← Principles
第 1 章 · 20 分钟
A principle: upgrading one judgement into a rule
'What do I do this time' becomes 'what do I do with this kind', and the judgement is paid for once.
概念地图
Principle
Upgrades 'what now' into 'what with this kind'; the judgement is paid for once.
深色为本章概念，浅色为其他章
陪练模式
不想只是读？让它带你走一遍。
导师把这一章拆成小步，每一步先问你，再讲；你用自己的话答，它先认下对的部分再纠偏。用你自己的 AI，对话只存在这台设备上。
Locus01
延伸阅读
The maxim tradition · from ancient proverbs to early modern rules of conduct
Compresses experience into short memorable lines. It argued against leaving experience locked in individual memory; the cost is that when two maxims conflict nothing adjudicates — 'look before you leap' and 'strike while the iron is hot' sit side by side, and neither says when to use which.
Foundational
Dalio · Principles · the 2011 free PDF
Opens not with entries but with what a principle is and where principles come from, and asks the reader not to adopt anyone's principles unthinkingly. He argued against taking a ready-made package on trust; the cost is that this sits in tension with the two hundred-odd ready-made entries that follow, which chapter 7 counts.
Turn
The 2017 trade book, Principles: Life and Work
The same material rewritten and expanded, with different numbering and wording. Any reading that turns on a particular phrase must name the version it uses — that is this course's version discipline.
Revision
Checklists versus expert judgement · later twentieth century onward
One line holds that writing judgement into explicit checklists reliably lowers error rates, another that a practised expert's judgement beats any checklist when information is incomplete. This dispute sets the boundary of this chapter's mechanism, and the debate block lands on it.
Counterexample
机制02
Upgrading a case into a rule
All two hundred-odd entries in the book are products of one move, and the opening pages set that move out. The word carrying it is principle. Its place in the argument is the entrance: meet something, sort it into a kind, then ask what to do with that kind. What gets saved is not thinking but repeated thinking. So what the book is really teaching is not the entries but how the entries are made.
机制One judgement, through being written as a rule for a whole kind of case, makes the same thinking cost payable once.
可迁移性测试
Move it to incident response: after one investigation, write the test that worked into the runbook, and the next similar alert follows it directly. The isomorphism breaks in that a runbook entry gets checked against later incidents, while a life principle is rarely revisited to see whether it worked.
机制03
A principle has to be violable
The book does not state this one; this course added it, and it is marked because without it the chapter cannot be checked: a principle must let you name what would count as violating it, or it carries no content. 'Be honest' rules out almost nothing; 'say it even when you know they do not want to hear it' rules out a whole class of behaviour. This test connects to chapter 6's radical truth: that one has content precisely because it forbids editing the facts to spare feelings.
机制A rule, through explicitly ruling out some class of behaviour, acquires content that can be checked.
可迁移性测试
Move it to acceptance criteria: 'write high-quality code' changes nobody's behaviour, 'any function over 50 lines must be split' does. The isomorphism breaks in that line counts check themselves, while most principles need a person to judge violations — and that judgement needs rules of its own.
Derivation04
What having principles actually saves
What is being reconstructed is the shape of the book's opening claim: why acting on principles beats deciding afresh, and the range that conclusion is confined to.
Situations of the same kind recur
A premise the author chose. He takes recurrence as the ordinary condition of life and does not defend it; this chapter's boundary grows on it.
One episode of hard thinking costs far more than one lookup
A premise the author chose, put in the book as not having to think it through from scratch each time a hard choice arrives.
A rule written down still applies at the next case of that kind
A premise the author chose. Drift in the situation breaks it, and the book gives no way to detect drift.
推导 · 0 / 4
Worked derivation05
Turning a slogan into a principle
'Communicate earlier next time' — is that post-mortem conclusion a principle?
Apply the test from the second block: what class of behaviour does it rule out?
裂缝06
Cracks in this reading
争议地形07
本书主张
Writing how you handle things into explicit rules and reusing them beats judging afresh each time; whoever does not write them down is paying the same thinking cost over and over.
另一种看法
One line holds that where information is incomplete, time is short and the patterns are complex, an experienced person's judgement beats an explicit rule, and that writing judgement into entries loses the cues that work without being articulable.
分歧扎在
The disagreement is rooted in which kind of situation, not in whether rules help. Both sides grant that recurring situations with clear feedback suit rules; they differ on situations with slow feedback, few samples and unstable patterns.
什么证据能裁决
What would settle it is comparison by setting: in domains with fast feedback and stable regularities, do explicit checklists beat expert judgement on error rates; then run it again in domains with slow feedback and unstable conditions. Criteria fixed in advance, recheckable by a third party.
The balance falls differently by setting: where feedback is fast and regularities stable (verification procedures, surgical checklists and the like) the evidence for explicit rules is fairly strong; where feedback is slow and conditions unstable neither side shows a stable advantage. Most of this book's principles are of the second kind while the author writes in the register of the first. What is missing is a version that writes the applicable setting into every principle.
Falsification08
Turning 'writing rules is cheaper' into a testable specification
The claim under test: writing down in advance how a recurring kind of decision is handled lowers the rework rate for that kind. The book runs no such test; below is how it would have to be run.
Data source: a team's own tickets, incidents or review records, with decision time, decision content and whether rework followed
It has to be a record already being kept; a reconstructed one gets picked to fit the conclusion.
Sample range: at least 60 decisions of the same kind, split into before and after the rule was written
Same kind only; comparison across kinds means nothing. 60 is a rough floor for seeing a difference in rework rates.
Time window: 6 months before the rule was written and 6 months after, with a 1-month transition excluded
The transition is excluded so that 'extra care right after writing it' does not get counted in.
Decision threshold: a relative fall in rework of 20% or more counts as support
Fixed in advance. Below that it cannot be separated from staff turnover and business variation.
Failure condition: rework does not fall, or falls as much in a control group with no written rule
Either counts as the claim being refuted.
推导 · 0 / 3
接口09
Which slot it hangs on
挂在哪个槽位The conclusion you already hold is probably 'run a post-mortem afterwards' — a habit about reviewing. This chapter goes at what the review should produce.
Does this chapter (a) replace 'run a post-mortem', or (b) constrain it: the review is cashed out as one rule that forbids some class of behaviour and states its range?
慢变量Register two slow variables: how many reviews you have run recently, and in how many of them you produced a rule whose violation could be named. Look at the second first.
小结10
本章小结
01A principle is a reusable way of handling a kind of case; what it saves is repeated thinking, not thinking.
02'Writing rules is cheaper' is arithmetic once the count exceeds one, not an insight about life.
03A principle must let you name a violator or it has no content — that test is added by this course.
04Two premises are the fragile ones: no drift, and the rule still applies next time; the book gives no way to detect drift.
05'I act on principle' is consistent with any evidence, so this course books it as unadjudicable.
提取练习 · 合上书，先自己答一遍。
?Is 'writing the rule is cheaper once the count exceeds one' (a) the arithmetic of 'reusable' or (b) an empirical claim observation could overturn?
?Is this chapter's debate settled by (a) whether rules help at all or (b) which kind of situation is in play?
Forced choice11
A · Identify the mechanism
Which of these is a mechanism?
二选一
Forced choice12
C · Locate the crack
In which situation does this chapter's reading give a confident and wrong answer?
二选一
读到这里，把它记为已读。