It only looks like nothing at the time.

One event, three ages

At age zero it is an instruction, or something that resembles one. Contract Week 6 takes apart who can actually give one and why the person standing on the slab may not be able to, and there is nothing to add to that.

At age one it is a variation: a change to the scope with a valuation attached, argued in commercial terms, settled by people who were not present when it happened. Contract Week 7 owns what that is.

At age two it is a claim, which needs entitlement, causation and quantum, and is assembled from records made for other reasons. That is Claims Week 26.

Three subjects, three tracks, one event. What none of them examines is the passage between the three, which is where this week sits.

And the passage isn't a process anybody runs. Nothing converts an instruction into a variation; it becomes one when somebody notices it should have been. Nothing converts a variation into a claim; it becomes one when the valuation fails to settle. The event ages whether or not anybody is attending to it, which is what makes the first hour the only reliable point of intervention.

What each renaming drops

Follow the same change across its three names and watch what survives.

The instruction carries a reason. Somebody wanted the change for a purpose, and at the moment of asking that purpose is in the room and gets stated out loud, because asking for something requires giving a reason.

The variation carries scope and price. It describes what is being done differently and what it is worth. It has no field for why anybody wanted it, and none is required.

The claim carries consequence. It describes what followed — the delay, the disruption, the cost — and reconstructs the causation from documents.

So the reason is present at age zero, absent from the form at age one, and unrecoverable at age two. Which matters more than it sounds, because at age two the argument turns on whether the change was necessary, and the answer to that was said aloud on the first afternoon by somebody who has since left.

The cost of recording inverts

Set the cost of making the record against the value of having it, and the two run in opposite directions.

In the first hour the record costs a sentence. Everybody involved is present, nobody is defending anything, and the facts are uncontested because nothing yet depends on them. Its value at that moment, to the person holding the pen, is nil.

In the second year the same record would be worth a great deal, and making it now means interviewing people who have left, reading emails written for other purposes, and inferring what was said from what was done. The result is an argument about an inference rather than a statement of a fact.

The crossing between those two curves is the whole practical content of this week, and it explains why the discipline is hard. The moment when recording is cheapest is precisely the moment when it looks least worth doing.

The people who know are the people who leave

There is a second inversion running alongside the first, and it is structural rather than behavioural.

At age zero the people who know what happened are on site: the foreman, the engineer who agreed it, the person who asked. At age two the event is being assembled by a commercial team, possibly by a consultant, from documents.

Between those two points the site team demobilises. That isn't a failure — it is what a project does.

But it means the record is compiled by the people with least direct knowledge, and that is a certainty rather than a risk, so it can be planned for.

Planning for it means one thing only: the record has to be made by the people who know, at the time they know it, in a form that survives them.

Everything else that gets proposed — a lessons workshop before demobilisation, an interview programme, a handover pack — is a reconstruction, done late, by people already half gone. Those have their uses and none of them recovers a sentence that was never written.

Four facts, and what they aren't

What the first hour needs is smaller than people assume, and confusing it with something larger is why it doesn't get done.

It isn't a notice. Notices have contractual requirements, time limits and consequences, and Contract Week 10 takes them apart properly. It isn't a claim, not a variation request, and not an accusation.

It is four facts. What was said. Who said it. When. And what it changed — the physical difference between what would have happened and what did.

Those four fit in one sentence, require no judgement, allege nothing, and are the only things a reconstruction two years later can't recover on its own.

Everything else in a claim can be rebuilt. Programmes can be re-analysed, costs can be re-priced, drawings can be compared. What was said in a conversation on a Thursday can't be, and it is the cheapest of the lot to preserve.

System design

Row one is the only entry in this whole dictionary with nothing in the verified against column, and that is the point. A contemporaneous note is a primary record: there is nothing behind it to check it against, which is exactly why it carries weight and exactly why it can't be reconstructed later.

Record	Produced by	Required quality	Verified against	Feeds
Contemporaneous note	Whoever was there, on the day	Four facts, no judgement, no allegation	Nothing — it is the primary record	Every later version of the event
The reason given	The same note, in the words used	Recorded as said rather than as interpreted	The person who said it, while present	Whether the change was necessary
The physical difference	Site, at the time	What would have happened, against what did	The work in front of you	Causation · valuation · quantum
Date first known	You, distinguished from the date raised	Two separate dates, because they are two separate facts	The note that carries it	Time bars · whether notice was timely
Who was present	The note	Names, because they will be needed after they leave	The day it happened	Who can be asked in two years

Row four separates two dates that get written as one. When something was first known and when it was raised are different facts, they diverge by days or weeks, and the gap between them is what a time bar argument is about.

Practical insight

Write your sentence in whatever you already use, on the day, and write it particularly when it seems unnecessary.

An email to a colleague. A line in your daily report. A note in your diary. It need not go to the other party and it need not be formal — a contemporaneous record you made for your own purposes carries weight precisely because you didn't make it for an argument.

The test for whether something needs the sentence is one question: if this turns out to matter, will anybody be able to tell what happened from the documents that exist? Where your answer is no, spend the thirty seconds.

You will apply that test twenty times in a month and it will say yes to three of them, which is the right rate. A project where everything gets a note has a note nobody reads.

And expect it to feel like fuss. Writing down something that took ten minutes and upset nobody will make you feel slightly foolish, every time you do it. That feeling is doing the work of a judgement without being one, and noticing the difference is most of what it takes to write your sentence anyway.

Key takeaways

• An instruction, a variation and a claim are one event at three ages, and nothing happens between them except time.
• Each name admits different evidence and requires a different person to settle it.
• The instruction carries the reason, the variation carries scope and price, the claim carries consequence.
• So the reason is present at age zero, has no field at age one, and is unrecoverable at age two.
• Which matters because the age-two argument turns on whether the change was necessary, and that was answered aloud on the first afternoon.
• The cost of recording is lowest in the first hour and the value of the record, to the person holding the pen, is nil at that moment.
• The moment recording is cheapest is exactly the moment it looks least worth doing.
• The people who know are on site at age zero and demobilised by age two, so the record gets assembled by whoever knows least.
• That is a certainty rather than a risk, which means it can be planned for.
• The first hour needs four facts: what was said, by whom, when, and what it changed.
• It isn't a notice, not a claim and not an accusation, and confusing it with one of those is why it goes unwritten.

Records born here. The contemporaneous note — what was said, by whom, when, what changed · the reason given, in the words used · the physical difference between what would have happened and what did · the date the change was first known, as distinct from the date it was raised · who was present.

What is coming next

Phase D ends here. The site is running, work is being packaged and handed out, and things are changing daily.

What runs alongside all of it is a structure of permissions: who can approve what, up to what value, and what happens when the answer is nobody present. That structure was set up in the first weeks, hasn't been looked at since, and decides how long every decision on the project takes.