Haute Lumière

Commerce · V.11 · MMXXVI · daylight

La Bourse  /  Volume V  /  Nº V.11  /  Workbook — the student

WORKBOOK — THE STUDENT

Chapter V.11 · When Flourishing and Output Diverge

For the person studying this alone, or in a seminar, with no organisation to change yet. You have something the executive does not: a system you fully control, and a horizon long enough to see a logarithm move.

---

WHY THIS WORKBOOK IS DIFFERENT

The chapter is about two instruments disagreeing. It reads as though you need a firm with a reporting pack before any of it applies.

You do not, and the reason is the most useful thing here: you already run two measures on yourself, and you have almost certainly never put them on the same page. One is output — work submitted, hours logged, marks, applications sent, words written, money earned. The other is the condition of the stock that produces it — sleep, attention, health, friendship, curiosity, the appetite to begin. You track both. You track them in different places, at different cadences, and you consult whichever one is currently more flattering.

That is precisely the failure the chapter describes in firms. The whole apparatus — the pair, the threshold, the enquiry — works at the scale of one person, costs nothing, and is the single most transferable thing in Volume V.

A second reason this chapter is worth your time specifically: it is a worked example of how to read a disagreement between two serious literatures. You will meet a hundred of those. Almost none of them will be resolved by picking the side you already liked. This one resolves by computing a slope and asking what window each camp was looking through, and that move generalises further than any finding in the chapter.

---

PART ONE — DISCOVERY

Days 1–30: find the disagreement that is already conducted well

Exercise 1.1 — Reproduce the regression (2 hours)

Do not take the line on trust. Open lib/verify/V_11.py, read the twenty-four rows, and rebuild the regression yourself — by hand on ten of the rows, or in a spreadsheet on all twenty-four.

You are looking to confirm:

1. Slope b ≈ 0.92 ladder points per natural log unit.
2. R² ≈ 0.86.
3. That the residuals for Costa Rica (+0.60), Finland (+0.99) and Singapore (−0.92) are what the module says.

Then do the thing that matters most: change the sample. Drop the four poorest countries and run it again. Drop the four richest. Watch the slope move.

That is not a flaw in the regression; it is the chapter's entire thesis appearing in your own hands. A log relation read across a narrow range looks like nothing, and the same relation read across a wide range looks like a law. Write down what the slope becomes in each case. Bring both to your seminar.

Exercise 1.2 — The two camps, steelmanned (90 minutes)

Write two paragraphs of 150 words each. The first is the strongest possible case for Easterlin, written by someone who believes it. The second is the strongest possible case for Stevenson and Wolfers, written by someone who believes that.

Rules: no straw version, no "of course, they ignore…", and neither paragraph may use the word obviously. Each must contain one number.

If you cannot write both convincingly, you do not yet understand either. That is not a criticism; it is the assignment, and it is the most valuable ninety minutes in this workbook.

Exercise 1.3 — The appreciative interview (45 minutes, with another person)

Find someone ten or more years into working life and ask exactly this:

"Tell me about a period at work when you were producing your best output. Not the busiest — the best. What was true about the rest of your life during that period? And what did it cost, if anything?"

Stay quiet. Take notes on the conditions, not the achievements.

You are collecting primary data on the stock–flow relationship from someone who has lived it. Do this twenty times over a year and you will know something about work that no reading list contains.

---

PART TWO — THE ARITHMETIC

Days 31–45: compute before you argue

Exercise 2.1 — The price of a point (45 minutes)

Work these by hand, then check against the module.

1. Gain per doubling: b · ln 2. Confirm 0.64 ladder points at b = 0.92 and 0.24 at the controlled slope of 0.35.
2. Income multiple for one full point: exp(1 / b). Confirm 2.9× and 17×.
3. Years at 1, 2 and 3 percent real growth. Confirm 109 / 55 / 37 on the bivariate slope.
4. The window table: 10, 20, 30 and 50 years at 2 percent. Confirm 0.18 / 0.37 / 0.55 / 0.92 ladder points.

Then write one sentence explaining how 55 years and 0.18 points over a decade are the same fact. If the sentence takes more than thirty words, you have not got it yet.

Exercise 2.2 — The resolution problem (45 minutes)

Compute the survey arithmetic yourself:

  predicted US gain, controlled slope    0.29 ladder points
  on a 1-3 scale                          0.29 × (2/10) = 0.058
  standard error, n = 1,500, sd = 0.65    0.65 / √1,500 = 0.0168
  t                                       0.058 / 0.0168 = 3.5

Now answer, in writing: if t = 3.5, why is the question not settled?

The answer is the most important methodological idea in the chapter and it will serve you in every empirical field you ever touch. A standard error contains sampling error and nothing else. Fifty years of changes in question wording, survey mode, response rate and sample composition are not in it, and any one of them can move a mean by the 2.9 percent of scale that is in dispute.

Then design the study that would settle it. One page. What would you fix, what would you anchor, what would you measure behaviourally? This is a real research proposal and it is within the reach of a student who has done Exercise 2.1.

Exercise 2.3 — Your own honest negative (30 minutes)

The chapter's honest negative is the set of permanent divergences: shift work, unpaid care, commuting. Each names who bears the cost.

Now do it for yourself. Name one thing you rely on that costs somebody something permanent, that no design you can imagine removes. Write who bears it. Use a name if you can, not a category.

Then — and this is the harder half — write what you would be willing to pay, in money, time or convenience, to change the distribution. Do not write "more awareness." Write a number or an hour.

---

PART THREE — DREAM AND DESIGN

Days 46–70: build the pair

Exercise 3.1 — Your two measures (90 minutes)

Choose one flow measure and one stock measure for your own life. They must be capable of disagreeing with each other; if they are mechanically linked, pick again.

Both go on one page, at the same cadence, in the same place. That layout decision is 80 percent of the mechanism and it is the part people skip.

Define each precisely enough that another person would compute it identically. "Reading" is not a measure. "Pages of primary source read per week, logged the day of" is.

Exercise 3.2 — Your threshold, written cold (30 minutes)

Before you have any data, write the rule that declares a divergence.

1. Collect eight weeks of both series without changing anything.
2. Compute the standard deviation of each.
3. Declare a divergence when the two move in opposite directions and both moves clear one standard deviation, two weeks running.

Write it down, date it, and show it to one other person. A threshold agreed after a number has moved is a negotiation about that number.

The arithmetic for why the gate matters is Brief 9, and you should be able to produce it on request: sign alone fires 25 percent of the time on two series doing nothing; gated, 0.25 percent — about 99× better.

Exercise 3.3 — The enquiry, rehearsed (45 minutes)

When your threshold fires, you get one question and three weeks: what is the flow being drawn from?

Rehearse it now on a past divergence you already remember — a term when the marks were good and everything else was not, or the reverse. Write 400 words answering that one question about that past period, honestly, and end with the sentence you would have needed to read at the time.

Exercise 3.4 — The present-tense description (60 minutes)

Write 500 words describing your working life ten years from now, in the present tense, as description rather than aspiration.

Constraints, and they are the exercise:

• Describe an ordinary Wednesday, hour by hour.
• Name both of your measures and say what they read.
• Include one permanent divergence you have accepted, and say who bears it — including where the answer is you.
• No adjectives of achievement.

The third constraint defeats almost everyone, and that is the finding.

---

PART FOUR — DESTINY AND DELIGHT

Days 71–90: make it hold, and enjoy it

Exercise 4.1 — The second reader (this week)

One person who sees both of your series and will say something when they diverge. Not a supporter — a reader. Ask explicitly, and give them the credit the first time they catch something.

The chapter's version of this is an enquiry owner who owns neither number. Yours is a friend who is not invested in either one being good.

Exercise 4.2 — Let it fire on you once (ongoing)

The first time your threshold fires and you do not move it is the moment the whole apparatus becomes real. Expect to want to adjust it. Do not.

Write, on the day: the threshold fired, and here is what I did not change.

Exercise 4.3 — Delight, on purpose (ongoing)

Delight is the adoption mechanism, not the reward. Make the weekly page genuinely pleasant to fill in — the specific notebook, the particular hour, the café, the pen.

Then complete this sentence: the part of this I look forward to is ___. If you cannot complete it, redesign the practice until you can. A practice with no pleasure in it has a half-life.

---

THE TERM PROJECT

One disagreement, carried the whole way

Take a live empirical disagreement in any field — not this one — where two serious camps read the same data differently. Then do to it what the chapter did to Easterlin and Stevenson–Wolfers.

Deliverables.

1. The steelman pair. 300 words on each side, each written by someone who believes it, each containing a number.
2. The computation. Find the quantity both camps are implicitly claiming, and compute it yourself from published data. Show every input and every intermediate term, as lib/verify/V_11.py does. This is the heart of the project.
3. The window. State what range, horizon or sample each camp is looking across, and compute what each camp's own model predicts inside the other's window. If they converge, say so. If they do not converge, that is a better result and you should say that too.
4. The honest negative. One thing your analysis cannot settle, and what evidence would settle it.
5. The design. One page: the study that would decide it, with its cost roughly estimated.
6. The reflection. 600 words. Which side you expected to win, which side you now think is right, and what the gap tells you about how you were reading.

How it is assessed. Not on whether you resolved it. On whether the arithmetic is reproducible and whether you reported a result that did not flatter the side you started on.

An analysis that failed to resolve the dispute and printed all its inputs is a first-class piece of work. One that resolved it and cannot be checked is not.

---

SELF-ASSESSMENT

The two that matter most are the third and the seventh. The rest can be learned in a fortnight. Those two are habits of honesty, and habits take a term.

---

CARRYING IT FORWARD

In ninety days, done properly, you have:

• A regression you built yourself, and the experience of watching its slope move when you changed the sample — which is an inoculation against every over-claimed correlation you will meet.
• Eight weeks of a paired series on your own life, with a threshold written cold.
• Twenty appreciative interviews on what actually produced people's best work.
• The ability to say, in a room, that is not a null result, it is a resolution limit — and then show the arithmetic.

That last one is worth more than it sounds. Most rooms cannot distinguish "we looked and found nothing" from "we could not have seen it if it were there." The person who can, changes what the room is able to decide, and you can be that person before anyone has given you any authority at all.

---

APPRECIATIVE QUESTIONS FOR YOUR SEMINAR

1. When has one of us changed our mind because of a computation rather than an argument, and what made that possible?
2. Which disagreement in our field would dissolve if both sides computed the same quantity — and what would it take to get them to?
3. What is already working about how this group disagrees well, and how would we get more of it?
4. If every one of us kept a paired series for a term, what would we know at the end of it that no reading could give us?