Quality Software Management: Vol 3

QSM의 3권. 관리자의 일치적인 행동에 대한 책이다. 리더십의 MOIJ 모델, Congruence 등의 주제에 대해 다룬다.

Part I. 나 자신을 관리하기

Chapter 1. 왜 일치성이 관리에 핵심적인가 (Why Congruence is Essential for Managing)

Summary

1. This book is about how to become the kind of high-quality, effective software engineering manager needed to produce high-quality software. First and foremost among the requirements for such a manager is the ability to act in congruence with your beliefs.
2. Cybernetics says managers can be seen as controllers of feedback systems. To manage an engineering system by feedback control, a manager as controller needs to
   • plan what should happen
   • observe what significant things are really happening
   • compare the observed with the planned
   • take actions needed to bring the actual closer to the planned
3. Effective managers must know what to do, but they must also be able to act in accordance with that knowledge.
4. Ashby's Law of Requisite Variety says the action taken by the controller must be congruent with the situation. When people are not tapping their full variety of potential actions, they are coping incongruently.
5. For control purposes, it doesn't matter why managers are unable to exhibit the requisite variety of action. Managers acting incongruently may not be capable of controlling the system they are trying to control. If so, the organization they manage will be unlikely to produce high-quality software, regardless of the reasons for their incongruence.
6. Technology is obviously important to the consistent delivery of high-quality software and software services, but in today's software organizations, management is the number one random process element. Not only that, but incongruent management stands in the way of improving all the other random process elements.
7. The personal effectiveness of people is what integrates all the other components of software engineering management. You'll never get a Pattern 3 organization with Pattern 2 managers. Instead, you start by getting the effective managers, then they lead the others.
8. The task of this volume is to address the major obstacles managers face in attempting to use full and appropriate variety in their actions.

Chapter 2. 관리를 선택하기 (Choosing Management)

Summary

1. Congruent managers are hard to find in software engineering, partly because of the way our organizations choose and develop their managers.
2. Boehm says, "Poor management can increase software costs more rapidly than any other factor." A cost driver of 64 would be a conservative estimate for management, because of the many ways poor management can increase costs—or even create total project failure.
3. Managers often seem to allocate improvement efforts in reverse order of cost-driver impact, putting least effort into the most important driver, management itself.
4. The One-Dimensional Selection Model for managers is based on three faulty assumptions, namely:
   • Managers are born, not made.
   • People can be ranked on a one-dimensional scale.
   • The scale for programming is the same as the scale for management.
   • As a result of this model, we tend to move the strongest technical people to management, a practice that weakens both management and the technical staff.
5. Team leaders can be effective at improving software quality, but the job of team leader is not the same as the job of manager. Nor will the best team leaders necessarily make the best managers.
6. People who don't want to be managers in the first place are starting their management careers in an incongruent position, one that will be difficult to improve over time.
7. People who go into management to pursue an unworthy vision will be unworthy managers. I hope this book won't help them.

Chapter 3. 대처의 유형들 (Styles of Coping)

Summary

1. Since everybody in an organization is responsible for controlling something, and since incongruent coping reduces the variety needed for effective control, it's possible to measure an organization's health through the people's characteristic coping styles.
2. When feelings of self-esteem are low, they are manifest in characteristic incongruent coping styles: blaming, placating, being superreasonable, loving/hating, or acting irrelevant.
3. In order to cope effectively with the world, we must be able to take into account three areas—self, other, and context—and balance their requirements all at the same time. To do this is to behave congruently.
4. When people fail to take other people into account, they fall into a blaming posture. When blaming, a person is saying, in effect, "I am everything, you are nothing."
5. When people forget to take themselves into account, they fall into a placating posture, and are effectively saying "I am nothing, you are everything."
6. A very common pattern is a blaming boss locked in a never-ending cycle with a placating employee.
7. Another common variation of the blaming-placating dynamic is the sudden switch in roles when the placater has swallowed enough abuse, and suddenly throws it all up onto the blamer.
8. In the superreasonable style of coping, people are entirely excluded from consideration. The superreasonable stance says, in effect, "It is everything; you and I are nothing."
9. From the viewpoint of congruence, love and hate relationships have the same structure—the total exclusion of context. The loving/hating stance says, in effect, "It is nothing; you and I are everything."
10. In the irrelevant style of coping, everything is missing, which leads to entirely unpredictable behavior. This behavior has a purely negative power—not to get things done, but to prevent things from getting done. In effect, the irrelevant behavior says, "Nothing counts for anything."
11. Incongruent coping styles each settle for less than they could get if they worked well. They do "work" to the extent they sometimes give some protection, so they are used when self-esteem is low.
12. Managers acting incongruently may get locked into their stance by a positive feedback loop connecting ineffectiveness and low-self esteem.

Chapter 4. 비일치성에서 일치성으로 (From Incongruence to Congruence)

Summary

1. Congruent behavior is not stereotyped behavior, but behavior that fits—the context, the others, and the self—so many congruent behaviors exist for any one situation.
2. You must experience the total interaction to know if it's congruent, which is not as hard as it sounds. When people experience the total interaction, they know when the message is not congruent.
3. Congruence is critically important to effective management. Managers must not only know how to recognize it, but also have confidence in their recognition. One way to gain such confidence is to watch for subtle incongruence between verbal and nonverbal response. Another is by listening to the content for certain characteristic patterns.
4. To become a higher-quality manager of software engineering, learn how the energy in incongruent coping can be transformed into something more useful.
5. Each incongruent behavior is based on a behavior actually effective in some survival situations, and may even have a genetic component. What makes it incongruent is its application when survival is not at stake.
6. Blaming is based on the survival behavior of attack. Blaming says, in effect, "If you continue to do that, I'm going to do what is necessary to stop you, no matter what the consequences to you." The impulse to blame can be transformed into an effective coping strategy—becoming assertive, direct, honest, frank, candid, forthright, or open.
7. The urge to placate can be congruently transformed into caring, yielding to the inevitable, giving in gracefully, or being a good loser. These behaviors are congruent when they take the future context into account.
8. The congruent transformation of the superreasonable style corresponds to the managerial behavior of staying cool, focused, and reasonable, especially in an emergency.
9. The congruent transformation of a loving posture is the ability to form beneficial alliances. The congruent transformation of a hating posture is the ability to participate in friendly rivalries.
10. Congruent behavior resembling irrelevance may be used as a desperate measure when everything rational has been tried. It may be diversionary, amusing, creative, or all three.

Chapter 5. 일치성을 향해 가기 (Moving Toward Congruence)

Summary

1. An important part of the task of new managers is learning to manage their own emotions, without becoming incongruent.
2. Beneath every incongruent coping, there is a survival rule, so-called because we respond as if our very survival were at stake. These rules function as unconscious programs controlling our behavior.
3. A reliable signal of our own incongruence is found in the internal messages we send ourselves, especially those messages wrapped in strong emotional packages.
4. By listening to your own internal messages, then transforming them in the light of high self-esteem, you can thwart incongruent actions before they happen.
5. Feelings are nature's way of telling you what's important and what's not so important. This information about you is not reliably available to other people unless you tell them.
6. To be congruent, you must set a "speak-up threshold" that measures your own feelings against those of others and the dictates of the context. Speaking congruently for yourself increases your chances of finding a solution you can live with.
7. The context in which you speak up is particularly important. Don't use third parties, hoping the word will get back. Don't claim to speak for third parties—if you have an issue, speak for yourself. Don't gossip about third parties. Raise any issue directly with the people involved. Don't issue general memos that secretly address individual cases. Such messages fool nobody—and make you look like a cowardly fool.
8. Always treat the other person with the honesty, dignity, and respect you would like for yourself. If you are too overwrought to control the tactics you use, then pull away until you regain control of yourself.
9. There is a step-by-step process you can use to recover your lost congruence. Roughly, you must first notice the incongruence, then make adjustments to your breathing, posture, and movement. Finally, you need to make contact with the other person, speaking in "I-statements," and wait for the other person to respond. You repeat this process as often as necessary either to get congruent or to discover it's not possible at the moment.
10. Congruence means feeling good enough to use the full variety of your action possibilities, so you have an excellent chance to be a professional software engineering manager.

Part II. 다른 사람을 관리하기 (Managing Others)

Chapter 6. The Manager's Job

6.7 What Congruent Managers Do (일치적인 관리자가 하는 것들)

Leadership is the ability to create an environment in which everyone is empowered to contribute creatively to solving the problems.

리더십은 문제를 해결하는데 있어 모든 사람이 창조적으로 기여할 수 있도록 환경을 조성하는 능력이다.

• 긍정적 강화를 제공해준다. offering positive reinforcement
• 정확하고 명확하게 지시한다. 그리고 지시받는 사람이 잘 이해하지 못할 때는 기꺼이 이해시켜준다. giving precise and clear instructions, and always being willing to clarify when they're not clear
• 꼭 필요한 것 이외에는 일하는 사람을 제약하지 않는다. not constraining workers any more than is essential
• 사람들이 가능성을 충분히 탐색하도록 해준다. letting people fully explore the possibilities
• 가능하면 작업을 단순화한다. 하지만 모욕적일만큼 너무 쉽게 하지는 않는다. simplifying tasks whenever possible, yet making sure the tasks aren't insultingly easy
• 시간 제약을 명확하게 하고, 시간제약의 이유를 설명해준다. making the time frames clear and giving the reasoning behind them
• 사람들의 스킬에 관심을 가진다. paying attention to people's skills
• 모든 직원들간의 업무부하를 고르게 조정한다. balancing the workload among all employees
• 모든 사람에게는 자신이 정말로 잘 할 수 있는 분야가 있다고 확신한다. ensuring there is some real part for everyone to play
• 스스로 직원들과 다른 사람들에게 지원적(supportive)이 됨으로서 어떻게 지원적이 될 수 있는지 가르친다 teaching how to be supportive by being supportive of employees and of each other
• 다른 사람과 고객들을 신뢰함으로써 어떻게 신뢰하는지 가르친다 teaching how to trust by trusting each other and the customers
• 직원이 된다는 것과 관리받는다는 것이 어떠한지 기억한다 remembering what it's like to be an employee and to be managed
• 신뢰를 구축하기 위해서 질문에 정확하고 정직하게 대답한다 answering questions correctly and honestly to build trust
• 좋은 컨설팅 조언을 받고 그것을 사용한다 getting good consulting advice and using it
• 문제에 대한 비전을 만들고 그것을 모두에게 명확하게 커뮤니케이션한다 creating a vision of the problem and communicating it clearly to everyone
• 직원이 필요로 할 때는 언제든 조직 차원의 가이드를 제공한다 providing organizational guidance whenever employees need it
• 사람들이 초기부터 성공을 경험할 수 있도록 환경을 조성한다 setting things up so people can experience early success
• 사람들에게, 그들이 할 수 없거나 하고 싶지 않은 일들을 요구하지 않는다 not asking people to do things they aren't able or willing to do
• 즐겨도 괜찮은 환경을 만든다 creating an environment in which it's okay to have fun

Summary

1. In an effective software organization, the manager's job is getting more people involved (and getting people more involved) in decisions about what is to be done, and in doing it.
2. For example, to recover from a crisis, a manager has to mobilize sidelined people in order to have the resources to keep the ship afloat.
3. As an organization moves deeper into crisis, the best-informed people tend to become overloaded. Managers may unknowingly contribute to this piling on, and can counteract it by congruent management action.
4. In Pattern 2 (Routine) organizations, managers prescribe the way things should be done, rather than describe what outcomes are desired (a more Pattern 3 behavior). Managers who prescribe tend to listen to employees in a superreasonable or blaming way, without hearing them.
5. A major job of the software engineering manager is to develop openness and trust among all the workers who are contributing to successful software activities. This requires honesty and true listening.
6. Right or wrong, the reasons given by employees always contain the information an effective manager needs to steer the organization.
7. The Pattern 3 (Steering) manager does not generally evaluate quality, but establishes the processes, not people, to evaluate quality. But when the manager does evaluate the quality of work, the main purpose of the evaluation is to help the employees develop their own skills.
8. Managers who believe the One-Dimensional Selection Model have no room in their repertoire for coaching, teaching, or training.
9. Congruent managers are not locked into blaming, partly because they know they are not passive victims of their employees, their bosses, or the dynamics of software quality. Instead, they use all of these factors intelligently as resources.
10. Leadership is the ability to create an environment in which everyone is empowered to contribute creatively to solving the problems. In this model, the manager's job may be evaluated by one and only one measure: the success of the people being managed.

Chapter 7. 선호 차이 (Preference Differences)

Summary

1. Poor managers tend to deal with variations among people by pretending they don't exist. Effective managers recognize differences and know how to deal with them congruently.
2. Treating everyone the same does not mean treating everyone equally. The astute manager notices the differences among people and knows how to use them to manage effectively—equally, but not unfairly.
3. Although we tend to think of all software work as logical, many actions are chosen on the basis of emotion. One of the mildest yet most important emotions is preference.
4. When an unconscious preference reduces choices, we have a tougher time with Ashby's Law of Requisite Variety. Just because there is a preference, however, such a loss of variety need not occur—because we need not act on our preferences.
5. The four dimensions of the Myers-Briggs Type Indicator (MBTI) are significant in the workplace because they describe how four elements determine much of a person's working style.
6. For each dimension of the MBTI, there is a pair of letters to choose from:
   • I or E according to how I prefer to get energy
   • S or N according to how I prefer to obtain information
   • T or F according to how I prefer to make decisions
   • J or P according to how I prefer to take action
7. Failure to take the I/E difference into account leads to underperformance by one group or the other. One of the manager's jobs is to design meetings to accommodate the environmental preferences of both Internals and Externals.
8. Sensors want the facts, lots of facts, while Intuitives want the big picture. A manager who has communication problems with employees should explore this difference as a prime candidate for the root cause.
9. Thinkers and Feelers are often intolerant of each others' preferred style. In an organization, you can see the T/F preference in action whenever decisions are to be made. Both types want good decisions, but they differ in what attributes make a decision good. Many T/F problems can be solved by designing the correct environment for decision making.
10. The Judging (J) preference is to have things settled, while the Perceiving (P) preference is to keep options open on the chance more information will affect the choice. Judging/Perceiving differences are often the source of great conflict, as well as the source of great attraction, because each needs the other.

Chapter 8. 기질 차이 (Temperament Differences)

Summary

1. The combination of MBTI dimensions produces 16 identifiable personality types. Sub-combinations produce other useful views of personality, such as the four temperaments of Kiersey and Bates.
2. We can conveniently identify four different kinds of control: intellectual, physical, emergency control, and emotional, which we can relate to the four temperaments.
3. Most of the major innovations in software have been in the area of improving intellectual control; at least, they were the innovations receiving the most attention.
4. Physical control is introduced into software to deal with real world deviations from the pure intellect model of software, either by prevention or detection and correction.
5. As we develop routines to handle intellectual and physical problems, we find our ability to manage well depends not on our ability to handle routine situations, but on our ability to handle exceptional situations.
6. Sometimes, the situation is physically and intellectually simple with no emergency conditions, yet people still fail to perform perfectly. As long as people are involved, the other three types of control become meaningless without emotional control.
7. Strong emotions produce stereotyped behavior, which means behavioral variety is reduced. Anything that reduces variety reduces the manager's ability to control.
8. The Visionary (NT) likes working with ideas. The Catalyst (NF) likes working with people to help them grow, but is concerned that people not suffer. The Organizer (SJ) likes order and system. The Trouble-shooter (SP) likes getting the job done.
9. Because each software cultural pattern favors different kinds of control, each temperament reacts differently to each pattern. Each temperament if unchecked will contribute to overruns in characteristic ways.
10. People of all temperaments react to errors, but they tend to react in different ways. Managers can use these reactions to motivate people to prevent errors.
11. When observing, the SJs and SPs easily take the "self" position; the NFs easily take the "other" Position; and the NTs favor the "context" Position.
12. In interactions with other people, the NTs tend to bypass the Intake step and go instantly to meaning, while the NFs tend to jump immediately to significance. The SJs tend to stay in Intake too long, gathering too many facts; while the SPs actually use the whole process rather well, but tend to go so fast it looks to others as if they leap instantly to Response.

Chapter 9. 차이를 자산으로 여기기 (Differences as Assets)

Summary

1. Any difference can become important when it is not understood, accepted, or handled well.
2. The software business is unlikely to be controlled effectively by one manager telling a whole lot of technical people exactly what to do. Control must reside within every person involved, meaning a great many individual differences in the way things are handled.
3. One lesson of the MBTI studies is there is generally no perfect answer to the question "Which type is right for this job?" Generally these studies show people of all types can do almost any job, although there are preferences for certain types in certain kinds of work.
4. Because software work consists of so many varied tasks, it's unlikely any one personality type, one set of skills, or one point of view would be best suited to all parts of the total software job. That's why we need differences among software people.
5. The Management by Selection approach is based on the One-Dimensional Selection Model applied to the technical staff. This approach identifies the "bad" programmers (or any other technical position), gets rid of the worst ones, and repeats this process so the average ability rises.
6. This approach is slow to show improvement, and produces only mediocre results. The approach requires a continuing supply of new programmers, and may damage your organization's reputation among job applicants. Your best performers may leave because this method doesn't really focus on their performance.
7. The systematic improvement approach is based on multi-dimensional thinking. The model is applied by identifying the "good" programmers (or any other technical position), analyzing the performance of the best to determine why they are doing so well, and developing systems (training, technical reviews, teams, mentoring, and modeling) for passing these best processes on to large numbers of people.
8. The systematic improvement approach says attention to process will increase awareness of what's effective; training increases penetration of existing effective processes; and identifying effective processes leads to abandoning the ineffective.
9. Organizations importing technical and managerial staff from other cultures add greatly to the richness of their working environment. This cultural richness, however, doesn't always seem like an advantage to the managers, who seldom have the training or experience to deal with it.
10. For whatever reason, men and women of most cultures operate—on average—differently in some areas. Managers who favor "male" or "female" values may find themselves cut off from half the best information and ideas available to them.
11. People differ greatly in what neurolinguistic programming calls their strategies. These are the programs by which people order the ways they take in information—their internal and external pictures, sounds, smells, tastes, and feelings—when they solve a problem. A toolkit of different strategies can be great assets in many difficult aspects of software work.
12. Most software people believe it's okay to treat people differently if they have different abilities, but they don't know how to recognize ability. Most don't believe it's okay to treat people differently if they are differently advantaged, but fail to use what they could learn from them. Finally, discrimination based on age (either favoring young or old) is so universal in software we seldom notice it, let alone take advantage of what learnings age differences can provide.

Chapter 10. 비일치성의 패턴들 (Patterns of Incongruence)

Summary

1. In many organizations, managers lack time because they lack the ability to deal congruently with incongruence in others. Different cultural patterns have their characteristic patterns of incongruence, so each pattern puts a different load on its managers.
2. Ineffective actions leave problems unresolved, so one effect of incongruence is an increase in the number of problems someone other than the originator must deal with. Thus, more incongruence means more problem-solving or fire-fighting time.
3. A major task of the software manager is to help people in the organization develop their social skills, not just because it's better when people are nice, but because social skills are becoming more and more important as a basis for technical success.
4. Organizations tend to lock on to either a pattern of incongruent behavior or a pattern of congruence. What makes the difference is the decisions of the managers—decisions which influence their behavior and become an example for others.
5. Feeling unable to trade-off either quality or schedule, software managers are too often tempted to sacrifice the quality of human interaction. If they yield to that temptation, they soon pay the price in both quality and schedule deficiencies.
6. The placating pattern is especially common in Pattern 1 cultures, which largely explains why they are called Variable cultures. To take one example, a project having no review system finds faults at the latest, most costly stage of the project, with maximum impact on schedule. Managers then use this delay in schedule to justify the omission of further reviews and tests.
7. In blaming cultures, especially in Pattern 2 (Routine) organizations, yielding to the temptation to punish the offenders for failures soon leads to extra effort to escape the abuse. On the other hand, the workers increasingly resent the abusive manager and find more ways to be unresponsive or even to sabotage the manager, which then leads to more failures.
8. The addiction cycle works like this: There is a short cycle in which doing X relieves symptoms. There is a longer cycle in which doing X makes those same symptoms worse. The short cycle of addiction contains the strength of belief in X, so the strength of belief in X grows the more X is used, although in the long cycle, X leads to a worse situation.
9. Addiction is, at bottom, the belief there is only one way to do something and it must be done that way.

Chapter 11. 인간 행동에 대한 기술 (The Technology of Human Behavior)

Summary

1. In order to work with people on a practical level, the working manager needs a model usable as a day-to-day guide to achieving more congruent management.
2. In the Satir interaction model, Response is not actually limited to the last step in the model. During Intake, for instance, I take in some data, then respond by deciding whether to open up for more, reduce the data by filtering, or proceed with making meaning of what I already have.
3. The Meaning step contains a response as well. After breaking the data down into major categories, I open, close, or modify the Intake process to fit the perceived meaning according to my personal style.
4. The significance of each possible meaning can be considered in terms of the possible consequences to me, and the possibility chosen determines the general pattern of my response.
5. The human brain seems to operate not as one mind, but as a team of minds. If I am coping well, I have many different minds to put in charge of different situations, as Ashby's Law of Requisite Variety says I must do if am to be an effective controller of complex systems.
6. The many members of my internal "team" can be thought of as distinct personalities, and (half seriously) identified with well-known characters. All characters, good and bad, are parts of the whole me, and if you work with me long enough, you'll eventually see most of the cast.
7. Underlying the Satir interaction model, we can imagine a tree whose branches lead to the various meanings I might make in a given situation. The root of the tree is my self. The main trunk is formed by my yearnings, each of which branches into one or more expectations.
8. Expectations are the translation of universal yearnings into specific ideas about how the world works to produce or withhold what is yearned for. Expectations are often held in the form of rules that express the way I expect the world to work. Even though a rule may not apply to the present situation, I do not easily relinquish it, because it contains valuable information from my earliest experiences about how to survive in order to achieve my deepest yearnings.
9. Standing in the way of your recognizing my intent in an interaction is my personal style, the surface formed by my coping posture, preferences, learnings, habits, addictions, and culture. To reach my intent, you need to use a model of human behavior to penetrate that style and find what lies beneath it.
10. The technology of human behavior is many times more complex than the technology of software. When I am able to see through this complexity to deeper levels, I am able to achieve a more complete rapport, and to manage people (myself and others) with minimal energy and disruption. This is what I mean by becoming a skilled technologist of human behavior.

# QSM Vol 3-1. Managing Yourself and Others

## Part I. Managing Yourself

### Chapter 1. Why Congruence is Essential for Managing

왜 (굳이) 행동과 생각이 일치(congruent)해야 하는가?

아는 것 vs 행동하는 것.

이전 볼륨에서, 엔지니어링 시스템을 피드백 제어를 통해 관리하기 위해서, 컨트롤러로서 매니저는 아래와 같은 것들을 해야 한다고 말했다:

* 무엇이 일어나야 하는지 계획하고
* 어떤 중요한 일들이 실제로 일어나는지 관찰하고
* 계획된 것과 관찰된 것을 비교하고
* 실제를 계획에 좀 더 가까워지도록 행동을 취하고

이 새 QSM 시리즈의 첫 두 권은 계획(planning)에 중점을 둔다. 3,4권은 관찰(ovserving)과 비교(comparing)에 중점을 둔다. 이 권(5,6)은 행동(action)에 중점을 둔다.

행동에 중심을 두는게 좀 이상해보일지 모른다. 어떤 사람들은 일단 계획이 세워지면, 행동은 으례 따라오는 것이라고 생각한다. 하지만 나는 관리자들이 무엇을 해야 하는지 꽤 잘 알지만 뭔가 그것을 행동하는데 부족한 상황을 많이 목격했다. 이를테면,

* 어떤 관리자는 프로젝트를 제 시간에 끝내는 것은 전혀 가망 없는 일이라는 것을 알았지만, 그녀의 상사에게 대안에 대해 이야기하기 위해 논의를 시작할 수 없었다.
* 어떤 관리자는 지연된 프로젝트에 개발자를 더 투입하는 것이 일을 느리게 만들 뿐이라는 것을 알았지만, 아무 것도 하지 않는 것처럼 보이는 위험을 감수할 수 없었다.
* 어떤 관리자는 사람들에게 소리지르는 것이 상황을 나쁘게 만든다는걸 알았지만, 그들에게 소리지르는 것을 멈출 수가 없었다.
* 어떤 관리자는 한 직원이 냄새가 너무 많이 나서 다른 사람이 그와 함께 일하지 못할 지경이라는 것을 알았지만, 그 직원에게 그에 대해 이야기할 수가 없었다.
* 어떤 관리자는 어느 직원이 임무에 적격이라는 것을 알았지만, 다른 사람을 선택했는데, 그가 최선의 후보자를 싫어했기 때문이었다.
* 어떤 관리자는 그들이 해결하려는 문제를 명확하게 이해하지 않고서는 프로젝트가 진행되지 않을 것이라는걸 알았지만, 개발자들이 코드를 작성하기 시작하고자 하는 열망에 저항할 수가 없었다.
* 어떤 관리자는 그녀가 지키지 못할 약속은 하지 말아야 한다는 것을 알았지만, 인간관계 상황의 어려움을 타개하기 위해서 계속 약속을 해나갔다.