Series ID: EDKS-BASIC-ZH-0099 · Lesson No.099 · C2-Direction Decision Writing
Writing for Policy, Decisions and Recommendations | 政策、决策与建议写作
A recommendation is not just an opinion with formal language. It is a decision claim tied to criteria, evidence, constraints, alternatives and consequences.
At advanced levels, policy and decision writing must make the reasoning inspectable. Readers should be able to see what problem is being solved, which criteria matter, what evidence supports each option, what uncertainties remain and what conditions would cause the decision to change.
高级建议写作的核心，是把“为什么这样决定”写得可以检查：问题是什么？标准是什么？有什么替代方案？证据多强？风险在哪里？什么时候应该重新评估？
Chinese Edition Hub · 中文版入口 · ← Lesson No.098 · Negotiation, Consensus and Productive Disagreement
The decision-writing map | 决策写作地图
- problem definition
- objective
- criteria
- constraints
- options
- evidence
- risk
- recommendation
- implementation
- review trigger
1. Define the decision | 先定义要决定什么
Weak:
We need a better system.
Stronger:
We need to decide whether to replace the current system before the next academic year.
2. Define the problem separately | 问题不等于方案
Do not write the preferred solution into the problem statement.
3. Objective | 目标
State what success should achieve.
- reduce error
- reduce cost
- increase access
- improve reliability
4. Criteria | 判据
Which dimensions will options be judged by?
5. Constraints | 约束
What cannot be violated?
- budget ceiling
- deadline
- legal rule
- safety standard
6. Options | 备选方案
Always include the status quo where appropriate.
7. Do-nothing option | 不改变也是一个选项
Sometimes the current system remains preferable under uncertainty.
8. Option generation | 生成方案
Do not compare only your preferred option with an obviously weak alternative.
9. Decision matrix | 决策矩阵
| Criterion | Weight | A | B | Status quo |
|---|---|---|---|---|
| reliability | 40% | 8 | 9 | 6 |
| cost | 30% | 7 | 5 | 9 |
| speed | 30% | 9 | 8 | 5 |
10. Matrix does not make values disappear | 矩阵不会消除价值判断
Weights themselves need justification.
11. Evidence per option | 每个方案都要有证据
Do not compare one option with detailed evidence against another described vaguely.
12. Evidence quality | 证据质量
Note:
- source
- method
- scope
- uncertainty
13. Risk | 风险
Risk writing should distinguish:
- likelihood
- impact
- detectability
- recoverability
14. Risk ≠ possibility only | 风险不只是“可能发生”
A rare catastrophic event and a common minor event require different treatment.
15. Risk register | 风险表
| Risk | Likelihood | Impact | Mitigation |
|---|---|---|---|
| implementation delay | medium | high | pilot first |
16. Uncertainty | 不确定性
Distinguish:
- known risk
- measurement uncertainty
- unknown information
- model uncertainty
17. Decision under uncertainty | 不确定下也要决策
You do not need certainty to decide. You need a rule proportionate to the stakes.
18. Reversible vs irreversible decisions | 可逆 vs 不可逆
Reversible decisions can tolerate more uncertainty than irreversible high-cost decisions.
19. Pilot | 试点
A pilot is useful when the decision can be staged and uncertainty can be reduced.
20. Pilot is not permanent delay | 试点不能变成无限拖延
Define:
- duration
- success criteria
- stop criteria
- decision date
21. Recommendation | 建议
Recommendation should answer:
What should be done now?
22. Recommendation strength | 建议强度
- proceed
- proceed conditionally
- pilot
- defer
- reject
- maintain status quo
23. Match strength to evidence and stakes | 强度匹配证据和风险
Weak evidence + high stakes often supports pilot/defer rather than full commitment.
24. Conditions | 条件
Proceed only if…
- budget stays below X
- error rate remains below Y
- staffing requirement is met
25. Dependencies | 依赖项
A recommendation may depend on another event or capability.
26. Implementation | 实施
State:
- owner
- sequence
- resources
- timeline
- monitoring
27. Implementation owner | 责任人
Passive language can hide responsibility.
Weak:
Monitoring should be conducted.
Better:
The operations team should monitor error rate weekly.
28. Decision owner | 决策权
Separate:
- who recommends
- who decides
- who implements
29. Review trigger | 复审触发器
State what evidence would cause the decision to be reconsidered.
30. Example review trigger | 例子
Review the rollout if error rate exceeds 3% for two consecutive weeks.
31. Stop rule | 停止规则
Define conditions under which implementation should stop.
32. Escalation rule | 升级规则
Define when an issue must move to a higher decision level.
33. Sunset clause | 失效/到期机制
A temporary policy can expire automatically unless renewed.
34. Policy language | 政策语言
Policy may need:
- scope
- definitions
- requirements
- exceptions
- responsibility
- review
35. Requirements vs recommendations | 要求 vs 建议
- must
- should
- may
Choose deliberately.
36. Must | must
Use for mandatory requirements where authority supports it.
37. Should | should
Use for strong recommendation or expected practice.
38. May | may
Use for permission or possibility depending context.
39. Ambiguous policy modals | 模糊情态词
High-stakes policy writing should avoid ambiguity over whether something is required.
40. Exceptions | 例外
Specify:
- who qualifies
- who approves
- how documented
41. Avoid exception loopholes | 避免例外漏洞
Vague exceptions can undermine the standard.
42. Recommendation rationale | 建议理由
Structure:
- decision
- criteria
- evidence
- trade-offs
- risk
- conditions
43. Executive decision memo | 行政决策备忘
Possible headings:
- Decision Required
- Recommendation
- Evidence
- Risks
- Options
- Next Steps
44. Public policy explanation | 公众政策解释
Explain:
- what changes
- why
- who is affected
- when
- where to get help
45. Technical decision appendix | 技术附录
Preserve detailed methods and uncertainty outside the executive summary.
46. Recommendation ≠ certainty claim | 建议不等于确定性
You can recommend action even while acknowledging uncertainty.
47. Explain risk appetite | 说明风险偏好
Different organisations may reasonably choose differently under the same evidence.
48. Values and priorities | 价值与优先级
State when recommendation depends on a normative priority.
49. Fairness vs efficiency | 公平 vs 效率
If the trade-off matters, name it explicitly.
50. Short-term vs long-term | 短期 vs 长期
Recommendations should distinguish immediate gain from long-term consequence.
51. Distributional effects | 分配影响
Who benefits? Who bears cost?
52. Average effect can hide subgroups | 平均值会隐藏不同群体
Check whether a policy helps some and harms others.
53. Implementation feasibility | 实施可行性
A theoretically good recommendation can fail if staff, time or infrastructure are unavailable.
54. Operational detail | 操作细节
Enough detail to act, not so much that the recommendation becomes unreadable.
55. Decision log | 决策记录
Record:
- decision
- date
- evidence
- assumptions
- owner
- review trigger
56. Why decision logs matter | 为什么要记录
Future reviewers can distinguish bad process from bad luck.
57. Outcome ≠ decision quality | 结果不等于决策质量
A good decision under uncertainty can still produce a bad outcome.
58. Evaluate process too | 也要评估过程
Ask whether the decision used the best available evidence and reasonable criteria at the time.
59. Mandarin transfer: “建议”可对应 suggest/recommend/propose/advice | 功能不同
Choose the English form based on whether you are making a formal recommendation, informal suggestion or policy proposal.
60. Mandarin transfer: 政策英语需要明确 must/should/may | 不要模糊
Modal force matters operationally.
61. Practice scenario A | 场景 A
A school is considering a new reading platform. Evidence shows higher engagement but no clear score improvement.
Write a recommendation.
62. Practice scenario B | 场景 B
A company pilot reduces response time but has uncertain error effects.
Choose: scale, pilot longer, or stop.
63. Practice scenario C | 场景 C
A policy helps the average user but disadvantages one subgroup.
Address distributional effects.
64. Error clinic | 常见问题
| Problem | Repair |
|---|---|
| Recommendation appears first with no criteria. | Show decision logic. |
| Only preferred option is analysed. | Compare credible alternatives. |
| Risk listed but no mitigation. | Add response. |
| Pilot has no success rule. | Define threshold/date. |
| Policy modals ambiguous. | Clarify must/should/may. |
65. First weak link diagnosis | 第一个卡点诊断
- problem vague → decision definition.
- recommendation feels subjective → criteria/evidence.
- uncertainty blocks action → staged decision design.
- implementation fails → feasibility/ownership.
- policy misunderstood → modal/scope clarity.
66. Seven-day training cycle | 七天训练
| Day 1 | problem/objective | 问题目标 |
| Day 2 | criteria/options | 判据方案 |
| Day 3 | evidence/risk | 证据风险 |
| Day 4 | recommendation | 建议 |
| Day 5 | implementation | 实施 |
| Day 6 | review/stop rules | 复审停止 |
| Day 7 | full decision memo | 综合 |
67. Self-test | 自测
Define a decision, establish criteria and constraints, compare credible options, weigh evidence and risk, recommend proportionately, specify implementation, and write review/stop triggers.
68. For parents and teachers | 给家长和老师
Ask advanced learners to write the rejected options fairly, not only the winner.
Require review triggers so recommendations remain revisable.
For Mandarin speakers, practise precise modal force in policy and decision English.
69. Final real-world challenge | 最终真实任务
- Choose a real decision.
- Define problem and objective.
- List three options including status quo.
- Set criteria and weights.
- Build evidence/risk table.
- Write a recommendation.
- Set implementation owner.
- Set one review trigger.
- Set one stop rule.
- Write a 500-word decision memo.
Next: Lesson No.100 | 下一课
The next lesson performs the full-course mastery review: from the first beginner sentence to independent expert English, using one system of diagnosis, transfer, evidence, audience control and lifelong maintenance.
Lesson No.100 · Mastery Review: From Beginner to Independent Expert English · 总复习：从零基础到独立专家英语
Reference floor: C2-direction decision writing. Recommendations are explicitly tied to evidence, criteria, constraints and review conditions.
