Before / After Behavior Monitor¶
Monitoring protocol — instantiates Compensation-Aware Safeguard Design
Measures the risk-relevant behaviors before and after a safeguard so offset shows up as a change in conduct, not only in the final harm rate.
Before / After Behavior Monitor is the measurement instrument of the archetype: it freezes a record of how people actually behave before a safeguard is scaled, instruments the same behaviors continuously after, and reports the difference as a distribution shift. Its defining commitment is that it watches conduct, not outcomes — following distance, leverage, usage intensity, maintenance cadence — because those leading indicators move long before the lagging harm rate does. That lets it detect a risk budget being re-spent while the protected failure mode still looks improved, which is exactly the window in which the archetype says offset must be caught to be correctable.
Example¶
A long-haul trucking fleet is fitting forward-collision-avoidance braking across 800 tractors. The obvious success metric is rear-end collisions, but those are rare and lagging, so the safety team stands up a behavior monitor first. Using existing telematics, they capture a two-month baseline of the behaviors the system could tempt a driver to relax: headway (following distance in seconds), following-distance time below two seconds, average speed over posted limits, and hard-brake events per thousand miles.
After the braking system goes live on a staggered schedule, the same telemetry keeps flowing. Within six weeks the equipped tractors show median headway shrinking from 2.4 to 2.0 seconds and time-under-two-seconds up by a fifth, while the unequipped control group holds flat. No crash rate has moved yet — but the monitor has already surfaced the reallocation: drivers are quietly cashing part of the new margin into closer following. That signal, not a collision, is what triggers the fleet to add a headway coaching nudge before the offset shows up as a wreck.
How it works¶
- Freeze a baseline on the tempted behaviors. Using the premortem's predicted pathways, pick the specific conduct the safeguard could relax and record its full distribution over a pre-deployment window.
- Stand up an observability channel independent of harm. Instrument those behaviors continuously — telematics, logs, sensor traces — so detection does not wait for an incident.
- Compare distributions, not just averages. After deployment, test the whole distribution against baseline; drift in the tail (the closest followers, the highest leverage) matters more than the mean.
- Attribute the shift. Isolate the safeguard's effect with a staggered rollout or control group so seasonal or demand-driven drift is not mistaken for compensation.
Tuning parameters¶
- Baseline window length — a longer pre-deployment window is more stable and less fooled by seasonality, but delays the rollout and risks a stale baseline if conditions change.
- Metric lead vs. lag — leading behavioral indicators (headway) fire early but are noisy; lagging harm indicators (crashes) are unambiguous but arrive too late to correct.
- Attribution design — a naive before/after is cheap but confounded; a staggered rollout with a control group isolates the safeguard at the cost of coordination.
- Observation salience — overt measurement is easy to deploy but can itself change behavior; unobtrusive measurement avoids that but raises privacy and consent stakes.
- Alert threshold — how much drift counts as a signal. Tight thresholds catch offset early but raise false alarms; loose ones wait for certainty and lose the correction window.
When it helps, and when it misleads¶
Its strength is turning "did behavior change?" from anecdote into a measured distribution shift, and doing so on leading indicators — so offset is visible while it is still a change in conduct rather than a change in the casualty list.
Its honest limit is that it measures behavior, not net safety: a real shift toward the boundary may still leave the system safer overall, and the monitor alone cannot say — it hands that verdict to the dashboard. Worse, the act of overt monitoring can itself suppress the very offset it means to detect, so a reassuring "after" period may be an artifact of being watched rather than genuine care.[n1] The classic misuse is reading a flat harm rate as proof of no compensation while the leading behavioral indicators drift unmeasured. The guarding discipline is to instrument leading conduct, keep measurement as unobtrusive as ethics allow, and route the signal onward for a net judgment rather than declaring victory on the harm rate.
How it implements the components¶
baseline_risk_behavior_profile— its first job is to freeze the pre-safeguard distribution of the behaviors the safeguard could tempt, the reference every later comparison hangs on.risk_budget_reallocation_signal— the flagged drift toward the boundary is the early-warning signal that the freed margin is being re-spent.offset_observability_channel— it stands up and maintains the instrumented feed that measures conduct continuously and independently of final harm.
It computes no net safety figure — that is the Safety-Gain Offset Dashboard — and it installs no counter-incentive, which the Exposure Cap or Rate Limiter and Shared Downside or Deductible Rule provide. It observes; it does not intervene.
Related¶
- Instantiates: Compensation-Aware Safeguard Design — it supplies the measured behavioral evidence the whole compensation-control loop depends on.
- Consumes: Risk Compensation Premortem — its predicted offset pathways tell the monitor which behaviors to instrument.
- Sibling mechanisms: Risk Compensation Premortem · Safety-Gain Offset Dashboard · Shared Downside or Deductible Rule · Use-Conditioned Protection Policy · Exposure Cap or Rate Limiter · Adaptive Safeguard Recalibration Gate · Post-Safeguard Incentive Audit
Editorial Notes¶
Form Classification¶
Form family: Monitoring, Sensing & Alerting
Rationale: Measures the risk-relevant behaviors before and after a safeguard so offset shows up as a change in conduct, not only in the final harm rate, making its operative form repeated observation of actual state that emits measurements, status, or alerts.
Independent corroboration: The frozen evidence defines Before / After Behavior Monitor as 'Measures the risk-relevant behaviors before and after a safeguard so offset shows up as a change in conduct, not only in the final harm rate', so its operative form is Monitoring, Sensing & Alerting.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Psychology
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Behavioral psychology measures observable conduct before and after environmental contingencies or safeguards change.
Related originating lineages:
- Behavioral Economics — Risk compensation and moral-hazard research predict offsetting behavioral adaptation.
- Statistics & Experimental Design — Longitudinal measurement and unobtrusive observation help separate change from noise and Hawthorne effects.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Independent reviewer agreement; high confidence.
Notes¶
[n1] The Hawthorne effect — the tendency of people to alter their behavior when they know they are being observed. For a behavior monitor this cuts both ways: overt measurement can mask the offset it is meant to catch, which is why unobtrusive channels are preferred where they are ethical. ↩