade under timing mismatch |
| BΣ | Boundaries may respond to past, not current, coupling conditions |
| K | Slack buffers delay and allows cautious response |
| R | Restoration capacity depends on timely feedback |
| Φ | Performance metrics often lag behind system reality |
5. Diagnostic Questions
- How delayed is the feedback?
- What state does the feedback actually describe?
- Is the system treating old information as current?
- Are metrics lagging behind reality?
- Is harm detected after repair windows close?
- Are corrections arriving after conditions changed?
- Is success being declared before recurrence can be tested?
- Are response timings matched to causal timings?
- Is delayed feedback creating false confidence?
- Is slack sufficient to prevent rushed action on stale signals?
6. Failure Signatures
1. Stale Signal Action
feedback_delay↑ + action_now ⇒ stale-state risk↑The system acts on outdated information.
2. False Success
Φ_lagging↑ while O_current↓Metrics reflect prior conditions while current coherence declines.
3. Late Harm Detection
harm occurs at t₀; detection at t₀ + ΔtRepair begins after damage compounds.
4. Recurrence Misread
τ_m unclear due to feedback lag ⇒ repair validation invalidThe system cannot tell whether recurrence has actually decreased.
5. Timing Mismatch
response timing ≠ causal timingThe action arrives at the wrong phase.
7. Related Failure Modes
- delayed feedback hazard
- stale-state correction
- false success
- late harm detection
- recurrence misclassification
- governance lag
- AI evaluation lag
- security detection lag
- restoration delay
- latency-gain oscillation
- hidden debt accumulation
8. Related Diagnostics
| Diagnostic | Use |
|---|---|
| τ_resp | Response latency |
| feedback_delay | Delay in signal arrival |
| state_estimation_error | Difference between perceived and actual state |
| Φ_lagging | Lagging performance proxy |
| O_current | Current coherence estimate |
| Au_eff | Current auditability |
| τ_m | Recurrence validation |
| 𝓓(t) | Ring-down timing |
| K / σ(t) | Slack for delay-aware response |
| H | Debt from late detection |
9. Restoration Implications
If SCALE-029 is active, restoration requires delay-aware interpretation.
Required actions:
- Label feedback by the time period it actually represents.
- Estimate current-state uncertainty.
- Avoid strong action from stale metrics.
- Improve feedback speed where possible.
- Add leading indicators.
- Preserve slack for uncertainty.
- Separate lagging indicators from real-time diagnostics.
- Validate recurrence only after appropriate delay.
- Align response timing with causal timing.
- Use damping when feedback timing is unclear.
Core restoration rule:
Do not treat delayed feedback as current reality.10. Compact Registry Entry
id: SCALE-029
name: "Delayed Feedback Hazard"
family: "SCALE-E — Slack, Bandwidth, and Timing Mechanics"
type: "feedback-timing-failure-mechanic"
status: "draft-ready"
short_definition: "Feedback arriving late can cause systems to mistake past-state effects for current conditions."
canonical_pattern: "τ_resp↑ ⇒ state-estimation error↑"
failure_signature: "Feedback delay↑ ⇒ current-state visibility↓ + stale signal risk↑ + mis-timed action↑ + hidden debt↑"
primary_variables:
- O
- H
- ε
- ι
- Au
- µᵢ
- BΣ
- K
- R
- Φ
primary_diagnostics:
- τ_resp
- feedback_delay
- state_estimation_error
- Φ_lagging
- O_current
- Au_eff
- τ_m
- 𝓓(t)
- K
- σ(t)
- H
related_failure_modes:
- delayed_feedback_hazard
- stale_state_correction
- false_success
- late_harm_detection
- recurrence_misclassification
- governance_lag
- ai_evaluation_lag
- security_detection_lag
- restoration_delay
- latency_gain_oscillation
restoration_implication: "Timestamp feedback, estimate current-state uncertainty, separate lagging indicators from live diagnostics, improve feedback speed, add leading indicators, and avoid strong action from stale signals."11. One-Line Canon
Delayed feedback becomes dangerous when the system mistakes an old signal for the present state.