软件工程的核心在于管理复杂性

近日，Hacker News 发表观点文章指出，软件工程的核心在于管理复杂性，而非单纯编写代码 1。文章认为，尽管 AI 能够高效生成代码，但无法替代工程师在架构权衡、上下文理解以及系统演进方面作出的判断 1。因此，在 AI 时代，算法思维变得比以往更加重要 1。

在具体应用方面，AI 可作为一种加速器，协助开发者生成样板代码、探索 API、提出实现方案、发现潜在缺陷、解释陌生代码、生成测试、比较不同方法，并减少将想法转化为工作代码所需的机械性工作 1。然而，文章也警告了 AI 生成代码所带来的代码所有权和认知负担问题，指出 AI 在让编程变得更容易的同时，也可能让软件工程变得更加困难 1。作者强调“代码行数很廉价，理解却很昂贵”，并呼吁开发者“永远不要发布自己不拥有的代码” 1。基于此，文章提出生产力的衡量标准不应是软件或代码的产出量，而应是在有效控制复杂性的前提下解决问题的有效性 1。

The core of software engineering lies in managing complexity rather than merely writing code 1. Artificial intelligence is extraordinarily useful for software development, serving as an accelerator for generating boilerplate, exploring APIs, proposing implementations, finding potential bugs, explaining unfamiliar code, producing tests, comparing approaches, and reducing the amount of mechanical work required to turn an idea into working code 1. However, while AI makes programming easier, it potentially makes software engineering harder 1.

AI cannot replace engineers' judgment in architectural trade-offs, context understanding, and system evolution, making algorithmic thinking even more crucial in the AI era 1. The author warns against the issues of code ownership and cognitive burden brought by AI-generated code, emphasizing the rule to never ship code you do not own 1. Furthermore, lines of code are cheap, but understanding is expensive 1. Consequently, productivity should not be measured by how much software is generated, but by how effectively problems are solved while keeping the resulting complexity under control 1.