Research note
When did the route change—and when did anyone know?
In some projects, the evidence changes before the dashboard or decision record does. This note separates those moments so a team can reconstruct delay without inventing a single false date.
This way of reconstructing a timeline is still experimental. Independent reviewers have not yet shown that they can reach similar date ranges from the same engineering records.
Four moments
Separate the change, the evidence, the symptom, and the decision.
They may occur in a different order than expected. One may never occur. The records may support only a date range rather than an exact timestamp.
Use a range when the record demands it. “Supported at review A; unsupported by review B” is better than an exact date invented after the fact.
Three useful delays
Each interval tells you something different.
ROUTE CHANGE → SYMPTOM
Visibility delay
How long did ordinary performance indicators remain normal, ambiguous, or silent after the route changed?
EVIDENCE AVAILABLE → RECOGNIZED
Recognition delay
How long did sufficient information exist before the responsible decision process recorded its implication?
ROUTE CHANGE → RECOGNIZED
Total delay to recognition
How much time passed between the supported change in the route and the decision record catching up?
Define before dating
The dates are meaningless until the question is fixed.
What result is being protected, and how will success be judged?
Which budget, schedule, configuration, policy, and available actions define the route?
Which dated records existed at each point, and what important records are missing?
What rule supports the conclusion, and who is technically qualified to make it?
Who can approve, reverse, or reopen the decision, and was that authority available in time?
What competing account still fits the evidence, and what would distinguish it?
Illustrative example
A performance failure may be the third event, not the first.
Imagine a development team that removes an inspection feature to meet packaging constraints. The dates below are invented to show the distinction.
Design changed
The released geometry no longer permits the required in-process measurement without rework.
Evidence existed
The updated drawing and access check together supported that conclusion.
Symptom appeared
A validation run produced an unexplained result that the missing measurement could have resolved.
Decision caught up
The review formally recognized that the current verification route was no longer available.
The useful question is not who should have known. It is which evidence was available, what decisions were made during the delay, and what would prevent the same lag in the next project.
Try the method with two reviewers
Give two people the same time-ordered records without the later outcome. Ask each person to date the four moments independently, then compare their ranges. Large disagreement means the records are too weak to support a precise date. If the reconstruction does not improve a real decision, narrow or discard the method.