Haute Lumière

Commerce · VI.01 · MMXXVI · daylight

La Bourse  /  Volume VI  /  Nº VI.01  /  Workbook — the student

Four colleagues walking toward a long table set with dried grasses, the room bright with sun.
Plate VI.01 · Workbook — the studentThe Table Before the Meeting.A decision rule is furniture. It decides who can be in the room, how long they will be there, and who is not in the room at all — and it does this before anybody has said a word.

WORKBOOK — THE STUDENT

Chapter VI.01 · Who Decides

For the personal student. A term of practice on one question: what does it cost to decide a thing, and who pays when it is decided? Everything here is applied to a life you are actually living — a household, a flat share, a student society, a band, a research group, a family.

WHY THIS WORKBOOK IS DIFFERENT

Most governance teaching gives you a taxonomy: here is majority rule, here is consensus, here is a board. A taxonomy is a thing you can be examined on and cannot use.

This workbook gives you an instrument instead. By the end of a term you will be able to take any group you belong to, count its members, estimate two parameters from your own observation, and compute the threshold that group should be using for a given class of decision — and then say, out loud and with a figure, who is paying for the rule it currently uses.

That is a small skill and it is unreasonably powerful, because almost nobody has it. Committees you will sit on for the rest of your life set their rules by instinct and defend them by principle. You will be the person in the room who can say this rule costs us about this much, and it costs them about that much, and here is the one that costs less than both.

PART ONE — DISCOVERY

Days 1–30: find the rules you are already living under

Exercise 1.1 — The rule census (2 hours)

Pick four groups you belong to. Make them different in size: something with three or four people, something with nine or ten, something with fifty, something with hundreds. For each, write down in one page:

  1. Who is formally allowed to decide, and what is the written rule?
  2. Who actually decides, and how do you know?
  3. Name three decisions the group made in the last six months and, for each, how long it took from first raising to settled.
  4. For each of those three, answer the hardest question in this chapter: what would it have cost to get it wrong? Use money where money applies and hours where it does not. Guess if you must, but write a number.

That fourth column is the exercise. It is the loss term in the external-cost curve and it is the only input most groups never estimate.

Exercise 1.2 — Find the blocking coalition (1 hour)

For each of the four groups, compute n − k + 1 — the smallest number of people who can stop something. Then name them. Not their role: their names. Under a 6-of-9 rule it takes four people to block. Under a unanimity rule it takes one.

Write one sentence for each group: "In this group, the following people, acting together, can stop anything." If that sentence surprises you, you have just learned what your group's rule actually is, as opposed to what it says.

Exercise 1.3 — Appreciative interviews (3 hours)

Ask three people in one of these groups the same three questions, and write down what they say rather than what you expected:

You are looking for the positive core: this group already decides some things well, and the conditions that make that possible are in the answers.

PART TWO — THE ARITHMETIC

Days 31–50: learn to compute before you argue

Exercise 2.1 — Estimate p by observation (1 week)

p is the probability that a given member accepts a given proposal. You cannot look it up; you estimate it. For one group, over one week, record every proposal made and how many members supported it on first hearing. Divide total supporters by total (proposals × members). That is your p.

Most student groups land somewhere between a half and three quarters. The chapter uses 0.60. If your figure is far from that, your subsequent numbers will be different from the chapter's and that is correct — the shape of the curve is the claim, and the level belongs to your parameters.

Exercise 2.2 — Build the table by hand (2 hours)

For a group of nine at p = 0.60, with one round costing £3,240 and twelve contested decisions a year, and a loss of £30,000 to a member left outside the winning coalition, reproduce this table yourself. Use the binomial tail; a spreadsheet is fine, and so is lib/verify/VI_01.py.

   k    P(X>=k)   rounds    decision £/yr   external £/yr     total £/yr
   4     0.9006     1.11           43,169         200,000        243,169
   5     0.7334     1.36           53,011         160,000        213,011
   6     0.4826     2.07           80,562         120,000        200,562
   7     0.2318     4.31          167,740          80,000        247,740
   8     0.0705    14.18          551,146          40,000        591,146
   9     0.0101    99.23        3,858,025               0      3,858,025

Do not move on until your numbers match. The point is not the arithmetic; it is that you have now personally produced 99.23 rounds for unanimity and will never again hear someone propose it without seeing that figure.

Exercise 2.3 — Move one parameter (1 hour)

Now change only the loss and watch the optimum move: £5,000 → 4 of 9 (44.4 per cent), £10,000 → 5 of 9 (55.6 per cent), £30,000 → 6 of 9 (66.7 per cent), £100,000 → 7 of 9 (77.8 per cent), £300,000 → 8 of 9 (88.9 per cent).

Write two sentences on what this does to the idea that a group should have a decision rule.

Exercise 2.4 — The consent computation (1 hour)

Compute 1 / 0.9⁹. You should get 2.58 rounds against unanimity's 99.23 — faster by 38.4 times. Then find the threshold rule with the same expected rounds: 6.23 of 9, or 69.2 per cent.

Write one paragraph answering: if consent behaves like a two-thirds rule, who in my group is paying the difference, and can they say so?

Exercise 2.5 — The delegation break-even (1 hour)

Most groups you belong to delegate without ever saying so: a committee, a treasurer, a captain, one person who "just handles it." Price it.

Take an assembly of 1,200 deciding directly at a 60 per cent threshold. It expects 1.95 rounds at £432,000 a round, so twelve decisions a year cost £10,114,446. Delegate the same twelve to a board of nine at 6-of-9 and they cost £80,562. The saving is £10,033,884. At a divergence rate of 25 per cent, the loss per divergent decision at which delegating stops paying is £3,344,628.

Now do it for your own group, with your own numbers, and write the one sentence that falls out: delegate everything whose loss-if-wrong is below X, and reserve everything above it. That sentence is a constitution, and you have just derived it from four inputs.

Exercise 2.6 — The status quo has a price too (1 hour)

Buchanan and Tullock measure every cost from the status quo, so in their diagram the status quo is free and the external-cost curve reaches zero at unanimity. Test that. Add an annual harm the unmade decision leaves standing, borne by whoever the current arrangement is already costing, plus the delay each rule creates. At zero the optimum is 6 of 9 (66.7 per cent); at £300,000 a year it falls to 5 of 9 (55.6 per cent).

Write one paragraph on a decision your group has not made, and on who is paying for it not being made. That paragraph is usually the most useful page in the whole term.

PART THREE — DREAM AND DESIGN

Days 51–70: build something a group will actually use

Exercise 3.1 — The priced schedule (4 hours)

Take the group where you have the most standing. Produce one page with four columns: class of decision · loss-if-wrong · computed threshold · who decides. Three or four bands is enough. Compute each threshold rather than choosing it.

Exercise 3.2 — The objection grammar (1 hour)

If your schedule uses consent anywhere, write the sentence that defines an admissible objection. One sentence. It sets q, which sets both curves, so it is the most load-bearing sentence in the document.

Then build the objection log: date · proposal · objection raised · was it about workability or preference · how resolved. Twenty minutes a month, and it is the only way you will ever measure q rather than assume it.

Exercise 3.3 — Present it and survive the room (2 hours)

Take the schedule to the group. Expect two objections and prepare for both.

"This is bureaucracy." The answer is the arithmetic: the current rule costs the group a computed number of hours a year, and the schedule lowers it. Bring the figure.

"You are trying to reduce our say." The answer is that the schedule raises the threshold on everything above the loss band and lowers it only below — and that it was written before anyone knew which question would come up, which is what makes it fair. A rule set before the question is a rule you can lose to without injury.

PART FOUR — DESTINY AND DELIGHT

Days 71–90: make it hold, and notice what changed

Exercise 4.1 — The amendment threshold (30 minutes)

Set the threshold for amending the schedule higher than any threshold inside it, and set it now, before anybody wants to amend it. This is the single move that determines whether the work survives its first contested question.

Exercise 4.2 — The drift check (1 hour, then monthly)

Read the objection log and count what share of objections were about workability rather than preference. If that share is falling, q is widening, your consent rule is sliding toward unanimity, and nobody has amended a word. That is the most common quiet death of a governance system and the log is the only instrument that sees it.

Exercise 4.3 — Notice the delight (30 minutes)

At the end of the term, write half a page on the first meeting that ended early because the threshold was met and everyone could see it. Describe what the rest of that afternoon was like. Attention is the currency this whole chapter is about, and you have just got some back.

Exercise 4.4 — Read one real institution against the model (2 hours)

Pick one of the three the chapter cites and write a page on it in the model's own terms — threshold, blocking coalition, who pays.

Article V. Thirty-eight of fifty states ratify, so 13 can block, and the thirteen least populous hold 14,618,613 people, 4.41 per cent of the country. Roughly 11,848 amendments have been proposed and 27 ratified — 0.228 per cent. Is that rule working, or failing? Defend your answer with the external-cost curve rather than with an opinion about politics.

Cloture. The 1917 rule took 67 of 100 and let 34 block; the 1975 rule takes 60 and lets 41 block — a blocking coalition 20.6 per cent larger, drawn from at least 21 states holding 37,174,921 people or 11.22 per cent of the population. Explain why use of the block rose anyway.

Buurtzorg. Teams of at most 12, 10,000 nurses, a central staff of about 50 — a support ratio of 200 to 1 across roughly 833 teams, and KPMG found 40 per cent fewer care hours per client. Show why twelve is the number: at q = 0.10 a team of twelve expects 3.54 rounds, and a body of fifty does not converge at all.

THE TERM PROJECT

One group, carried the whole way

Choose one group in week one and stay with it. Produce, by the end of term, a single document of eight to twelve pages:

  1. The rule census. What the group's written rule is and what its real one is.
  2. The decision register. Every decision in the term, with time-to-settle and loss-if-wrong.
  3. Your measured p and q, with the observation method stated and its weaknesses named.
  4. The computed schedule, with the arithmetic shown for at least two bands.
  5. The blocking coalition, named, for the current rule and for yours.
  6. Who pays. One page identifying the group that bears the external cost of your proposed schedule, and what it costs them. This section is the project. A schedule that cannot name its payer has hidden them.
  7. What happened when you took it to the group.

Assessment weights section six most heavily. A student who computes a beautiful optimum and cannot say who pays for it has learned the model and missed the chapter.

SELF-ASSESSMENT

Score yourself honestly, out of four, on each.

Twenty-four and above means you can use this. Below sixteen, the missing piece is almost always the loss-if-wrong column, and the fix is one hour with the minutes.

CARRYING IT FORWARD

The instrument scales without modification. Volume VI goes on to the commons, to polycentric governance, and to Ostrom's design principles — which are, read correctly, a specification of the conditions under which a decision rule holds and the conditions under which it collapses. You are now equipped to read them as engineering rather than as ethics.

And one habit worth keeping past the term: whenever anyone proposes a rule, ask two questions in this order. What is the loss if we get this class of decision wrong? And who can block, and what are their names? Those two questions settle most governance arguments before they start, and almost nobody asks them.

APPRECIATIVE QUESTIONS FOR YOUR SEMINAR

  1. When has a group you belong to decided something quickly and well? What made that possible?
  2. Which of our groups has the healthiest relationship with disagreement, and what are they doing that the others are not?
  3. If every proposal in this seminar arrived with its own threshold printed on it, what would our discussions be about instead?
  4. What would it take for the people who lose a vote here to still feel dealt with fairly a month later?
  5. Who in our shared life pays for a rule the rest of us benefit from, and how would we find out what it costs them?
  6. What is the smallest change to how we decide that would give us back the most attention?