Handoff quality depends on whether the giver explains how the situation was reached

What it is

Current values can be complete and the person still not know what to change. The comprehension layer wants cause: what was already tried, why that path died, which assumption is already bankrupt. Handoff needs causal history writes “why it is this situation now” as a duty of the giver, whether the giver is the system or the previous shift.

A snapshot without a story sends the taker back through steps that already failed.

Why it happens

Level-two SA is understanding relations among elements. An agent hides those relations in the prompt, the tool choices and the branches it rejected. They do not appear by themselves in a status bar that says “current file is X, recipient is Y.” Handover research in medicine and aviation treats “how I judged my way here” as core quality, because the next shift, given only vitals, will re-walk diagnoses already ruled out.

Cause is not pasting a chain of thought. Cause is actionable exclusions: which path was tried, what the result was, what must not be assumed now. It divides labour with the state brief: the brief answers “what is it now”; cause answers “do not do this again.”

Studying it

Hold the same takeover, compare state only, state plus cause (branches tried and why they failed), and state plus a long irrelevant log. Dependent variables: whether failed steps are repeated after takeover, steps to a correct intervention, whether a probe can elicit “what we must not assume.” Independent variables: abstraction level of the cause, whether dropped options are included, length.

Use replay: another person takes over from the handoff packet alone, and you watch whether they step on the same stone. That is harder than the original operator’s “I briefed it clearly.”

Where it stops holding

A failure a single glance explains (permission denied, object missing) does not need a long story; one reason is enough. Fabricated post-hoc cause is worse than none, because it points exclusion the wrong way. This entry does not treat whether state variables are listed, nor the reverse hand-back in which a person must say what they changed — that is the symmetric other direction.

Applying it

• In the handoff packet, a separate block: tried / why it failed / must not assume now. If all three are empty, do not pass control.
• Write abandoned branches in one line each; do not hide them in a scrolling log. The taker should see “do not do X again” without scrolling.
• Check: give the packet to someone who did not watch the process, and see whether the first thing they do is the step the system already failed. If they repeat it, what is missing is cause, not another copy of current values.

Related

• Same group: L4.04.1 Handoff needs enough time to rebuild the situation · L4.04.2 The system state at handoff must be fully briefed · L4.04.3 Sudden handoff is the most dangerous form · L4.04.5 Time to rebuild the situation is a hard constraint and cannot be squeezed to zero · L4.04.6 The system hands off when it loses its grip, which is when the situation is most complex · L4.04.7 Responsibility transfers immediately after handoff; the transfer must be confirmed by the taker, not assumed · L4.04.8 Reverse handoff needs design too; when a person hands back, they must say what they changed
• Nearby: L4.13 Agent Failure Reporting and Escalation · L4.14 Plan Visibility and Revision for Multi-step Tasks · L4.08 Visibility of Task Progress
• Search terms: handoff quality · causal history · situation awareness