Haute Lumière

Commerce · IV.07 · MMXXVI · daylight

La Bourse  /  Volume IV  /  Nº IV.07  /  Workbook — the student

WORKBOOK — THE STUDENT

Chapter IV.07 · Supply Chains as Ecologies

Volume IV — Production and Regeneration

A term of practice, a term project, and a self-assessment. Everything here can be done with a spreadsheet, a notebook and a willingness to ask people where things come from.

---

WHY THIS WORKBOOK IS DIFFERENT

Most supply-chain teaching asks you to optimise a chain somebody else drew. This one asks you to find out whether the chain exists.

The claim of the chapter is not that supply chains are complicated. It is that the object is a population and is routinely managed as a line, and that the gap between those two descriptions can be measured in dollars, in tiers and in days. You are going to measure it — first on something small enough that you can actually finish, then on something that matters to you.

You will finish the term able to do four things almost nobody can do on demand: state how many firms are upstream of a product and what it would cost to see them, compute which cheap component gates the most expensive output, compute how many sources a given risk justifies, and compute how deep a standard can be enforced before enforcing it costs more than it is worth. Those are four divisions. They are not hard. Almost nobody does them.

---

PART ONE — DISCOVERY

Weeks 1–4: find the population

Exercise 1.1 — Pick one object. Choose a physical thing you own that you can hold: a pair of shoes, a coffee, a bicycle inner tube, a phone charger, a jar of honey. Not a service, not software. The constraint matters: physical things have suppliers who can be named.

Write down, without research, everything you believe about where it came from. Date the page. You will come back to it in week twelve and the difference is half of what you learn this term.

Exercise 1.2 — Find tier one. Identify the company whose name is on the object. Find their public statements about manufacturing: annual report, sustainability report, supplier list if they publish one. Some do — Apple, Patagonia, Nike, H&M, and an increasing number of others publish factory lists. Record what you find and, crucially, record what you looked for and did not find. That second list is the finding.

Exercise 1.3 — Try tier two. Pick one named tier-one factory and try to find out who supplies it. You will usually fail. Note precisely where the trail went cold: the company does not publish, the factory is a contract manufacturer serving many brands, the material is a commodity with no traceable origin.

Compute your own disclosure rate: of the tier-one nodes you identified, what fraction named at least one of their own suppliers? The chapter assumes sixty percent for a firm asking as a customer. You are asking as a member of the public, which is a different and much weaker position, and your number will show it. Write your number down with the denominator beside it.

Exercise 1.4 — The population estimate. With whatever b your research supports — and if it supports nothing, use 5 and say so — compute:

  N₁ = ____    N₂ = N₁ × b = ____    N₃ = ____    N₄ = ____

Exercise 1.5 — Find the recruitability. Ask one question of one person who makes something: if your main supplier vanished tomorrow, who would you call? Bakers, mechanics, printers, farmers, jewellers, brewers all answer this question readily and the answers are consistently more interesting than any corporate disclosure. Record the answer verbatim. You have just measured something no database holds.

---

PART TWO — THE ARITHMETIC

Weeks 5–8: compute before you argue

Exercise 2.1 — The visibility bill. Using $1,200 a node and your own disclosure rate d:

  cost at tier n       =  N₁ · b^(n-1) · 1,200
  fraction visible     =  d^(n-1)
  cost per node FOUND  =  1,200 / d^(n-1)

Build the four-row table. Then answer in one sentence: at what tier does this stop being worth doing for this object, and why?

Exercise 2.2 — Exposure ratios in your own life. This is where the chapter becomes personal and it is the exercise students remember.

        E  =  value of what it gates  /  the thing's own cost

Worked example, and you will find worse ones:

  a $60 charger gating a $1,400 laptop                    23.3 x
  a $12 cable gating the same laptop                     116.7 x
  one password gating every account you have               — undefined, and
                                                             that is the point

Build your own table of five. Rank them by E, not by price. Then look at where your money and attention actually go, and note the mismatch. This is the same mismatch a procurement department has, at a different scale, and you now have direct experience of why it persists: the expensive thing feels important.

Exercise 2.3 — Your own common-mode term. List everything you depend on to finish this term's work. Now find the q: the single failure that takes out several at once. One flat. One laptop. One bank account. One cloud account that holds your notes, your backups and your email recovery.

Model it. Take p = 0.03 as the chance any one copy of your work is lost in a term and q = 0.01 as the chance all copies go together:

  P(all m fail)  =  q + (1 - q)·p^m

  m = 1   0.039700
  m = 2   0.010891
  m = 3   0.010027

The second copy buys you 2.88 percentage points. The third buys 0.086 — about thirty times less. And nothing you do gets below q = 0.01 until you break the shared dependency, which usually means one copy somewhere that does not share your account, your building or your device.

Exercise 2.4 — The breakeven. If losing a term's work costs you $4,000 in time and repeated effort, and each backup costs $30 a year:

  copy 2   ΔP 0.028809   saves $115.24/yr against $30   — take it
  copy 3   ΔP 0.000864   saves   $3.46/yr against $30   — do not

Notice what just happened: the arithmetic told you to stop. Most risk advice never does, which is how people end up with five backups of their essays and one copy of the only thing that matters.

---

PART THREE — DREAM AND DESIGN

Weeks 9–11: build something that would hold

Exercise 3.1 — The enforcement depth of a rule you care about. Pick a standard you would want applied — no forced labour, no deforestation, living wage, recycled content. Pick a company. Estimate N₁ from their published supplier list, take b = 5, take c = $6,000 an audit, and estimate V₁ as whatever you think full compliance is worth to them annually in avoided disruption, remediation and lost contracts. Compute:

  n*  =  1 + ln(V₁/C₁) / ln(b/λ)

Try λ = 0.6, 0.3 and 0.15. Write the three answers down. Now go and read what the company's own code of conduct claims about its reach. The gap between that sentence and your three numbers is the most useful thing you will produce this term, and it is not a gotcha: the company is not lying, it is describing an intention with no arithmetic attached. Your job as an analyst is to supply the arithmetic.

Exercise 3.2 — Find the waist. For your chosen product, find where the population narrows. Refining, smelting, certification, port, a single machine type, a single seed variety, a single accreditation body. Count the nodes at the narrowest point. Compute the audit cost at the mouth and at the waist and take the ratio. In the chapter's mineral case it is 26.7× and a 96.25 percent saving.

Exercise 3.3 — Design one intervention. One page. Name the standard, name the tier you would enforce it at, name where you would stand in the network, and give the cost with its denominator — what your proposal does not cover. A proposal that claims full coverage is a proposal nobody experienced will read.

---

PART FOUR — DESTINY AND DELIGHT

Week 12: make it hold, and enjoy it

Exercise 4.1 — Date everything. Go back through your map and put a date on every node. Anything older than your first week is already a claim rather than a fact. This habit — a node not re-verified reads as unknown, not as unchanged — is the single most transferable thing in this workbook and it applies to every dataset you will ever inherit.

Exercise 4.2 — Open your week-one page. Read what you believed before you started. Write half a page on what changed and, more importantly, on which of your original beliefs turned out to be right for the wrong reason. Those are the dangerous ones.

Exercise 4.3 — Tell one person. Find someone who owns the same object and tell them one thing you found out about it, in under a minute, with a number in it. If they ask a follow-up question, the number was the right one.

On delight. The pleasure in this work is specific and it takes about six weeks to arrive. It is the moment an ordinary object stops being opaque — when you look at a coffee cup and see a population with a shape, a waist, a set of exposure ratios and about four people who could tell you the rest if you asked them nicely. The world does not get heavier when you can see through it. It gets considerably more interesting, and you stop being frightened of complexity you can count.

---

THE TERM PROJECT

One object, carried all the way

The deliverable. Eight to twelve pages on one physical object.

1. The map. As deep as you got, with every node dated and the disclosure rate stated with its denominator.
2. The population estimate. N₁ through N₄, with your b justified.
3. The visibility bill. The four-row table and the tier at which you would stop.
4. The exposure ratios. At least five components ranked by E, with the mismatch against price named.
5. The common-mode analysis. What the real q is, and what would reduce it.
6. The enforcement depth. n* at three values of λ, against the company's own claim.
7. The waist. Where it is, how many nodes, and what enforcement there would cost.
8. One page of design. Your intervention, with its denominator.

How it is marked. Not on how deep you got. Depth is mostly a function of which company you chose. It is marked on whether every number has its inputs printed beside it, whether every coverage claim states what it excludes, and whether the conclusion follows from the arithmetic rather than from the position you held in week one.

The one rule. If a figure in your project does not appear in your own calculation sheet, delete it. A number you copied is a number you cannot defend, and the first one a reader checks decides whether they check the second.

---

SELF-ASSESSMENT

Score each honestly, 0 to 3. Nothing here is graded by anyone else.

Twelve or below after a full term is a signal about practice, not about capacity. The fix is a smaller object, done completely, rather than a larger one done partly.

---

CARRYING IT FORWARD

Three things worth keeping after the term ends.

The habit of asking where. It costs one question, it is almost always welcomed, and it consistently returns information no database holds.

The habit of the denominator. Every claim about coverage names what it did not look at. You will find this makes you more persuasive rather than less, because the person you are talking to already suspected the gap and you just became the one who could be trusted about it.

The habit of computing the stopping point. Most analysis tells you to do more. Arithmetic that can tell you to stop is rarer and worth considerably more, and you now have three examples of it — the third backup, the third source, and the fourth tier.

---

ONE MORE HABIT, AND IT IS THE ONE THAT TRAVELS

Everything above is about objects, and objects are only the practice ground. The transferable skill is narrower and stranger than it looks: you have learned to ask what a claim did not count.

Almost every confident statement you will meet in a career — a coverage figure, a market share, a compliance claim, a survey result, a model's accuracy — is a numerator with an unstated denominator. Thirty-six percent of tier three is an honest sentence because the denominator is printed beside it. We require compliance throughout our supply chain is not a sentence with a truth value at all, because nothing in it can be counted.

So carry three questions out of this term and use them on anything, in any subject:

1. What is the denominator? What did this number not look at?
2. What is the shared term? Which of these supposedly independent things would fail together, and for what single reason?
3. Where does it stop paying? Every good analysis can tell you when to stop. An analysis that only ever recommends more of itself is advertising.

You have now done all three with real numbers on a real object, which means they are yours rather than something you read. That is the whole point of the term.

---

APPRECIATIVE QUESTIONS FOR YOUR SEMINAR

1. When has one of you found out something surprising simply by asking a maker where their materials came from? What made them willing to answer?
2. Which object in this room has the highest exposure ratio, and how would we settle the argument with a number?
3. Where in our own studies do we have a q — one shared dependency behind several things we think are independent — and what would break it?
4. If our university published its supplier list tomorrow, what would we want to look at first, and what would we build with it?
5. What would it take for the maps we have built this term to still be useful to next year's students rather than starting again?