Lab attachment
Review that runs alongside the programme
A single review at the end catches errors when they are most expensive to correct. Lab attachment places a standing external review function next to an ongoing programme, so claims are examined as they form.
- Continuous, scoped engagement
- Read-only, laboratory-defined
- Set with the laboratory
- Advisory; the laboratory decides
Purpose
What attachment changes
In an attached engagement, the Institute maintains a live claim ledger and contradiction register for the programme. New results enter the ledger as they are produced, and each is classified according to what the evidence currently supports. When a decision approaches — a publication, a design freeze, a scale-up — the release state is already documented rather than assembled retrospectively.
Attachment does not place the Institute inside the programme. Review remains external: the Institute does not design experiments, does not analyse data on the laboratory's behalf, and does not co-author the claims it examines. The laboratory retains every decision.
Structure
How the engagement is arranged
- The programme, claim classes, and instruments in scope are listed in writing, and are not widened unilaterally.
- Read-only access to a laboratory-defined subset of records, established only after eligibility review. The laboratory sets and can withdraw the scope at any time.
- Classified, export-controlled, or embargoed material is handled under separate arrangements outside this website, or excluded from scope entirely.
- A standing review interval, plus decision-triggered review when the programme reaches a gate.
- A maintained claim ledger, a contradiction register, and a release-state statement at each gate, with dissent recorded.
- Either party may end the engagement. Records produced remain available to the laboratory.
Gating
Where the review intersects the programme
Attached review against programme gates
ConceptualThe programme proceeds through its own stages; the attached review supplies the release state at each gate. Where a stage is unsatisfied, the default is a hold and a return for repair and retest.
Layered flow showing programme stages with attached review gates, a repair-and-retest return path, and a fail-closed default at each gate.
Represents structure of reasoning, not a specific calculation.
The gate does not stop the programme; it records what the evidence permits at that point. A hold is a statement about evidence, not an instruction.
Attachment makes the release state a continuously maintained object rather than an end-of-project exercise.
Suitability
When attachment is the wrong choice
- Where there is no consequential decision ahead, a single scoped review is cheaper and sufficient.
- Where the laboratory cannot make any records available, review would have nothing to examine.
- Where the request is for endorsement rather than examination, the Institute is not the right counterparty.