Assess whether Accepted-TC-centered accounting improves measurement of agentic software work

Determine whether using Accepted Trustworthy Change as the accounting boundary improves the measurement of agentic software work compared with activity measures such as lines of code, commits, pull requests, or agent-task counts.

Background

The paper argues that scalable agent execution weakens the relationship between implementation activity and engineering value. It therefore proposes Accepted-TC accounting, which distinguishes changes that have entered the formal baseline from candidates, rejected work, abandoned work, and eligible but unaccepted changes.

The unresolved empirical issue is whether this accounting boundary provides a more valid basis for evaluating trustworthy output, resource use, lifecycle cost, and agentic software productivity than conventional activity metrics.

References

As of the freeze date, the most important open questions for this framework remain empirical: whether Responsibility Topology adds explanatory power beyond team size, whether Minimum Sufficient Specification can be identified in practice, whether Context Invalidation reduces stale-context failures, and whether Accepted-TC-centered accounting improves measurement of agentic software work.

— Software Engineering in the Agent Era From Trustworthy Change to Human Agent Software Organizations  (2609.04630 - Wang et al., 4 Sep 2026) in Section 10.3, final paragraph; Section 11.5, “Benchmark and Reporting Protocol”