The Log
Eight pieces on the infrastructure of engineering decisions — how it works, where it fails, and why it matters long after delivery.
When class raises a query on a structural detail, it routes — inevitably — to the engineer who designed it. This is not a workflow decision. It is a structural tax on seniority, invisible in any project plan, that accumulates across hundreds of comments on a newbuild.
Every ship design office maintains a comment register. Ask the PM and the lead engineer whether it is current. Neither will say yes. Your register tracks the state of comments as logged — not as they stand. The live status lives in email. This is structural, not a discipline failure, and it is the origin of most correspondence disputes.
The liability a design office carries for a newbuild decision runs ten, fifteen, twenty years into service. The correspondence record that would explain that decision typically survives three. This is not a compliance gap — it is a structural mismatch between how long the industry holds you accountable and how long your records last.
Every major party in a newbuild project runs dedicated infrastructure for managing its records. The design office — which generates most of the project's technical correspondence and carries professional liability for every decision behind it — typically runs email, a shared drive, and a spreadsheet. This piece names the structural problem that asymmetry creates.
ClassNK published its Guidelines for 3D Model-Based Approval in May 2026, alongside the NK-PASS submission platform and a ten-company pilot test behind it. The guidelines define four areas — scope, submission units, revision handling, and the 3D–2D relationship during transition — and that last area is the one most likely to be misread by offices still working in 2D.
A complete account of the class comment sheet — what it contains, how dispositions work, and what open comments mean for every review stage that follows. The mechanics are standard. The part that compounds over a career is the disposition strategy.
Approval cycle time is not a function of design quality. The controllable bottleneck is administrative: routing, acknowledgment, follow-up, comment reconciliation. These activities require no engineering judgment, yet they consume engineering time — and most offices have made them structurally invisible.
AI tools are entering ship design offices faster than the accountability frameworks governing them. The question is not whether AI can draft a class comment disposition — it can. The question is whether the design office's approval chain permits it to issue one, and the answer is unambiguous.