A package is an intersection, not a slice.

The second dimension

The cost engineering framework describes it as a matrix. On one axis the breakdown of scope; on the other the breakdown of the organisation — who exists, and what each of them is capable of taking on. A work package is the intersection of a cell from each.

Read that as a test rather than a diagram and it becomes useful. A branch of the breakdown structure lacking an owner remains scope. An owner holding unbounded scope remains a resource. Only where the two cross does something exist that can be planned, handed over and managed.

Which explains a failure that otherwise looks like bad luck. Decompose scope on its own, get to pieces that look the right size, and then assign them afterwards — and the assignments won't line up, because the pieces were made without reference to who could take them. A parcel spanning two trades leaves the ownership question open. A parcel amounting to one trade's morning costs more to administer than to do.

Schedule Week 10 builds a breakdown structure and week 12 handles the identifiers that join it to everything else. What is being added here is that either operation alone leaves you one axis short of something anybody can be given.

A boundary somebody can walk

The second requirement is that the boundary be findable in the world, not only in a document.

A package bounded by a grid line, a level, a room, a chainage or a system can be walked. Somebody can stand at the edge of it. A package bounded by a phrase — the mechanical works to the east block — resists that, because two people reading the phrase draw the line in two places, and each believes there is agreement until both have priced it.

The consequence lands on measurement. A boundary that can be walked can be measured, and a package that can be measured produces a number. A boundary that exists only in a sentence produces a percentage, and a percentage against a fuzzy boundary is an opinion with a decimal point.

Scope has no geometry

Here is the failure that packaging alone can't solve, and it is worth knowing that it can't.

Two packages can be perfectly bounded, perfectly owned, correctly sized, and scheduled in the same week — and be in the same room. Nothing in a breakdown structure prevents it, because a breakdown structure has no space in it. It has parents and children and no walls.

So the packaging exercise produces a set of parcels that are logically disjoint and physically overlapping, and the overlap is discovered by two foremen at the same door. This is a property of dividing work in one representation and performing it in another, rather than a planning failure in the sense of somebody having been careless.

What that argues for is a second pass over the packages with a drawing rather than a structure — not to repackage, but to find the pairs sharing a space and note them as a relationship the structure has no field for.

It is a cheap pass and an unusual one. The packaging exercise is finished by then, everybody has moved on to sequencing, and going back to look at the same set through a different representation feels like rework. It takes an afternoon and it finds things no amount of re-reading the structure will.

How big is a package

Size is decided by two opposing failures, and the space between them is narrower than it looks.

Too large to report: a package spanning four months and three areas stays partly done throughout, and every month somebody estimates how partly. Too small to administer: a package taking a crew a day still needs an owner, a start, a completion, an inspection and a record, and that overhead stays fixed as the work shrinks.

The same trade-off appeared in week 13 on the level of detail in a programme, which is no coincidence — it is the same decision seen from the scope side rather than the time side. And it has the same property: easy on day one, and by month twelve there is history attached to the packages and the size is settled.

Which suggests deciding the convention deliberately and writing it down, rather than letting it emerge package by package. A stated rule — one floor, one trade, one month, or whatever fits the work — produces a set that can be compared with itself. A set that emerged produces packages of six different sizes and a reporting problem that looks like a data problem.

What this week isn't about

A package existing on paper remains some distance from a package a crew can start. The gap between those two is the whole subject of Reporting Week 9, which takes apart what the word ready means when six functions each mean something different by it.

Nothing of that is repeated here, and the two should not be confused. Packaging is a design question: what are the parcels, who owns them, where do they stop. Readiness is an operational one: is this parcel startable on Monday. A project can do the first well and the second badly, and the symptom looks identical from a distance.

System design

Row four asks for something a matrix produces free and nobody keeps: the cells that stayed empty. Scope with no owner and owners with no bounded scope are both findable on the day the matrix is drawn and invisible afterwards, once the packages have been listed and the matrix thrown away.

Record	Produced by	Required quality	Verified against	Feeds
Package list	The project, from the matrix	One named owner per package, never a company or a discipline	Somebody accepting it	Who is asked when it stops
Physical boundary	The same exercise	Expressed in something walkable — grid, level, room, chainage	Standing at the edge of it	Measurement · payment · handover
Measurement rule	You, per package	What number, from what source, counted by whom	One month of actual counting	Whether progress is a figure or a view
Scope-to-organisation matrix	Planning, once	Shows the cells that produced no package, as well as those that did	The organisation that exists	What has been left unassigned
Spatial overlap list	You, with a drawing	Names the pairs sharing a space, which no structure can hold	The layout from week 20	Sequencing between owners

Row five holds the thing the whole apparatus can't: a relationship between two packages that exists in space rather than in logic. It has to be written down separately, because there is nowhere in a breakdown structure to put it.

Practical insight

Take one package from your programme and ask three questions of it. You need ten minutes and no permission from anybody.

Who owns it — one name, not a company and not a discipline. Where does it stop — can you walk to your boundary and stand on it. How will it be measured — what number, from what source, counted by whom.

If any of the three has no answer, what you are holding is a scope element rather than a work package, and the difference will surface as an argument in about four months.

Do it on a package that is already running rather than one still in the future. A package in progress has real answers or visibly lacks them, and the exercise takes ten minutes instead of an afternoon of speculation.

Then do the second pass. Print your packages for one month onto a drawing, coloured by owner, and look for the parcels sharing a space. You will find pairs, they will surprise you, and every pair you find now is a conversation between two of your foremen that won't happen at a door.

Key takeaways

• Nobody can be handed a floor, a system, or a level of a breakdown structure.
• A work package is the intersection of a scope element and an organisational one, not a slice of either.
• A branch with no owner isn't a package; an owner with no bounded scope isn't a package either.
• Decomposing scope alone and assigning afterwards produces parcels that span two trades or amount to one trade’s morning.
• A boundary has to be findable in the world — a grid line, a level, a room, a chainage — rather than described in a phrase.
• Two people reading a phrase draw the line in two places and neither knows they disagree until both have priced it.
• A boundary that can be walked can be measured; one that exists in a sentence produces an opinion with a decimal point.
• A breakdown structure has parents and children and no walls, so packages can be logically disjoint and physically overlapping.
• That overlap is discovered by two foremen at the same door, and packaging alone can't prevent it.
• Size sits between two failures: too large to report, too small to administer, and the overhead of administering doesn't shrink with the work.
• Packaging is a design question and readiness is an operational one, and a project can do the first well and the second badly.

Records born here. The package list, each with a named owner · the physical boundary of each, expressed in something walkable · the measurement rule per package · the scope-to-organisation matrix the packages came out of · the spatial overlaps the structure can't hold · the size convention and the reason it was chosen.

What is coming next

With packages defined and owners named, the project acquires a rhythm, and the rhythm is a week.

Somebody decides on a Thursday what a crew will do on Monday, and by Wednesday half of it has changed. What separates a project where that works from one where it doesn't isn't the quality of the plan but what happens to the things that stop it — who receives them, on what day, and whether anybody ever hears the answer.