Type: url
Original: https://arxiv.org/pdf/2609.04630
Status: generated
Summary
A theory-building preprint from Harbin Institute of Technology arguing that coding agents make digital execution elastic while judgment, verification, integration, attention and acceptance of residual risk stay bounded, so software engineering should shift from controlling human activity to governing the software change itself. It proposes Trustworthy Change as a lifecycle-spanning object with Candidate, Eligible and Accepted states, Responsibility Topology as a classification of organizations by how independent residual-risk acceptance authority is distributed (single-center or multi-anchor), and the Human–Agent Cell as the execution unit, and derives mechanisms such as authoritative engineering state and context invalidation. The authors present it as conceptualization and operationalization only, with falsifiable propositions left for future empirical study.