Give-Way Inaction in Approach Channels Produces 0.251nm Average CPA — the Tightest Geometry in 1.97 Million Encounters.

The Setup

COLREGS Rule 17 assigns the give-way vessel the obligation to act first. When it does not — when it holds course and speed despite the encounter — the stand-on vessel faces a choice governed by Rule 17(a)(ii): take independent action to avoid collision. That moment is the handoff. The encounter record captures it as a rule17_handoff_trigger.

Two distinct failure modes produce a handoff. giveway_inaction: the give-way vessel has not maneuvered and the encounter geometry is deteriorating. extremis: the situation has progressed to the point where both vessels must maneuver immediately regardless of role assignment.

Both triggers operate across three encounter contexts: open sea, approach channel, and anchorage. Across 1,973,422 encounters tracked in the 45 days ending June 16, each of the six trigger-context combinations produces a distinct average minimum range at closest point of approach.

The worst is approach-channel give-way inaction: 0.251nm average CPA.

The Chain

The six trigger-context pairings, ordered by average minimum range:

| Trigger | Context | Encounters | Avg Min Range | |---|---|---|---| | giveway_inaction | approach | 88,694 | 0.251 nm | | giveway_inaction | open_sea | 160,852 | 0.273 nm | | giveway_inaction | anchorage | 33,483 | 0.271 nm | | extremis | open_sea | 472,124 | 0.317 nm | | extremis | approach | 250,019 | 0.360 nm | | extremis | anchorage | 108,814 | 0.424 nm |

The pattern is not what the trigger types alone would predict. Extremis — the more severe failure mode by definition — consistently produces wider average CPA than give-way inaction. The 472,124 open-sea extremis encounters average 0.317nm; the 160,852 open-sea give-way inaction encounters average 0.273nm. Same context, different trigger, 16% worse geometry in the inaction case.

The course-change data partially explains this. Extremis events produce large mutual maneuvers: both vessels in open-sea extremis average 55.5° and 55.6° course changes. Give-way inaction events produce smaller mutual course changes: 51.6° and 51.7° in open sea.

The extremis maneuvers are bigger. They produce wider average CPA in open water because two large simultaneous course changes in unconfined water tend to diverge. Both vessels move away from each other's original track. The geometry opens.

In approach channels, the dynamic reverses. Extremis in approach channels averages 0.360nm CPA — better than open-sea extremis — because the channel walls constrain the geometry. Two vessels executing large course changes inside a bounded lane are pushed toward parallel vectors, not crossing paths. The channel is a constraint that helps.

Give-way inaction in approach channels averages 0.251nm — the worst combination in the dataset — because the stand-on vessel facing an unresponsive give-way cannot move freely. The channel boundary limits its escape path. The give-way vessel has not acted. The stand-on's only available course changes are constrained by the very geometry that helps during extremis mutual maneuver. The stand-on ends up with a limited escape corridor and a vessel that is not clearing it.

Total handoff encounters in the 45-day window: approximately 1,115,000 — 56.5% of all encounters. The majority of close-quarters events in the tracked fleet involve a Rule 17 handoff.

The Implication

The directional claim: give-way failure in approach channels is the highest-geometry-risk encounter configuration in the tracked fleet, and it is distinct from the more-discussed extremis failure mode. The 88,694 giveway_inaction/approach encounters in 45 days represent a geographically concentrated failure class — approach channels are finite, mappable, and monitored. This failure mode is more predictable before it reaches handoff than open-sea encounters.

The implication for triage: extremis is more numerous in approach channels (250,019 events vs. 88,694 inaction events), but inaction produces 30% worse CPA. Alert queues that prioritize extremis events by count will surface the more common failure while deprioritizing the geometrically worse one.

What to Watch

The ratio of giveway_inaction to extremis within approach contexts is the prioritization indicator. The current 45-day split: 88,694 inaction vs. 250,019 extremis — inaction is 35.5% of approach-channel handoffs. If that share drifts above 40%, give-way compliance in tracked approach channels is deteriorating as a distinct behavioral pattern, not merely tracking encounter density.

Average course change at inaction events is the secondary signal. If the 42.4°/42.5° mutual course change for approach inaction rises toward the open-sea inaction levels (51.6°/51.7°), it indicates the stand-on vessel is executing larger emergency maneuvers — a sign the handoff is occurring later in the encounter, with less time to act.

Limitations

Average minimum range is a geometric outcome, not a collision probability. The 0.251nm figure does not imply any specific incident rate. No collision or near-miss incident data exists in this dataset to establish an outcome base rate against encounter count.

Context tags (open_sea, approach, anchorage) are assigned algorithmically and are not field-verified. Encounter boundaries near navigable channel edges may produce misclassifications. A vessel technically in an approach channel tagged as open_sea would shift the approach-channel averages if reclassified.

The 45-day window captures a specific seasonal and traffic-density period. Encounter density and give-way compliance rates may vary with season, port congestion cycles, and fleet composition changes.

Data as of 2026-06-16. Sources: pairwise_encounter (rule17_handoff_trigger, context_tag, min_range_nm, max_course_change_a_deg, max_course_change_b_deg), 45-day rolling window, n=1,973,422 total encounters, n=~1,115,000 handoff events.