ATTENTION, PEOPLE WHO HAVE ASKED “BUT IS IT LEGAL?” AS THOUGH MORALITY WERE A CUSTOMER-SERVICE CHATBOT THAT CLOSES THE TICKET WHEN IT FINDS NO PROHIBITED KEYWORD!
Are you tired of not having a waiver?
Have you sat in a room where somebody said “technically we’re fine,” and watched four adults visibly decide that the word technically was doing enough work to go home on?
Then congratulations.
You qualify for the Five Number One Rules of Being Ethical After the Easy Case — five Number Ones for the moment when legality has answered its narrow question and everybody discovers there are still people in the building.
“Wait,” you’re yelling at the screen, “they can’t all be Number One!”
That’s exactly what Big Numbering wants, while it sells a front-page test as a substitute for a stakeholder map, and a stakeholder map as a substitute for an independent reviewer.
Ethical failure is a specialist.
Naming every affected party won’t tell you what a headline would reveal.
Passing a disclosure screen won’t trace what a decision causes once other people respond.
Seeing a second-order effect won’t control a self-interested waiver.
And an impeccable code is worth nothing if the person who benefits from the exception also gets to approve it.
Five different ways a cost disappears: out of the map, behind a specific rule, after the first consequence, inside a waiver, or underneath the authority of whoever gains.
So here they are.
Five rules.
Five Number Ones.
No substitutions. No “technically.” No stamping APPROVED on a cost merely because somebody moved it one column to the right.
RULE #1: NAME EVERY PARTY WHO ABSORBS A COST
Introducing STAKEHOLDER-SHRINK-RAY™, the remarkable device that reduces an entire decision to the people in the meeting and the people receiving the invoice.
Everyone else becomes a tasteful blur just past the edge of the slide.
Do the opposite. Before deciding, write down who can affect this, who’s affected by it, who carries a cost later, and whose claim is legitimate even though they have no market power at all.
Then put it on a map, because a map makes the judgment visible instead of instinctive.
Mendelow’s power-interest matrix crosses power to affect the decision against interest in its outcome, and the four quadrants give you a triage order. It doesn’t make attention infinite. It stops limited attention from pretending it has no allocation rule.
When the map has to survive a challenge, the Mitchell-Agle-Wood salience model adds legitimacy and urgency alongside power — three attributes, eight types, including the dormant, the dangerous and the definitive.
Which sounds academic until you notice what it makes sayable: a party can have a legitimate, urgent claim and no power whatsoever.
Merck’s Mectizan decision is what that looks like with money attached. Merck registered ivermectin for human use in 1987. The people facing river blindness across parts of Africa and Latin America couldn’t pay a commercial price, and no government or agency had committed to funding distribution at scale.
A shareholder-only screen sees no purchaser and stops there. It isn’t confused. It has genuinely finished.
In 1988 Merck committed to donate the ivermectin needed to eliminate the disease. The programme expanded to lymphatic filariasis, ran in more than 30 countries, and between 1995 and 2010 was estimated to prevent roughly seven million years of disability.
That doesn’t prove every map produces a heroic answer. Maps are incomplete and depend entirely on how the decision was scoped. It proves something smaller and harder to dodge: “cannot pay” and “does not count” are two different sentences, and only one of them is an argument.
A cost does not disappear because the person carrying it cannot reach the conference room.
The map is where an invisible invoice becomes an agenda item.
RULE #1: RUN THE FRONT-PAGE TEST AFTER THE SPECIFIC RULES
NOW AVAILABLE: NEWSPAPER-PROOF-O-MATIC™ — the machine that asks whether a decision would look fine in public, then uses that warm feeling to drive straight over the conflict-of-interest policy.
The front-page test isn’t a hall pass.
It’s a residual screen, and residual means it goes second.
Read the specific rule first. The conflict provision. The gift rule. The confidentiality duty. The rule on corporate opportunities. The legal duty that applies in this actual organisation, in this actual jurisdiction. Then use disclosure to test whatever those didn’t cleanly settle.
Berkshire Hathaway’s Code of Business Conduct and Ethics is arranged exactly that way — the front-page guidance sits after the provisions on conflicts, corporate opportunities, gifts, insider trading and confidentiality. A screen asks a broad question. A specific rule governs a known category. Reversing them means the broad question gets to overrule the known category, which is a strange thing to have built on purpose.
Then run the procedure around the screen, not just the screen. Get the facts. Ask whether the request seems improper. Clarify who shares responsibility. Discuss it with a supervisor. Use anonymous third-party reporting only when the normal channel genuinely can’t work.
That isn’t bureaucracy in a raincoat. It’s a mechanism for stopping a private rationalisation from becoming the only surviving record of the decision.
And the test works precisely because answering it badly is embarrassing. Could you disclose the decision, its affected parties, and the real reason for the exception — without the sentence going suspiciously passive? Would the explanation still hold if the most affected powerless party were named inside it?
A screen is a flashlight, not a permit.
If the policy says stop, a headline doesn’t get to overrule it with better lighting.
RULE #1: LOOK FOR THE SECOND-ORDER EFFECT
From the makers of FIRST-ORDER-ONLY™ comes the long-awaited sequel: “YES, BUT THEN WHAT?” — in which a decision is permitted to have a second consequence, and occasionally a third, before anyone signs anything.
Second-order thinking doesn’t ask whether the first result is attractive. It asks what that result itself causes, and what happens after everybody else responds to it.
Forrester’s Industrial Dynamics made this formal in 1961 with feedback, stocks, flows and delay: a small retail demand wobble amplifies into enormous swings upstream, with no external shock required. We call it the bullwhip effect. Howard Marks’s version for investors asks what happens once everyone else has reached the obvious conclusion.
Point that lens at an ethical shortcut and watch it stop being cheap.
The waiver first makes a transaction possible. Fine.
Then the beneficiary learns that disclosure is negotiable.
Colleagues learn the code has a side door.
A counterparty learns precisely which incentive is worth cultivating.
“Does this get the deal done?” quietly becomes “what system does this deal teach everyone to operate?”
Inversion arrives at the same place from the opposite side. Jacobi restated resistant problems backwards; Munger made it a general heuristic. So don’t ask how the exception can be defended. Ask what would guarantee an ethical failure here.
Leaving out the party with no power? Concealing the next-best alternative? Letting an interested beneficiary pick their own reviewer? Treating one exception as proof the rule was decorative all along?
Remove the failure conditions before polishing the forward story.
Opportunity cost makes the resistance concrete, because it counts what you gave up rather than what you paid. In one 2009 experiment, simply reminding shoppers what else the money could buy dropped stated willingness to buy a $14.99 DVD from 75% to 55%. That doesn’t prove ethical prompts improve every decision. It shows what happens when an alternative gets named instead of being silently compared against nothing.
So name what the shortcut displaces. An independent review. A fairer allocation. A relationship built on being predictable. A chance to stop a harm that hasn’t happened yet.
A non-financial cost is still a cost. It’s just much better at hiding in a budget meeting.
The first consequence is the sales pitch. The second consequence is the invoice.
An exception that looks cheap at signing may be buying a very expensive habit.
RULE #1: TREAT A WAIVER AS A HIGHER-RISK DECISION
BEHOLD WAIVER-WASH PLUS™: pour one exemption over a conflict of interest and watch the conflict become ethically unscented!
Board approval sold separately. Independent judgment may vary.
A waiver is not evidence that there was no conflict.
It’s evidence that a conflict was correctly identified, and that somebody with authority decided the normal control wouldn’t govern this particular instance.
That raises the scrutiny. It cannot possibly lower it.
Who benefits? Who reviews? What relationship connects those two? What does the exception displace? What happens if the decision gets disclosed alongside the exception that permitted it?
Enron supplies the named failure mode, and the details matter more than the moral. In 1999 CFO Andrew Fastow proposed LJM Cayman and LJM2 Co-Investment — limited partnerships designed to buy underperforming assets off Enron’s books. Enron’s own Code of Ethics would ordinarily have barred it outright: the CFO would be running entities transacting with the company he served.
The rule noticed exactly the conflict it was written to notice. The system worked.
Then the board formally exempted Fastow from the conflict-of-interest provision so he could run the partnerships anyway.
The waiver didn’t make the arrangement harmless. It removed the guardrail from the one transaction whose beneficiary had the clearest possible reason to want it gone. The code kept circulating internally in July 2000, a little over a year before the December 2001 bankruptcy. The LJM entities became central to the off-balance-sheet transactions that inflated earnings and hid debt.
This isn’t evidence that ethics codes are generally decorative. It’s a much narrower and more uncomfortable finding: the institution with the power to enforce a code is also the institution that can suspend it, for the person who gains.
A waiver is not a smaller conflict. It is a larger question about who gets to suspend a control.
If an exception needs a spotlight to look ordinary, it isn’t ordinary enough.
RULE #1: ESCALATE BEFORE SELF-INTEREST WRITES THE EXCEPTION
THIS IS THE ONE THEY TRIED TO KEEP IN THE LOCKED CABINET: CALL-SOMEONE-ELSE 3000™.
No glowing button. No chrome. Its only feature is that the person profiting from the exception doesn’t get a private conversation with their own conscience and a rubber stamp.
Write the escalation route before the pressure arrives.
State the request. The affected parties. The specific rule. The disclosure concern. The foreseeable second-order effect. The independent reviewer. The next channel if the normal route can’t work.
A process assembled after a lucrative exception has already appeared is a process being built while falling down the stairs.
The order is what does the work: facts, then impropriety, then shared responsibility, then supervisor, then external reporting if the internal one is compromised. That sequence creates a record and distributes responsibility before a single interested story hardens into “the business case.”
Escalating isn’t surrendering judgment. The reviewer still needs the map, the specific rule, the disclosure test, the inversion, the opportunity cost, the downstream chain. What it adds is distance between a decision and its immediate beneficiary — and that is the one ingredient a self-justifying exception can never manufacture for itself.
There are real limits here, and they’re worth stating. Ethical screens expose conflicts; they don’t mechanically resolve them. A power-interest map gives triage order, not truth. Codes and legal duties vary by organisation and jurisdiction, and none of these models is imposed by a regulator as general compliance.
All of which argues for making the route more explicit — not for letting the exception appoint its own referee.
Escalate while the decision is still a question, not after it has become somebody’s bonus.
The route isn’t mistrust of a person. It’s mistrust of the extremely convenient timing of a conflict.
BUT WAIT, THERE’S MORE!
“What if the affected party has no power?”
Same rules. Map the claim anyway. Power is one attribute; legitimacy and urgency are the other two, and Mectizan is what it looks like when someone who cannot pay still counts.
“What if the policy doesn’t expressly forbid it?”
Same rules. Check the applicable provisions first, then run the disclosure screen inside the procedure — facts, responsibility, supervisor, and the actual affected parties named out loud.
“What if the board approved the waiver?”
Same rules. Approval records that a control was suspended. It doesn’t erase the conflict, the beneficiary’s interest, the reviewer relationships, the displaced alternative, or the precedent now sitting in the filing cabinet.
“What if it genuinely helps the organisation?”
Same rules. Name what it displaces, then trace what it teaches everyone else to expect. The first benefit may be entirely real. It just isn’t the whole causal chain.
“What if it’s legal here?”
Same rules. Legal duties vary. The architecture doesn’t: map who bears the cost, test disclosure, trace the effects, raise scrutiny on exceptions, and keep the decision away from its beneficiary.
The details change.
The architecture doesn’t.
THE FIVE, WITHOUT THE WAIVER STAMP
Name every party who can affect or absorb the cost — including the one with a legitimate claim and no leverage.
Apply the specific rules first, then run the front-page test as a residual screen inside a real escalation procedure.
Trace what the first consequence causes next. Invert the problem to find what would guarantee failure. Count the better alternative you displaced.
Treat a waiver as a higher-risk governance decision, not as proof the conflict was small.
Escalate before the pressure arrives, in writing, with the reviewer named.
DO THIS TONIGHT, BEFORE THE LOOPHOLE COOLS
Pick one pending decision. A vendor choice, a hiring referral, a gift, an expense, a policy exception.
Write down every party affected by the result — and make yourself include the one with no leverage. Put them on a power-interest grid. If a low-power claim feels easy to skip, score its legitimacy and urgency separately, then write the sentence explaining why it does or doesn’t count.
Find the actual specific rule before reaching for a broad ethical slogan. There usually is one, and it’s usually more precise than the slogan.
Then write the front-page sentence: decision, exception, beneficiaries, affected parties, cost. If it needs the passive voice to sound decent, that sentence has just performed a public service.
Now invert it. Write the version of this decision that would be indefensible, and check how far away you actually are.
Trace the second-order chain — what the first result permits, signals, or teaches. Name the next-best alternative the shortcut displaces.
Finally, write the escalation route: supervisor, independent reviewer, and the channel to use if the normal one is compromised. Do it now, while nothing is at stake, because the whole value of that route is that it existed before somebody’s self-interest turned “we should check” into “we already promised.”
For the low, low price of refusing to let invisibility do moral work, the complete Five Number One system is yours.
No subscription. No ethics sticker. No complimentary loophole magnifier with every board packet.
And if you act now, we’ll include the only feature that has ever mattered, at no additional charge:
a decision that can survive being seen.
Operators are no longer standing by.
The operator is whoever has to explain this later.