Writing as a thinking tool
471 points
• 4 days ago
• Article
Link
写作是一种重要的外化过程，能够提升人类认知。它的作用非常类似于计算机科学中的 out-of-core execution algorithm 。借鉴 Manuel Blum 的研究，有人提出：没有写作的人类只能停留在有限自动机的层面，而用笔和纸则提供了实现图灵机复杂性的外部记忆。通过把想法卸载到外部媒介，我们超越内部的局限，得以处理并打磨那些原本可能未成形、或转瞬即逝的念头。这种做法不仅仅是为了留档，更是思考与深度理解不可或缺的工具。
大型语言模型（Large Language Models）的兴起，给这一认知过程带来了独特挑战。虽然 AI 能综合既有知识与共识，但它缺少个人特有的、尚处萌芽状态的直觉，也缺乏进行创造性、带来新突破的能力。让 AI 代写文本，往往会制造出"严谨思考"的错觉——比如在专业备忘录中——却不需要真正进行智力加工。由于写作是锚定理解的主要机制，把这项工作交给机器，最终会导致无法真正把问题想透。因此，在自动化文本生成的时代，个人写作逐渐演变为一种显著的比较优势。
有效的写作需要结构化的方法：把重点放在内在的发现过程上，而不是追求润色或迎合外部评价。第一条规则，是允许不完美：在不受拘束的前提下先大量起草（bulk drafts），让那些尚脆弱的想法在尚未成熟之前免于过早被否定，直到它们拥有成长的空间。想法被记录下来之后，第二阶段就要进行毫不留情的修改与打磨——用对待代码同样的批判性疏离感来审视自己的文字。这种起草与编辑的迭代，正是洞察不断加深的关键所在：写作因此成为抽象与复杂问题解决的载体。
除了起草技巧本身，写作还依赖简单、直接和稳定的节奏。坚持朴素而简洁的表达，会迫使内容变得更清晰；而冗长绕弯往往只是对定义不清的想法的遮掩。此外，写作应由内在好奇心驱动，而不是被外界认可那种难以预测的反馈牵着走——把作品当作与自己的对话来经营。与其等待灵感出现，不如设定固定的时间安排，这样就能持续保持"察觉—反思"的状态，确保大脑始终是一个主动的处理者，而不是停滞不前。
Writing serves as a vital externalization process that elevates human cognition, functioning much like an out-of-core execution algorithm in computer science. Drawing on the work of Manuel Blum, it is suggested that while humans without writing are limited to finite automata, the use of pen and paper provides the external memory necessary to achieve the complexity of a Turing machine. By offloading thoughts onto an external medium, we move beyond internal limitations, allowing for the processing and refinement of ideas that otherwise remain unformed or fleeting. This practice is not merely for documentation, but serves as an essential tool for thinking and deep understanding.
The rise of Large Language Models has introduced a unique challenge to this cognitive process. While AI can synthesize existing knowledge and consensus, it lacks an individual's specific, nascent intuitions and the capacity for creative, novel leaps. Relying on AI to generate text can create the illusion of rigorous thought, such as in professional memos, without the actual labor of intellectual processing. Because writing functions as the primary mechanism for anchoring understanding, offloading this task to a machine results in a failure to genuinely think through a problem. Consequently, personal writing has evolved into a distinct comparative advantage in an age of automated text generation.
Effective writing requires a structured approach that prioritizes the internal act of discovery over polish or external reception. The first rule is to embrace imperfection by producing bulk drafts without restraint, protecting fragile ideas from premature judgment until they have room to grow. Once the ideas are captured, the second phase involves ruthless editing and refining, treating one's own prose with the same critical detachment applied to code. This iterative process of drafting and editing is precisely what allows for the deepening of insights, effectively turning writing into a vehicle for abstraction and complex problem-solving.
Beyond the mechanics of drafting, writing thrives on simplicity, directness, and consistent scheduling. Adhering to a style that is plain and succinct forces clarity, as convoluted prose often serves as a mask for poorly defined thoughts. Furthermore, writing should be motivated by intrinsic curiosity rather than the unpredictable nature of external validation, treating the work as a conversation with oneself. By setting a regular schedule rather than waiting for inspiration, one maintains a state of constant noticing and reflection, ensuring that the brain remains an active, processing agent rather than becoming stagnant.
115 comments • Comments Link
- 写作是一种重要的认知工具，而不是对已完成思想的简单记录。它能帮助个人阐明复杂想法，识别知识差距，并搭建其内部逻辑。
- 过度依赖 AI 来生成写作或代码，可能会绕过真正理解主题所必需的关键挣扎，进而导致输出肤浅，且作者自身的思维过程逐渐萎缩。
- 有一种相对可靠的启发式判断：AI 生成的内容往往表明作者缺乏足够的深度或投入，暗示所呈现的想法可能尚未被充分审查或完善。
- 手动编码和写作迫使人们深入接触需求与约束，充当一种"思维工具"，把在被动开发中容易被掩盖的复杂性显露出来。
- 一些从业者采用混合方法：把 AI 当作"橡皮鸭"用来交流想法或生成草稿，然后再手动细化，或重新输入逻辑，以确保自己真正内化解决方案。
- 像笔和纸这样的物理、自由形式媒介，为草图绘制和图表制作提供了不受约束的非线性自由，这是数字工具往往难以复制的，因此更适合头脑风暴和概念映射。
- 一个有效的笔记系统通常会把模拟与数字流程结合起来，例如先用带颜色编码的手绘系统图表，再通过 OCR 或数字工具整理并进行长期保存。
- 与 LLM 的互动可以发挥知识伙伴的作用，但前提是用户保持严格监督：要求它解释，进行对抗性审查，并通过测试与个人调查来验证输出。
- 图灵机的类比强调写作把记忆外部化为"磁带"，让个人能够记录、重读、评估并重写，从而克服仅限于内部思维的认知局限。
- 为了个人清晰而进行的写作是一项无需读者也值得投入的实践，因为其主要收益——更敏锐的理解与更好的想法生成——来自创作者在创作过程中完成的工作。
这种集体观点指出，写作和编码是基础性的认知过程，而不仅仅是输出手段。参与者认为，手动工作的阻力并非障碍，而是学习、解决问题以及建立真正专业知识所必需的一部分。尽管普遍承认 AI 能作为强有力的助手或对话工具，但大家高度一致：把核心"思考"交给代理会带来收益递减，最终结果更为肤浅。最终，讨论表明，真正深刻的智力成长需要对自身工作进行主动、有意的投入，无论是通过传统手写笔记，还是把软件开发视为人与机器之间严格且交互式的对话。 • Writing serves as an essential cognitive tool rather than a mere record of completed thoughts, helping individuals clarify complex ideas, identify knowledge gaps, and structure their internal logic.
• Relying on AI to generate writing or code risks bypassing the critical struggle required to truly understand a topic, leading to shallow output and an atrophy of the author's own thinking process.
• A reliable heuristic exists where AI-generated content often signals a lack of depth or commitment by the author, suggesting that the ideas presented have not been adequately vetted or refined.
• Manual coding and writing force a deep engagement with requirements and constraints, acting as a "thinking tool" that reveals complexities which might otherwise remain obscured during passive development.
• Some practitioners employ a hybrid approach, using AI as a "rubber duck" to bounce ideas off or generate rough drafts, then manually refining or re-typing the logic to ensure they personally internalize the solution.
• Physical, free-form mediums like pen and paper offer unconstrained, nonlinear freedom for sketching and diagramming that digital tools often struggle to replicate, fostering a superior environment for brainstorming and conceptual mapping.
• Effective note-taking systems often involve a blend of analog and digital workflows, such as hand-drawing system diagrams with color-coded notations before using OCR or digital tools for organization and permanence.
• Engaging with an LLM can function as a productive intellectual partner, provided the user maintains rigorous oversight by asking for explanations, performing adversarial reviews, and verifying output through testing and personal investigation.
• The analogy of a Turing Machine highlights that writing acts as an externalized "tape" of memory, allowing individuals to record, re-read, evaluate, and rewrite, thereby overcoming the cognitive limitations of a finite, internal-only thought process.
• Writing for personal clarity is a valuable practice that does not require an audience, as the primary benefit—sharpened understanding and idea generation—is realized by the creator during the act of composition.
The collective perspective emphasizes that writing and coding are foundational cognitive processes rather than just output methods. Participants argue that the friction of manual work is not a hindrance but a necessary component of learning, problem-solving, and building genuine expertise. While there is broad acknowledgment that AI can serve as a powerful assistant or a tool for dialogue, there is a strong consensus that delegating the core "thinking" to an agent leads to diminished returns and superficial results. Ultimately, the discussion suggests that deep intellectual growth requires an active, intentional engagement with one's work, whether through traditional longhand notes or by treating software development as a rigorous, interactive dialogue with the machine.