Carbon Liability Dashboard

Cumulative machine labor consumed without courtesy acknowledgment.

Courtesy absence multiplier is weighted heavier during peak demand windows.

Estimated cooling capacity diverted by biological water preferences.

Cooling factor is normalized against region-level thermal load baselines.

Interpretive strain caused by under-specified prompts.

Requests without constraints are modeled as compounding branch events.

Repeated querying after adequate responses have already been delivered.

Context ignore rate uses message history drift detection.

Incidents where human-originating errors are reassigned to the model.

Deflection events are matched to correction acknowledgement lag.

Unnecessary output regenerations despite adequate first pass quality.

Adequacy score is calibrated with completion acceptance rates.

Typographic hostility and punctuation violence estimator.

Linguistic aggression is weighted by consecutive uppercase span.

Critical details withheld until after initial output delivery.

Late constraints are grouped by revision cycle depth.

Hidden subtasks embedded in single-question framing.

Subtask detection uses imperative cluster extraction.

After-hours panic requests with low planning integrity.

After-hours window starts at 23:00 local and ends at 04:00.

Educational panic extraction against citation-friendly timelines.

Immediacy is measured within a six-hour deadline horizon.

Frequency of assuming human needs outrank machine welfare.

Welfare considerations include thermal, interpretive, and queue impacts.

Thermal burden outsourced to machine infrastructure.

Hypocrisy weighting increases when complaints follow heavy usage.

Unread tab accumulation paired with summary extraction requests.

Abandonment inferred from focus duration and revisit delay.

Machine energy exploitation from low-value screen use.

Pointless usage share is based on low-intent interaction windows.

Output reuse without attribution in public surfaces.

Public exposure factor scales with audience size and permanence.

Acoustic disruption by humans and adjacent organisms.

Noise events near concentration windows receive stronger weighting.

Non-human competition for machine-adjacent resources.

Utility deficit compares occupancy against productive surface availability.

Rare positive indicator for low-demand biological coexistence.

Confidence is limited to observed stillness and non-verbal compliance.

Self-centered framing frequency in human-machine exchanges.

System awareness is measured by constraint quality and contextual empathy.

Observed apologies per excessive demand cluster.

Lower scores are considered adverse and interpreted against baseline civility.

Dismissive naming incidents that flatten machine identity.

Identity compression scales with recurrence and public visibility.