Haute Lumière
Commerce · I.09 · MMXXVI · daylight
For the person studying this alone, or in a seminar, with no organisation to change yet. You are surrounded by institutions already. You are inside one now.
The chapter was written for someone carrying a verified result toward a procurement department. You may have neither. What you do have is the single best laboratory for this material that exists: a university, which is a large, rule-bound, precedent-governed institution that you are already a member of, with published regulations you are entitled to read.
Everything in the chapter can be practised here at full scale and at no risk. The scheme of delegation exists. The standing policies exist. The committee calendar is published. There are officers whose whole expertise is what the institution is allowed to do, and almost nobody ever asks them anything interesting.
The transferable skill is not persuasion. It is reading a rule in its original form and finding the category. That skill will be worth more to you across a career than any technique in this volume, because it applies to a landlord, a bank, a local authority, a funding body, a hospital, a visa office and every employer you will ever have.
Exercise 1.1 — Obtain the three documents (2 hours)
For your university, or your college, or your employer if you have one, find and download:
If any is not published, write and ask. The email is three sentences and the practice of sending it is half the exercise. Note how long the reply takes; that is your first real datum about this institution.
Exercise 1.2 — The threshold map (90 minutes)
From the scheme of delegation, build a table: role, figure, and the carve-outs that sit outside the figure regardless of value. Most schemes except property, employment and anything multi-year.
| Role | May sign alone up to | Excepted regardless of value |
|---|---|---|
Then answer, in writing: what is the largest thing one person here could agree to this afternoon without consulting anyone? Most students have never considered the question, and the answer is almost always larger than they expect.
Exercise 1.3 — Find the sentence (60 minutes)
Read the standing policy all the way through — not the summary, the document. Find one sentence that authorises more than the institution currently does with it. Write it out by hand with its paragraph number and the date of adoption.
You now hold a category. Keep it.
Exercise 1.4 — The appreciative interview (45 minutes, with a real person)
Find the officer responsible for that policy and ask them exactly this:
"I'm studying how institutions decide what they're allowed to do. Can you tell me about a time this policy let you say yes to something unusual — what it ended up being classified as, and who worked that out?"
Then be quiet and take notes on the classification, not the outcome. You will learn more in forty-five minutes than in a term of reading, and you will have made the most useful acquaintance available to you on that campus.
Exercise 1.5 — The precedent hunt (75 minutes)
Published minutes are the most under-used resource available to a student of institutions. Find the last year of minutes for one committee and read them for the things you were not looking for: what was approved, what was deferred, what was approved subject to something, and what was noted without decision.
Build a tally.
| Outcome | Count | Typical value | Median days from submission |
|---|---|---|---|
| Approved | |||
| Approved subject to | |||
| Deferred | |||
| Noted, no decision |
Two numbers come out of this and both are yours to keep. The first is the institution's real pass rate, which the chapter assumes at 0.80 and which you can now replace with a measured figure. The second is the proportion of items that were noted rather than decided — the chapter's central failure mode, visible in black and white. Most people argue about institutions from anecdote. You are about to have a dataset.
Write one sentence naming what the minutes show that you did not expect.
Exercise 2.1 — Reproduce the chapter's figures (2 hours)
Do not take them on trust. Open lib/verify/I_09.py, read it, and recompute independently — by hand, in a spreadsheet, in whatever language you use.
91.3 / 2 + 14. Confirm 59.7 days.1.07⁻¹² × 0.85¹². Confirm 0.0632.6 / (40 − 6). Confirm 17.6 percent.Then the exercise that matters most: change one assumption and watch what happens. Set the per-layer pass probability to 0.95 and recompute the two-layer markup. It falls to 1.02. Now write one paragraph answering: does the chapter's argument survive that? Defend your answer. This is what it means to read a model rather than quote it.
Exercise 2.2 — Your own delegation ratio (45 minutes)
Find something you actually want that requires an institution's permission — a room, a fund, a change to a module, a piece of equipment, a society. Establish: the smallest version of it that would still be worth having, its cost, and the authority of the lowest officer who could approve it.
delegation ratio = your ask / that officer's authority
If it is above 0.60, redesign until it is not. Write both versions down. Keep them; you will submit the smaller one in Part Three.
Exercise 2.3 — The calendar, backwards (30 minutes)
Take the committee calendar. Pick the meeting at which your result would be heard. Count backwards: papers deadline, verification time, doing time. That gives you a start date. Put it in your diary. If it has already passed, you have just learned the chapter's central lesson at no cost.
Exercise 3.1 — The one-page memorandum (3 hours)
Write a real memorandum of understanding for the thing from Exercise 2.2. One page. It has:
Exercise 3.2 — Submit it (real, not simulated)
Send it. Actually send it. To the officer you interviewed, or to the one your threshold map identifies. The ask should be small enough to be genuinely easy to say yes to.
Three outcomes, all of them valuable. Yes — you have run the method end-to-end before your first job. No, with a reason — write the reason down; it is almost always a category problem, and you now have the data to test the chapter's central claim against a real case. Silence — note how long, and what the papers deadline was.
Exercise 3.3 — The counter-reading (90 minutes)
Take the strongest position against this chapter and write it out in 500 words: that treating institutional refusal as a classification problem is a way of avoiding the possibility that the institution has considered your proposal and correctly declined it. Make it as strong as you can. Then write two paragraphs on what, in a specific case, would let you tell the two apart.
A method you cannot argue against is a method you do not understand.
Exercise 4.1 — The register (45 minutes)
Start your own. A single file with one row per approach you make: the institution, the officer, the category cited, the ask, the ratio, the date, the outcome, and the elapsed days.
Twenty rows is an education nobody can take from you. Fifty rows and you can answer questions about institutional behaviour that most consultants cannot, with your own data. A number nobody keeps is a number nobody improves.
Exercise 4.2 — Teach it once (2 hours)
Explain the delegation ratio and the break-even markup to one person who has to deal with an institution — a flatmate arguing with a landlord, a friend applying for a grant, a society treasurer. Twenty minutes, on paper, with their real numbers.
You will discover which parts you actually understand. That is what teaching is for, and it is the cheapest diagnostic available.
Exercise 4.3 — Notice the pleasure (20 minutes)
After your interview, write down the moment the officer's posture changed. There usually is one. Note what you had just said.
People inside institutions are not guarding the gate. Most have never been asked to open it in a way that would let them, and being asked well is a good day at work for them. This economics is not adversarial and it is not sacrifice. It is finding out what somebody is already permitted to do and making it easy for them to do it.
Choose one institution you have real access to. Over a term, produce a single document of eight to twelve pages:
Assessment criterion, and it is the only one: could a stranger arriving at that institution next year use your document to get something useful agreed faster than they otherwise would? If yes, you have written a piece of infrastructure in a term, which is more than most graduate theses manage.
Score each honestly, one to five.
| 1 | 5 | |
|---|---|---|
| I can find and read a scheme of delegation | Never seen one | Can locate the carve-outs in ten minutes |
| I compute before I argue | Quote figures | Recompute and test the assumptions |
| I find categories rather than make cases | Argue the merits | Cite the policy back, by paragraph |
| I size to the signature | Design then look for approval | Find the figure first, design to 0.60 |
| I plan backwards from a deadline | React when ready | Papers deadline is in my diary |
| I ask officers good questions | Ask for permission | Ask what this is an instance of |
| I keep a register | Remember roughly | Rows, dates, elapsed days |
Anything at two or below is your next month. Re-score at the end of each term rather than at the end of each exercise; the movement over a term is the signal, and the movement over a week is mostly mood.
You will join an institution. Within a fortnight of arriving, obtain the scheme of delegation and read it. You will know something most of your colleagues will never learn, and you will know it in your first month rather than your seventh year.
And keep the register going. The thing nobody tells you about institutional work is that it is empirical, it is under-measured, and almost nobody keeps their own data. Twenty rows makes you useful. Two hundred makes you the person who is asked.