Chapter 4: Judgment: Adaptive Insight as an Organizational Asset

1. Two geniuses, two burial methods #

On August 18, 1992, Wangan Computer filed for bankruptcy protection. It has been two years and five months since its founder Wang An passed away.

This company deserves the first half of the word "great": Wang An started with calculators, and before electronic calculators were broken down by Japanese manufacturers, he decisively switched to word processors. Both transformations were at the right time. The company has maintained an annual growth of 40% for ten consecutive years, with revenue reaching a peak of US$3.07 billion in 1988, and more than 30,000 employees. Wang An is called "the Doctor" within the company. This title accurately describes the company's judgment structure: all important product and strategic judgments ultimately pass through that head. That brain worked very well, and it worked well for nearly forty years.

Then it stopped working. On March 24, 1990, Wang An died of esophageal cancer. The company he left behind never got a major judgment right again, and two and a half years later it was in bankruptcy court.

In the same era, another Chinese entrepreneur did the exact opposite experiment. Duan Yongping gradually faded out of daily operations after 2001, and BBK was split into three, evolving into OPPO, vivo and BBK Education. He later recounted: "I have never controlled OPPO or vivo." Without remote control, these companies have repeatedly made key judgments (converting functional phones to smart phones, deepening offline channels, and high-end breakthroughs) over the next two decades, and their judgment styles are surprisingly consistent, so consistent that people in the industry can recognize at a glance that "this is the BBK style of play."

The personal judgments of both founders are top-notch. The difference lies in where the judgment is stored: Wang An's judgment is stored in his mind and buried with him when he dies; Duan Yongping's judgment is written into standards ("duty" is expanded into four executable interpretations at OPPO) and stored in the organization. When people are gone, the standards are still in business. Personal judgment is luck, organizational judgment is ability: The former depends on whether the person is in good condition today and whether he will be there tomorrow, while the latter does not depend on any one person.

This chapter answers the question left at the end of Chapter 3: If individual adaptive insight cannot be scaled, how can it become the ability of a thousand people? The answer is in two steps: first look at the structure of this ability (dual ability model), and then look at how to test whether it is a real asset or fake prosperity (five indicators).

2. Dual ability model: the picture in the book #

If you could compress this book into one picture, it would be this one. The horizontal axis is the degree of unification of judgment standards: whether people (and AI) in the organization use the same ruler when making choices. The vertical axis is contextual unity: whether people (and AI) in the organization see the same picture of facts when making judgments. Two axes delineate four quadrants:

Bottom Left: Old Organization. The standards are not unified and the context is not unified. Everyone judges based on their own experience and acts based on the information they have. The picture of "a hundred arrows shooting in a hundred directions" in Chapter 2 lives in this quadrant.
Top left: There is consensus but no standards. The context is unified (everyone has the same understanding of the status quo, and the atmosphere is good), but there is no unified ruler. In this kind of organization, discussions are lively and decision-making relies on the voice or the boss. The typical experience is "everyone agrees at the meeting, and everyone does their own thing after the meeting."
Bottom right: There are standards but no consensus. The walls are covered with institutions and values, but members of the organization see a different picture of the facts. Standards operate in a distorted context, producing precise errors. Nokia in Chapter 9 will show how this quadrant dies: the standards are fine, the information fed to the standards is layer upon layer of glorification.
Upper right: AI-native organization. Same ruler, same picture. Judgment can occur at any node (frontline employees, newcomers, AI), because no matter who makes the judgment, the ruler used and the facts are the same.

This picture has an extra layer of meaning in the AI era that it did not have in the old era: Your AI is also an employee, and it also eats these two things. Whatever judgment criteria are given to AI, it will optimize in that direction. Klarna's AI adopted the "cost first" criterion and executed exactly this mistake at a scale of 2.3 million conversations per month. What context is given to the AI, what judgments will it make based on. MD Anderson's Watson did not share the same medical records system with the hospital until its death. The coordinates of the two samples in Chapter 2 can now be found on this picture: they did not recruit AI, but recruited AI into an organization in the lower left quadrant.Therefore, in the sentence "Great companies need and only need two abilities", the three words "only need" are the key points. Strategy, product, marketing, financing, these are all important, of course, but they are all outputs of judgment. If these two things, judgment standard and context, are right, the rest will be continuously produced; if these two are wrong, everything else will amplify the error. Implementation used to be high on the "need" list. Now that the AI has taken it away, all that's left on the list is exactly these two.

3. The ruler itself: Judge by Value #

The horizontal axis of the dual-ability model talks about "unified judgment standards", and the natural question is: What standard does it unify? There must be a bottom-level ruler.

My answer: Judge by Value. Don’t judge by emotion (I don’t like this plan), don’t judge by status (this is what the director mentioned), don’t judge by busyness (this team works the most overtime), don’t judge by short-term volume (this feature has the highest demand); only judge by what value a thing creates, for whom it is created, and when it will be realized. This is not a moral initiative, but an engineering conclusion: any problem has N solutions, and the N solutions must be sorted, and the sorting must have a basis; emotions, identity, busyness, and volume of voice can all be used as the basis for sorting, but only the results of sorting by value are directly linked to the survival of the organization. (The full version of this derivation is in Chapter 7, where it is shown how it is forced out of the methodology.)

Emotion, identity, busyness, and volume of voice, these four fake judgments are at least visible. The fifth and most hidden one is stance. I use a widely circulated joke to demonstrate it. I laughed the first time I heard it and broke into a cold sweat after laughing. Same thing: cheating. When someone you don't know cheats on you, your judgment is "This person has a problem with his character"; when a friend cheats on you, you give him a few words and you have already made up the reasons for him in your heart; when it's your turn, your conclusion becomes "there are some things that you really can't do anything about at the time." Same behavior, three answers. What changes is not the facts, but the distance between the judge and the facts. When the ruler is not written down, the position is the ruler. What's even more terrible is that the person changing the ruler has no feeling at all: in each of the three judgments, he is sure that he is objective. Social psychology calls this phenomenon actor-observer bias (Jones & Nisbett, 1972): to explain other people's behavior, use character; to explain one's own behavior, use situation.

When it comes to the organization, this joke is played out every day, but the "derailment" is replaced by delays, overspending, and lost orders: delays by other teams are due to poor execution, and delays by our own team are due to many objective difficulties; price reductions by competitors are vicious competition, and price reductions by ourselves are strategic profit concessions. Therefore, the real enemy of "Judge by Value" is not that there is no ruler in the organization, but that everyone has a ruler in their pocket that can expand and contract according to their position, and they can measure it with ease. This is also the deep reason why Chapter 6 insists on writing down the standards: explicitness is not only for inheritance, but also for taking the ruler out of everyone’s pocket and nailing it to the wall. **The ruler nailed to the wall has the same scale when measuring others and measuring yourself. **

The most common place where this ruler is used incorrectly is in the time dimension. There are two accounting periods for value: short-term value is usually about cash, survival, and limiting resources; long-term value is usually about capabilities, standards, talents, trust, and other things that will compound interest. "Judgment by value" requires putting the two into the same set of coordinates, rather than choosing sides. People who use "long-termism" to avoid short-term survival problems and people who use short-term income to overdraw long-term trust make the same mistake: they only look at one account period. Wangan Computer made two mistakes in its early days: insisting on borrowing to avoid dilution of equity (using a long-term ruler to calculate short-term survival), and launching fourteen products at a time, most of which were never delivered (exchanging long-term credit for short-term buzz).

The ruler also has energy to reach people. Business is not value, and position is not value. What is ultimately created and accumulated is value. When this sentence is applied to an organization, it means that when evaluating a person, one does not look at how full his schedule is or how high his title is, but rather what he leaves behind in the things he handles: How many more verified standards are there? How many more reusable assets are there? This is exactly the judgment version of the group of gene replacements in Chapter 2 "Performing Worship → Judgment by Value".

Finally, the complete structure of the word "judgment standard" is explained clearly, because it is not a lonely ruler, but a three-layer judgment system:

Value sorting (why): The lowest level of sorting basis, which is Judge by Value, has just been explained.
Thinking Framework (How to think): The process rules for how conclusions must be derived and evidenced. First principles, two-sided samples, and statistical medians all live on this floor. What it manages is not which one to choose, but the production process of options and conclusions.
Decision Rules (What to Choose): Trade-off terms in specific scenarios. Cash flow takes priority over beautiful statements, responsibilities are divided into four sections, and there is no meeting without DEMO.

All three levels meet the standard definition (can be written, borrowed, verified, inherited), so the horizontal axis of the dual-ability model says "unification of judgment standards", which refers to the unification of the entire system. The manufacturing industry has long had isomorphic divisions: process standards govern the process, and acceptance standards govern the results. Both are called standards. The most common symptom of an organization that misses the middle layer is that the values are consistent and the decision-making rules are there, but the two teams reason in completely different ways and reach irreconcilable conclusions. The ruler is the same, the process is different, and the output is still the same.## 4. From four actions to two abilities: organizational translation of adaptive insight

Chapter 3 breaks down adaptive insight into four actions, and declares at the beginning of this chapter that it cannot be scaled. Now connect these two sentences together and see clearly where the "unscalable" stuck is and how the dual-capability model solves it.

There is no problem with the four actions running in a person: he finds the gap by himself, borrows by himself, judges and transfers by himself, and verifies by himself. The problem lies in the fate of the results: the standards he has verified are stored in his mind. The second person encounters the same problem and starts all over again with four actions; in an organization with a thousand people, the same standard is repeated a thousand times, or worse: a thousand people borrow a thousand conflicting standards, each verify it, and each believes in it. The stronger the individual's adaptive insight, the more hidden this waste is, because every part looks smart.

What the dual-ability model does is to install organizational-level exits and entrances for the four actions:

Unification of judgment standards (horizontal axis), undertaking actions two and three: Standards borrowed and verified by anyone will enter the organization's public standard library, and will be distributed to everyone and all AIs after a unified caliber. The tuition paid by one person is free for a thousand people.
Unification of context (vertical axis), taking over Action 1 and Action 4: Where is the standard gap, which standard is being verified, which one has just been falsified, this information enters the shared context. Therefore, finding gaps no longer relies on someone's acumen, and correcting deviations no longer relies on someone's candor.

There is a gate that must be set up for this translation: not all experience is eligible for inclusion. Only experiences that have been compared, verified, reflected on, and can explain the migration conditions can be upgraded to organizational judgment; otherwise, the standard library will become an empiricist warehouse, and unified errors will be issued uniformly. (The complete process of this gate is the content of Chapter 7, so here are the rules first.)

In one sentence: personal adaptive insight production standards, organizational judgment storage, unification and compound interest. The former is a mine and the latter is a power grid. Without a power grid, no matter how many mines there are, they can only light the miners’ own headlamps.

5. Five indicators: test whether judgment is a real asset #

"Our company has judgment" is a statement that cannot be refuted, so it has no information. In order for judgment to become a manageable asset, it must first become testable. I use five indicators:

Win Rate: Can you win key competitions across cycles? Winning only once may be due to luck. Cross-cycle means that the environment has changed several times, but you still win. Luck will not be so patient.
Correction rate: After failure, the speed and cost of discovering errors, revising judgment, and resuming action. This metric is more honest than winning percentage because it can only be measured in losses. An organization that never fails has no correction rate data, only the risk of failure to deliver.
Replication rate: Without the founder or top leader, can the organization continue to produce people with similar judgment styles? This is a direct test of whether the judgment is in the organization or in the person.
Migration rate: Can the same judgment framework still produce results if it is changed to another industry, another region, or another task? If it can be transferred, it means that the organization has mastered the standard; if it cannot be transferred, it means that it only has the experience of one scenario.
Spillover rate: Has anyone outside the organization started to actively use your language, framework, and judgment standards? Pay attention to the difference between it and brand awareness: awareness is "others know you", and spillover is "others think with your ruler".

The first two tests are the quality of the judgment (win and can afford to make mistakes), and the last three tests are whether the judgment can get rid of the individual and form compound interest. A rule of thumb when assigning each of these five items a score of 1 to 10 is not to average the score: an average score will allow an organization with a replication rate of 1 to achieve a decent total by relying on a high win rate. First look at the median of the five items, and then focus on the lowest item: the lowest item is almost always a bottleneck that prevents judgment from forming compound interest. Wang An's computer's winning rate is 8 points, and the copy rate is 1 point. That 1 point was the entire reason for August 18, 1992.

6. Three sample libraries: Don’t compare scale, compare mechanism #

I have selected three positive examples of judgment capitalization, and first established the rules of comparison: the Chinese Communist Party is a very large-scale political organization, Duan Yongping’s circle of influence is a network of entrepreneurs derived from a common culture, and the Musk business group is an individual-centered cross-company network. The three are not of the same kind. They do not compare in terms of scale, performance, or "who is the strongest." They only compare one thing: the mechanism through which the judgment of a few people becomes an organizational asset. **Duan Yongping’s circle of influence, the load-bearing wall of this chapter (all pedigrees can be tested). Chain: In 1995, Duan Yongping left Xiaobawang and founded BBK, and the six middle-level executives he took away were Chen Mingyong, Shen Wei, and Jin Zhijiang; in 2001, BBK was divided into three parts; in 2004, Chen Mingyong founded OPPO based on the audiovisual business. The key move to capitalize judgment occurred in 2004-2005: OPPO expanded "duty" from one word into four executable interpretations: isolate external pressure and temptation, blame yourself first when problems arise, do not take advantage of others, and duty over integrity (you must do things you have not promised but should have done). Please pay attention to the essence of this action: a personal character is translated into a judgment algorithm: how to sort out temptations, who to check first if there is a problem, and where to draw boundaries, all of which are executable, testable, and can be passed on to the next person. Four of the five indicators passed (the list of events of the same caliber for correction rate is not complete yet, and the boundaries of judgment will be explained): winning rate (functional machines, smart phones, and high-end machines are all on the table in three cycles), replication rate (Chen Mingyong, Shen Wei, and Jin Zhijiang are each at the helm independently, and the judgment style is identifiable and consistent; Huang Zheng calls himself "the fourth disciple of the next generation". Duan himself has not formally accepted the disciple, so he is labeled here as he calls himself), migration rate (the same set of standards has moved from educational electronics to mobile phones to e-commerce), and spillover rate ("duty" and "dare to be the queen of the world" have become industry language, and some people blurt it out in meetings in the investment circle). **A strong man can win once, and a strong organization can replicate the winning method. **

Musk Enterprise Group, List of Mechanisms (I only quote mechanisms, bilateral verification of effect data is still in progress). There are four identifiable mechanisms: First principles serve as a unified thinking framework, that is, the process standard for judging the system, stipulating how conclusions must be derived (unverified industry practices are not accepted, and all costs are re-derived from physical boundaries). Thousands of engineers therefore share the same set of reasoning processes, not just the same A slogan; use extreme goals to compress context ("reduce launch costs by two orders of magnitude" synchronizes the choice direction of thousands of people); high-frequency controllable failure as a correction mechanism (exploding rockets is part of the test schedule, not an accident); mission selection of talents as a means of replication (replace management levels with target concentration). These four mechanisms point to the same effect: small core teams undertake tasks that only giant organizations could complete in the past, and judgment density replaces headcount density.

The Communist Party of China, the ultimate sample of scale (My research boundaries: I only study the four organizational mechanisms of how to unify judgment standards, how to create context on a large scale, how to correct deviations, and how to form spillovers through organizational language. I only use publicly published primary documents, and do not involve political positions and historical evaluations. The description of the mechanism does not make value judgments). Its value to this chapter lies in its scale: a set of value ordering and organizational language that spans more than half a century, covers tens of millions of members, and maintains a unified judgment framework and context-making capabilities through multiple major corrections, proving that the dual-capability model has no scale upper limit. The specific mechanism is left to the consensus engine in Chapter 9.

After the exhibits were removed from the three sample halls, the remaining common skeleton was the same: stable value ranking, explicit judgment standards, unified context, decentralized execution, and rapid correction. The scale is five orders of magnitude different, but the skeleton is exactly the same. This is the strongest evidence that "judgment is an organizational asset rather than a personal talent": if it is a talent, it is impossible for the same skeleton to grow in such different containers. The argument in this chapter also deliberately does not rely on any single sample: remove any one of the three museums as a whole, and the skeleton will still stand.

7. Wang An’s computer on the dissecting table: the complete pathology of strongman dependence #

Now go back to the sample at the beginning and do a complete dissection using the dual-ability model and five indicators. The popular version of Wang An's cause of death is "missing the PC", and this attribution stops at the symptom level. Missing the PC was an error of judgment; the real cause of death was that there was no mechanism in the organization to correct "the Doctor's" error of judgment.

There are three lesions, each with details that can be examined from multiple sources:

Less 1: Succession is based on blood, not standards. In 1986, Wang An appointed his 36-year-old son Fred as president despite the collective opposition of the senior management; John Cunningham, a veteran who had followed him for 18 years and served as president, resigned immediately, and a group of core executives left. Pay attention to the information leaked by this decision: in Wang An's judgment system, "my judgment can be continued by my bloodline." Judgment is understood as inheritance, and the way inheritance is passed on is inheritance, not reproduction. Compare Duan Yongping: BBK's successor is a middle-level employee selected and trained by the same set of standards. He has zero blood ties, but the consistency of his judgment style is high enough to be recognized by the industry.Lession 2: Standards never leave the mind. There is no doubt about Wang An’s judgment (the timing of the two transformations can be called a textbook), but in the past forty years, this set of judgments has never been written into a testable and inheritable format. So the company's strategy swayed with his personal intuition: Fourteen hardware and software products were released at once in 1983, most of which were never delivered, and their credibility was damaged; the interface of the dedicated word processor was copied by PCs running cheap software, and the price was broken down. There was no public ruler in the organization that could be used to question "Why hasn't the Doctor turned around yet?" Judgment is not in people’s minds, it’s in the organization’s standards. The kind in my head that I fall asleep every night and one day fall asleep forever.

Trouble 3: Even correction is an individual act. In fiscal 1989, when the company lost $424 million, Wang made exactly the right course of action: firing the president. It was his own son who was fired, just weeks after surgery for esophageal cancer. This detail is often told as a tragic story, but what is truly amazing about it is the mechanism level: the only person in the company who has the power to correct wrong judgments is the person who made the wrong judgments; the entire correction mechanism is the self-denial of a dying old man. After his death, the company no longer even had this mechanism. In FY1990, revenue was still US$2.5 billion, with a loss of US$715.9 million; it went bankrupt in August 1992, with liabilities of US$550 million.

This is the old gene that this chapter will replace: strongman dependence, that is, placing the organization's judgment on individual strongmen, and replacing the implementation of standards with loyalty to the strongman. The scary thing is that the stronger the strongman, the more comfortable this gene is: Wang An's mind has been easy to use for forty years, so good that no one (including himself) feels the need for a set of standards that can operate without him. If judgment only exists in people’s minds, the founder’s obituary is the company’s obituary. The new gene is what we have discussed in this chapter: the capitalization of judgment—refining it into standards, precipitating it into context, and allowing all employees (and all AI) to inherit it.

8. Boundaries of Judgment #

Three points, stop this conclusion in the direction that is most likely to go astray.

First, shared judgment standards are not equal to one word. What is unified is the sorting basis and boundaries (what counts as value, what priorities should be used when encountering conflicts, and what cannot be done), but not every specific answer. On the contrary, the better the standards are unified, the more decentralized the implementation can be: the premise of Duan Yongping's "never remote control" is that there is no need for remote control, and the ruler is already in everyone's hands. Yiyantang is a form of strongman dependence, which is the opposite rather than the extreme form of this assertion.

Second, the five indicators are rulers, not precision instruments. When scoring your own organization, the absolute number of scores is not important, but the gap between the five items is important; when comparing across organizations, you only compare whether the mechanism exists, not the score, because there is no comparable standard for winning rates in different industries.

Third, truthfully explain the quality of the sample. The genealogy and philosophy texts of Duan Yongping’s circle of influence are all available for examination, but I have not collected a complete list of failures and corrective events of the same caliber (OPPO’s early transformation setbacks, vivo’s hesitation in smartphone transformation), and I am still missing half of the homework to prevent survivor bias; the Musk Enterprise Group only cited mechanisms in this chapter, and the bilateral data on the effects are being verified; the first-hand document collection of Chinese Communist Party samples has just started. Therefore, the evidence structure of this chapter is: two samples of Duan Yongping’s circle of influence (positive) and Wang An’s computer (reverse) are used as load-bearing walls, and the other two museums serve as circumstantial evidence. The load-bearing walls are sufficient to support the assertion, but a more complete specimen would make it stronger. This account is recorded here, not hidden.

What to Do Monday Morning (No. 1 perspective) #

One piece of paper, two tests, forty minutes:

Test 1: Standard Audit. Write down the company’s three most recent important decisions, and answer four questions for each: What are the criteria based on? What value is created for whom and when will it be realized? What scarce resources are consumed? Do short-term gains harm long-term value? Failure to write a standard decision is on-site evidence that the judgment has not been capitalized: that decision is currently stored in someone's mind.

Test 2: Five-index physical examination. Winning rate, correction rate, replication rate, migration rate, spillover rate, each scored 1-10 points. Discipline: Not counting the average score, first look at the median to determine the overall water level, and then focus on the lowest item. Then ask a question about the lowest item: If I start taking a year-long vacation tomorrow, how many points will this item become? After scoring, you will know whether your company is more like the BBK system or more like Wang An Computer in 1989. It still had $3 billion in revenue that year, and it looked like nothing was going to happen.

Note (individual and team perspective): The replication rate among the five indicators can be asked to the individual: Among the people you have led, how many of them have not had a decline in judgment after they left you? Managers whose answer is zero, no matter how good their performance is, are becoming Wang An of their own teams.

Quotable Lines1. AI can do things for you, but it cannot decide for you what is worth doing - this is not a technical limitation, but a definition: once it is decided, the company belongs to it. #

Personal judgment is luck, and organizational judgment is ability.
Judgment is not in people’s minds, but in the organization’s standards.
Great organizations do not never fail, but they are able to win key competitions across cycles and correct mistakes faster than others after failure.
A strong man can win once, and a strong organization can copy the winning method.
Busyness is not value, and position is not value. What is ultimately created and accumulated is value.
Personal adaptive insight will retire, leave, and decline; organizational judgment will not.
If judgment only exists in people’s minds, the obituary of the founder is the obituary of the company.
The stronger a strongman is, the more comfortable he will be relying on this gene.
An organization that never fails has no correction rate data, only risks that have not yet been paid.
When the ruler is not written down, the position is the ruler.
The ruler nailed to the wall has the same scale for measuring others and measuring yourself.

Chapter Acceptance Self-Check (compare with the five acceptance standards of the chapter) #

The assertion can be restated in one sentence ✓, and is an inference of the core assertion (the core assertion asserts that adaptive insight needs to be solidified into organizational capabilities → this chapter gives the structure and test of solidification).
Whiteboard framework diagram ✓ (four quadrants of the dual capability model + five indicators of organizational strength, the main framework of the book).
External comparison and data ✓: Winner's side Duan Yongping's circle of influence (pedigree has been verified, load-bearing wall) + Musk (mechanism level, quality has been explained) + (circumstantial evidence within the boundary, quality has been explained); Loser side Wang An Computer (multiple sources have been verified: 3.07 billion peak, 424 million/715.9 million losses, complete timeline); P3 The control group has been balanced; the five-index scoring clearly "does not count the average score, look at the median and the lowest item"; Actor-Observer Bias (Jones & Nisbett, 1972) is the psychological anchor of the stance shift section. It is a textbook-level classic. Please check the original source before publication.
12 golden sentence candidates ✓ (The sentence "Cultural Spillover" belongs to Chapter 9 and has been removed; v1.2 adds two new sentences about stance ruler).
"What to Do Monday Morning" Two tests from the first position + personal notes ✓.