学习编程依然值得。
Learning to code is still worthwhile

原始链接: https://stevekrouse.com/learn-to-code

在人工智能与“氛围编程”(vibe coding)的时代,有人认为学习编程已不再必要。然而,Val Town 的创始人史蒂夫·克劳斯(Steve Krouse)主张,编程仍然是一项至关重要的追求——这不仅是为了职业发展,更是作为一种基础的教育工具。 克劳斯借鉴了西摩·帕普特(Seymour Papert)的“数学王国”(Mathland)愿景,认为编程是通往逻辑思维、调试能力和构建能力的强大门户。除了这些元技能之外,编程还培养了一种独特的能动性,使个人能够通过精确的语言将抽象的想象转化为现实——这就像是在“施展咒语”。 克劳斯将代码与其他博雅教育学科相提并论,并指出,尽管大语言模型(LLM)可以生成语法,但这并没有削弱理解支撑我们世界的形式语言的价值。归根结底,编程是一项具有创造性、令人愉悦且赋予力量的活动。无论是作为理解数学的视角,还是作为构建软件的工具,“真正的计算机革命”——即普及编程素养——依然是一个值得追求的目标,它提供了一种变革性的方式,让我们能够参与并塑造现代世界。

关于“学习编程是否仍有价值”的 Hacker News 讨论显示,人们在将编程视为一种**基础技能**还是**商品化劳动**之间存在巨大的分歧。 **支持编程仍具重要性的观点:** * **深刻的理解力:** 许多专家认为,如果不具备编程技能,就无法评估 AI 的输出、调试复杂系统或设计可扩展的架构。编程被视为一种形式化逻辑的“思维方式”,就像尽管有计算器,数学家仍需掌握基础数学一样。 * **“架构师”角色:** 支持者认为,随着 AI 处理“样板式”任务,人类程序员将转型为“指挥官”或架构师。这一层面的工作仍需要高水平的技术专长,以有效地引导 AI 智能体。 * **内在价值:** 有人认为,编程就像文学或音乐一样,是一种创造性的手艺。即便 AI 能生成平庸的代码,人类仍会珍视构建事物的过程,而不仅是其市场价值。 **认为编程重要性下降的观点:** * **市场现实:** 持怀疑态度者认为,专业编程正日益成为一种小众的“匠人”行当。他们认为对于大多数商业应用而言,“足够好”的 AI 生成代码将使手动编程对初学者来说在经济上不再划算。 * **抽象的局限性:** 许多人担心,过度依赖大语言模型会导致一代“只靠感觉写代码”的开发者出现;当抽象层失效时,他们缺乏修复系统的基础知识,从而导致未来出现难以维护且充满漏洞的软件。
相关文章

原文
Learning to code is still worthwhile

July 6, 2026

Is there any point to learn to code in the age of vibe coding?

Literally no one is uttering what we thought were the immortal lines “learn to code,” right? I don't think that sentence has been spoken for many months now in Silicon Valley

Making Sense with Sam Harris #481

I'm the founder of Val Town, a "Silicon Valley startup" for writing and deploying code, and I still believe everyone should learn to code.

Sam is right that “learn to code” is no longer trotted out as a quick path out of poverty. The ability to string together two lines of JavaScript no longer guarantees you a 6-figure-salary.

This is also true of math, literature, science, or any of the liberal arts. Like those, coding is worthwhile to learn on educational grounds, not merely vocational ones.

I grew up hating math. I stumbled into an after-school program that taught programming. Through it, I fell in love with math, and excelled at it beyond my wildest dreams.

Later, I learned this experience masterminded by a math & education researcher named Seymour Papert. He wanted children to learn math like they learn to speak: through exploration instead of instruction. He started from the assumption that we all know it’s impossible to be congenitally “bad at French”: if you grow up in France, you’ll learn it. So Papert tried to create “Mathland”, a place where anyone could grow up “speaking math.”

The Mathland he created was the LOGO programming language, where you can draw pictures by giving instructions to a turtle on the screen with ink on its feet. I recently made a version of it you can try out online. Can you figure out how to draw a circle?

Through learning to program, I learned so many meta-skills, like debugging, composition, and logic. Most importantly, I learned that there’s literally nothing that cannot be learned. These meta-learnings of programming may explain the unreasonable competence (and arrogance) of computer scientists: why we think we can solve all the world’s problems – even those far outside our original domains.

Coding is a delightful activity that combines the creativity of writing with the precision of mathematics and the instant feedback loops of a video game. It helps you sharpen your desires into precise language that computers can carry out.

It’s remarkably close to casting spells. I think of Hermione correcting Ron – "it's levi-O-sa” – as a good model of what it’s like learning the arcane syntax of code. But once you master that, “You are a wizard, Harry.” You can encode what you imagine into an arcane language that a computer will make real. We all can be wizards.

LLMs can write English as well as they can write code, yet we have no fear for the relevance of humanities. The same intuition holds true for code.

Many dismiss code in the same way they dismiss legalese as inscrutable, tedious details. But like law, code is what our world runs on, and an elegant line of code can literally change the world. Think of e=mc2 or “we hold these truths to be self-evident” if you doubt the power and majesty of precise formal language.

And finally, programming is simply fun. It's a joy. My calling in life is to spread the joy of programming, so if you have any inkling of interest in learning to code yourself, shoot me an email – [email protected].

The dream of universal code literacy – or the "real computer revolution” – lives on, even in the age of LLMs.

👋 Hi, I'm Steve. I build Val Town, a delightful way to write and deploy code in the cloud. You should try it! We're hiring — if you like my writing, maybe you'd like working together.

联系我们 contact @ memedata.com