16.4 — Principles, and Holding Them Under Pressure

13.8 covered how to write a code.

This chapter is about the harder half: what happens to it when it costs something, which is the only time it matters.

The three ways a principle fails

And it is worth knowing which one is happening, because the remedies differ.

It was never real

A principle nobody has ever tested is a preference.

And you will not know which you have until it costs something, which means most people's principles are unclassified.

The test: name a principle you hold. Now name the last time holding it cost you something specific. If nothing comes, that is information — not that you would fail, but that you do not know.

It is revised in the moment

The most common failure, and the most insidious.

You do not break the principle. You discover that it does not apply here, for reasons that are entirely convincing and that arrive in about four seconds.

Ravana's ten heads (2.14): a counter-argument for every warning, produced faster than the warnings could be delivered.

And 2.11's count: one reason is a reason, four reasons is a search.

The remedy is procedural because it cannot be cognitive. You cannot out-think a mind that is generating exceptions. What works: the exception procedure written down in advance (4.11, 13.8), and the requirement to say it out loud to one person before acting.

It is redefined afterwards

The worst, and 3.14 is the case.

Yudhishthira did not lie. He said "Ashvatthama is dead" and then, quietly, "the elephant."

And the reason that is worse than a plain lie is that it comes with a pre-built defence. **A person who lies plainly may stop. A person who has found a construction that preserves their self-image will do it again.

The tell: you are explaining, at length, why what you did was not the thing it obviously was.

And the rule from 13.8: a breach acknowledged as a breach leaves the principle standing. A breach smuggled in under a redefinition destroys it.

What makes a principle hold

Five things, and they are structural rather than about willpower.

It is specific

"Be honest" cannot be held because it cannot be checked.

I do not present a number I know to be wrong, however it is framed can be.

13.4 and 13.8 both said this and it bears repeating: a principle stated as a quality is not operable, and a principle stated as a behaviour in a recurring situation is.

It has been paid for once, cheaply

And this is the most useful thing in the chapter.

A principle you have upheld at small cost is a different object from one you have never tested.

Which suggests a deliberate practice: pay for your principles in the small cases, on purpose.

Say you were wrong about something minor, publicly. Refuse a small advantage that would require a small deception. Correct a number in your own favour. Admit you have not read the thing.

Each of these costs almost nothing and each one converts a preference into something that has held.

And 8.9's version: take a small humiliation deliberately and notice you survived.

Somebody knows about it

A principle held privately can be revised privately.

A principle one other person knows you hold cannot be revised silently, and the difference in holding rate is large.

This is the same mechanism as 13.7's ally and 8.6's naming it out loud.

The cost was priced in advance

Bhishma's failure was not that his vow was too demanding. It was that he never wrote down what it was for (3.2, 3.13), so the purpose dropped out and only the loyalty remained.

Every principle needs a purpose clause and an expected price.

I will not misrepresent a result. This may cost me a promotion. I accept that.

A principle whose price has been named in advance is much harder to abandon, because abandoning it now requires contradicting something you already said.

And it can bend rather than break

8's finding and 10.9's wall: an absolute rule will meet a case it cannot handle (16.1's item eight).

A principle held absolutely will eventually produce something monstrous, and then it will be abandoned entirely, because it had no provision for its own exception.

A principle with a written exception procedure survives being broken.

Which is counter-intuitive and correct: the flexible one is the durable one, provided the flexibility is specified in advance rather than improvised.

The pressures, and what each does

Four, and they attack differently.

Money. The slowest and the most effective. It does not ask you to break the principle; it arranges your circumstances so that holding it becomes progressively more expensive, over years, with each step small. Drona is the case (3.4, 3.8): a man who took a salary and was silent in the hall.

Belonging. The strongest. 13.7's mechanism — the room where nobody is objecting — and the discovery that holding the line means being outside.

Fear. The fastest. And 8.7's rule applies in reverse: nothing irreversible while it is running.

And love. The one nobody prepares for. Dhritarashtra is the case (3.18): a man who could not act against his son and told himself for forty years that he was powerless, and then crushed an iron statue with his bare arms.

The general defence against all four is the same: the decision made in advance, by a version of you who was not under any of them.

What to do when you have already broken one

Because you will, and this is where most people abandon the whole enterprise.

Four steps, from 13.8 and 14.7.

Name it accurately, in the harshest accurate description (2.4). Not softened, not inflated.

Say it out loud to one person.

Repair what can be repaired.

And do not adjust the principle to include what you just did. This is the one that matters. A rule revised to accommodate your last exception is no longer a rule, and the revision is nearly always done silently and feels like maturity.

And then continue. A principle broken once and re-affirmed is intact. A principle broken once and quietly retired is gone.

The person who has them

And what it actually looks like from outside, because it is not what people expect.

Not rigid. Rigidity is the sign of a principle that has never been examined and is being protected from examination.

Not loud. A person who announces their principles frequently is maintaining a position rather than holding a line (8.9).

Predictable. This is the observable feature. People know what they will do. And that predictability is the entire practical value of having principles — to other people, and to yourself.

And 13.8's closing point: a code is not for being good. It is for being predictable to yourself.

What to carry from this page

Name a principle and the last time it cost you something specific. If nothing comes, it is unclassified.

Pay for them cheaply on purpose. Small public corrections, small refused advantages. Each one converts a preference into something that has held.

Write the purpose clause and the expected price.

Do not adjust the rule to include your last exception. That is how they actually disappear — silently, and it feels like maturity.

And the observable feature is predictability, not rigidity or volume.

Next: 16.5 — wisdom, which is the capacity all of this has been circling, and which is a skill rather than a state.