A Probe is a measurable signal used to verify whether a recovery mode is working.
Without probes, governance becomes narrative. With probes, it becomes testable.
Every Civilisation OS run must define at least 3 probes:
- Short-horizon (days–weeks) — leading indicator
- Mid-horizon (weeks–months) — conversion indicator
- Long-horizon (months–years) — outcome integrity indicator
Probes must have:
- a metric definition
- data source
- sampling cadence
- pass/fail thresholds
- risk of gaming (and anti-gaming guardrails)
Probe Classes (What We Measure)
Class A — Depth Probes (D)
Measures real foundation and internal model strength.
Class B — Load Probes (L)
Measures stress pressure and constraint intensity.
Class C — Repairability Probes (R)
Measures ability to detect error, correct, and learn.
Class G — Governance Integrity Probes
Measures truth systems, auditability, and feedback health.
Class I — Incentive Alignment Probes
Measures gaming vs real outcome correlation.
Class P — Production Resilience Probes
Measures fragility, redundancy, and recovery time.
Class Cx — Constraint Probes
Measures resource pressure, overshoot, and regeneration.
OS-1 Education OS — Probe Registry
D-Probes (Depth)
E-D1 Explain-Back Accuracy
- Definition: learner explains concept in own words with causal reasoning
- Pass: ≥ 80% accuracy across 5 random items
- Fail: ≤ 60%
E-D2 Novel Problem Transfer Rate
- Definition: success on unseen problem types
- Pass: +20% improvement over 4–8 weeks
- Fail: flat or negative after intervention
E-D3 Error Type Shift
- Definition: % errors that are conceptual vs careless
- Pass: conceptual error % drops steadily
- Fail: conceptual errors persist unchanged
L-Probes (Load)
E-L1 Sleep and Recovery Stability
- Pass: stable sleep window for 14 days
- Fail: chronic sleep debt persists
E-L2 Attendance / Consistency
- Pass: ≥ 90% planned study sessions executed
- Fail: avoidance increasing
R-Probes (Repairability)
E-R1 Self-Correction Frequency
- Definition: learner identifies and fixes own errors without prompting
- Pass: increases week-to-week
- Fail: remains near zero
E-R2 Retest Willingness
- Definition: willingness to re-attempt after failure
- Pass: retest rate rises
- Fail: refusal persists (identity collapse risk)
E-R3 Voluntary Practice Time
- Pass: increases without external pressure
- Fail: effort only under coercion
OS-2 Governance OS — Probe Registry
G-Probes (Truth / Integrity)
G-G1 Audit Variance
- Definition: difference between reported metrics and independent audits
- Pass: variance decreases each cycle
- Fail: variance grows or audits blocked
G-G2 Bad News Flow Rate
- Definition: volume of surfaced issues / incidents / defects
- Pass: rises initially then stabilises (truth returning)
- Fail: drops to zero during visible decline (suppression)
G-G3 Correction Cadence
- Definition: time from detected failure to implemented correction
- Pass: cadence becomes predictable and shortens
- Fail: “review theatre” with no changes
I-Probes (Incentives)
G-I1 Metric–Outcome Correlation
- Pass: correlation improves
- Fail: metrics improve while outcomes worsen
G-I2 Gaming Incidence
- Definition: detected gaming behaviors per cycle
- Pass: decreases after incentive redesign
- Fail: increases despite “reforms”
Legitimacy Probes
G-Leg1 Trust Index
- Pass: trust improves across time windows
- Fail: trust collapses, compliance drops, coercion rises
G-Leg2 Voluntary Compliance Rate
- Pass: compliance improves without enforcement
- Fail: compliance only maintained by force
OS-3 Production/Technology OS — Probe Registry
P-Probes (Resilience)
P-R1 Mean Time to Recovery (MTTR)
- Pass: MTTR decreases after redundancy/maintenance
- Fail: MTTR worsens
P-R2 Single Point Failure Count
- Pass: decreases systematically
- Fail: unchanged
P-R3 Maintenance Backlog
- Pass: backlog shrinks steadily
- Fail: backlog grows
Output Integrity Probes
P-O1 Output Stability Under Shock
- Pass: output volatility reduces
- Fail: repeated collapses
OS-4 Constraint OS — Probe Registry
Cx-Probes (Constraint Pressure)
Cx-L1 Resource Intensity
- Definition: resources used per unit output
- Pass: declines
- Fail: rises
Cx-L2 Price / Volatility Index (for critical inputs)
- Pass: volatility stabilises
- Fail: volatility increases
Overshoot / Regeneration Probes
Cx-R1 Regeneration Rate vs Extraction Rate
- Pass: regeneration ≥ extraction (sustained)
- Fail: extraction > regeneration (persistent)
Cx-R2 Ecological Debt Trend
- Pass: debt decreases
- Fail: debt increases
Time Irreversibility Probes
Cx-T1 Feasible Set Shrink Rate
- Definition: how quickly options disappear
- Pass: shrink rate slows
- Fail: shrink rate accelerates
Mandatory Probe Set Template (for every run)
Every run must include:
- One R-probe (repairability)
- One truth/integrity probe (governance reality contact)
- One long-horizon constraint probe (sustainability)
This prevents “paper recovery” where systems look improved but are collapsing.
Anti-Gaming Rules (Probe Governance)
A probe is invalid if:
- it is easily gamed without changing reality
- it has no audit path
- it is measured too infrequently to detect drift
Every probe must declare:
- gaming risk (low/medium/high)
- an audit method
- a secondary confirmation probe