Where uncertainty is reduced: upstream architecture and the performance of precautions in attack

Abstract Article 57 of Additional Protocol I requires constant care, verification of military objectives, feasible precautions aimed at avoiding incidental civilian harm and, where avoidance is not feasible, minimizing it, effective advance warning unless circumstances do not permit, an excessiveness assessment, and cancellation or suspension. Article 57(3) governs choices between military objectives offering similar advantage. This article examines those duties when the information, alternatives, and safeguards available during attacks were shaped upstream by acquisition, design, testing, or information management. A bounded two-condition test identifies architectural deficits that cannot practicably be cured within the decision window and can be characterized ex ante through specifications, tests, records, or procedures. The test defines the article’s subject, not the limits of applicable law. Two applications distinguish technical positional assurance from legal verification and show how information about cumulative degradation of civilian services affects the precautionary sequence. For States bound by Protocol I, it identifies the roles of weapons review under Article 36, implementation measures under Article 80 and access to legal advice under Article 82. It separates attribution, breach, and the consequences of State responsibility, and examines architectural records as evidence of elements in proceedings for an attack without equating institutional possession with personal knowledge. Known limitations may require restrictions, another feasible course or non-use where the applicable legal conditions cannot be met. Neither a general duty to acquire superior technology nor automatic individual criminal responsibility follows; the legal result depends on the obligation, evidence, and contemplated use.