Haute Lumière

Commerce · VI.10 · MMXXVI · daylight

La Bourse  /  Volume VI  /  Nº VI.10  /  Workbook — the student

WORKBOOK — THE STUDENT

Chapter VI.10 · Auditing a Governance System

For the person studying this alone, or in a seminar, with no board to audit yet. You do not need one. You already belong to three or four governed bodies, all of which keep records, and one term is enough to run a real audit on one of them.

---

WHY THIS WORKBOOK IS DIFFERENT

The chapter was written for someone with access to five years of board minutes. You may have access to a student society's shared drive, a course committee's agendas, a sports club's WhatsApp group and a flat's kitchen rota.

That is enough, and in one respect it is better. The bodies you belong to are small enough that you can read the whole archive, which almost nobody auditing a corporation can do. You will spend this term learning an instrument on a system where you can check your own answer by knowing the people in it — and that is the only time in your life you will be able to do that.

One warning, and it is the ethical spine of the whole term: an audit of a body you belong to is an audit of your friends. Everything below is designed so that the finding is about the structure and never about a person, and so that the people in the record can read your report without flinching. If you cannot write it that way, you have not finished thinking about it.

---

PART ONE — DISCOVERY

Weeks 1–3: find a body and read its whole record

Exercise 1.1 — The five candidates (60 minutes)

List five governed bodies you belong to. A governed body is anything that (a) has more than three people, (b) makes decisions that bind the others, and (c) leaves a trace. Candidates people usually miss:

For each, write one line: what it decides, who is bound, and where the trace lives. Then choose the one with the longest readable record, not the one with the most interesting politics. The record is the instrument.

Exercise 1.2 — The appreciative read (2 hours)

Before you count anything, read the whole archive once, straight through, for one purpose only: find the three best decisions this body has ever made. Write a paragraph on each. Who first raised it, what happened in between, and what made it possible.

Do this first and do it properly. It is not a warm-up. An audit that begins by looking for faults finds faults, reports them, and is never invited back; an audit that begins by establishing what the body is good at produces findings people can act on, because they have somewhere to stand while they read them. You will also discover that the three best decisions share a mechanism, and that mechanism is usually your most valuable finding of the term.

Exercise 1.3 — Build the decision register (3 hours)

One row per decision, six columns:

  date of first trace | date decided | route | mover | outcome | downstream proof

First trace is the earliest external evidence the demand existed — a message, an email, an agenda item, a complaint. Route is standing process or exception. Downstream proof is anything showing the decision actually happened.

Most bodies your size will give you between forty and a hundred and fifty rows. Extraction takes an afternoon. This register is the term's real deliverable and everything else is computed from it.

---

PART TWO — THE ARITHMETIC

Weeks 4–7: compute the six

Exercise 2.1 — Weighted participation (90 minutes)

Identify the tiers. Most small bodies have two or three: a committee that decides, sometimes a wider group that proposes, and a membership that attends. Assign each tier a share of the decision weight, then compute

      N_eff  =  1 / Σ sᵢ²

Now do the thing the chapter insists on: compute it at three weightings and publish all three. If your answer moves by less than 1.5× across plausible weights, say so — that is a strong result and it means the structure, not your assumption, is driving the number.

Write one sentence: "This body has N people and about X voices."

Exercise 2.2 — Latency and the floor (60 minutes)

For twenty decisions, compute days from first trace to decided, and days from decided to downstream proof. Then find the floor: what cycle does this body sit inside? A term. A season. A funding round. An academic year.

Report excess over the floor. If a society's grant decisions all take five months because the union disburses twice a year, that is the union's calendar, not the society's governance, and saying so is the most useful thing in your report.

Exercise 2.3 — Non-implementation (90 minutes)

Count decisions with no downstream proof within a reasonable window — one term for most student bodies. Compute the share. Then find the silent reversals: decisions revisited under a different name. Match by subject, not wording.

Write down, before you look, what you expect the number to be. Comparing your prior with the result is where the actual learning is.

Exercise 2.4 — The exception share, with its denominator (45 minutes)

Two figures, always together: the share of decisions taken outside the standing process, and the share of the body's actual decisions that the standing process covers at all. The second is usually the shock. A society constitution may govern spending and elections while everything about what the society actually does gets settled in a chat.

Exercise 2.5 — The minutes pass (2 hours)

Count recorded interventions by name across one year. Compute the top decile's share, the number of people appearing exactly once, and the effective number of voices. Then sit with the singleton list for ten minutes before writing anything: those are people who turned up and did not speak, and every one of them has a reason.

Exercise 2.6 — Rule and practice (2 hours)

Inventory every provision in the governing document that creates a right or a duty. For each, ask the archive: was it exercised in the last twelve months? Compute dormancy. Then go the other way and list standing practices with no written basis.

Both numbers matter and the second one matters more, because an unwritten rule cannot be argued with by a newcomer — which is Jo Freeman's whole point and the reason most small organisations are harder to join than they think.

---

PART THREE — THE AUDIT'S OWN DESIGN

Weeks 8–9: what your sample can carry

Exercise 3.1 — Compute your own precision (45 minutes)

Take your register's size n and compute the half-width 1.96 × √(0.25/n). Then apply the design effect for clustering inside meetings: DEFF = 1 + (m−1) × 0.15, with m your typical decisions per meeting. Multiply your half-width by √DEFF.

Write the sentence out: "This audit supports a claim to ±___ points and no finer." Put it on the front page of your report, not in a footnote.

Exercise 3.2 — The observer test (30 minutes)

Attend two meetings. Then compare what you saw with what the minutes recorded. You are not checking whether the minutes are honest; you are checking whether they are informative — whether the disagreement in the room survives into the record at all.

This is a validation sample and it is the correct use of attendance. If the minutes record outcomes and never positions, that is itself a finding, and a large one: it means the archive can tell you what was decided but never who dissented, and half of your six measures are unavailable until the minuting changes.

Exercise 3.3 — Name your holdout (20 minutes)

Choose one number that nobody in this body may ever target — a count of distinct names in the record, say, or first-time speakers. Write down why, and get one other person to agree in writing. That agreement is the most durable thing you will produce this term.

---

THE TERM PROJECT

A two-page governance audit of a real body

Produce exactly two pages.

Page one: the six numbers, with history where you have it. Each with its weights or assumptions printed beside it. No score. No colour. No ranking.

Page two: the boundary statement. What you looked at, what you did not, your computed precision, and the three things this audit cannot see. Then, at the foot, the three best decisions this body has made, from Exercise 1.2, because the profile is only readable against them.

Give it to the body itself before you submit it. Not for approval — for a right of reply, which you print unedited. A person named in a finding should read it first from you and not from a noticeboard.

The assessment criteria, in order of weight:

1. Is every figure reproducible from your register by someone who does not have your notes?
2. Is your precision computed and stated, including the design effect?
3. Does the report name a structural cause for every finding, and no person?
4. Does the boundary statement name three things the audit cannot see?
5. Would the people in the body read it without flinching, and act on it?

---

SELF-ASSESSMENT

Score yourself honestly. These are capabilities, not marks.

The single test of the term. Hand your two pages to someone who was not involved and who does not care. If they can tell you, unprompted, what the body should do next — and if that action is about a rule rather than a person — you have built the instrument. If they say "so is it good or bad?", you have written a score after all, and the fix is to delete the sentence that invited the question.

---

WHERE THIS GOES NEXT

Three doors, all open to you before you graduate.

The comparative audit. The chapter's sharpest sampling result is that clustering lives in the meeting, so four bodies for two years beats one body for eight. Four students, four societies, one shared register format, one term: that is a genuinely publishable comparison and nobody has done it for student organisations.

The archive rescue. Most student bodies lose their records every time the committee turns over. Building a retention rule — where minutes live, who holds them, how far back — takes one afternoon and is the single highest-value governance act available to you. Do it while nobody has a stake in it.

The instrument itself. The six measures fit in a spreadsheet. Build it once, cleanly, with the weights exposed as cells rather than buried in formulas, and hand it on. A tool that outlives your membership is how a student changes an organisation they have already left.

---

PART FIVE — THE ETHICS OF AUDITING PEOPLE YOU LIKE

This is the part of the term most students find hardest, and it is worth its own week.

Rule one: every finding names a structure. "The room has nine voices in it" is a finding. "Priya and Tom dominate the meetings" is a complaint with a statistic attached, and it will end your access, your friendships and your project in that order. The same underlying fact supports both sentences. Choose the first one every time, and notice that it is also the more useful of the two: a structural finding suggests a change to the agenda, a rotation of the chair, a written speaking order. A personal one suggests only that somebody should be different.

Rule two: the singleton list is confidential and the singleton count is not. Publish how many people spoke once. Never publish who. Then go and ask two of them, privately and without an agenda, what would have made speaking easier. Their answers are frequently the best material in the whole report and they are almost never what you expected — in most small bodies the reason is that the item they cared about came last, at the point in the evening when everybody wanted to leave.

Rule three: give the body the right of reply and print it unedited. You are not a regulator. You have no power here and you should not pretend to any. What you have is arithmetic and access, and the report's authority comes entirely from the fact that anybody could check it.

Rule four: hand over the instrument. At the end of the term, give the spreadsheet, the register and the method to whoever is on the committee next year. An audit that only you can run is a piece of coursework. One that the body can run on itself every year is governance, and the difference between the two is an afternoon of documentation.

And the rule under all four. You are studying a body that exists because people volunteered their evenings. Whatever the numbers say, that fact was true before you arrived and will be true after you leave, and a report written as though it were not is a bad report regardless of how correct its arithmetic is. Chapter II.10's finding holds here with particular force: measurement is an intervention. Yours is the first this body has ever experienced, and what it learns about being measured will outlast anything it learns about its own minutes.