Maggie Appleton 写设计、人类学和智能体界面时，习惯先画清楚隐喻，再谈工具。把「A Picture Worth a Thousand Programmes」整理成可阅读的中文笔记：问题在哪、界面默认会把人带去哪、落地时该改什么。原站导航、评论回响和广告已去掉。

Transcript and Slides

Developers who are familiar with modern programming interfaces and tools. Ideally JavaScript, but not necessarily.

This talk – “A Picture Worth a Thousand Programmes” – is an exploration of why we don’t use more visual explanations and embodied knowledge in our programming tools.

I. What’s wrong with text?

I first presented it at React Advanced London in 2021 5ya . You can watch the recording on the official conference website:

Or if you prefer words, you can find the written version below.

II. What can visuals give us that text can’t?

Transcript and Slides This is about making pictures. Specifically, making pictures of programmes. Ideally pictures worth a thousand programmes, as the traditional saying goes.

I’ve spent my whole career so far creating visual representations of programmes.

Visual mediums reveal metaphors

During my time at egghead I made hundreds of cover illustrations for the courses we taught.

Each of these would start with an abstract programming concept that I had to find a way to visually represent in a single image

Visual mediums reveal space

I learned to rely heavily on visual metaphors for these. Styling with CSS became painting a house. And became organising types in typescript became sorting suits in a deck of cards.

I’ve also made a lot of illustrated diagrams over the years to explain topics like how JavaScript prototype inheritance works, or what happens when you flatten an array, or how APIs are like small robotic waiters.

Visual mediums reveal change over time

The goal of all these illustrations is to make complex technical topics relatable and easier to understand, a theme you’ll hear a lot about in this talk.

I also recently collaborated on a project called Just JavaScript with Dan Abramov

值得单独记下的观察

• more accessible to more people
• ‘What’s wrong with text?’
• ‘What can visuals do that text can’t?’
• “Haven’t we already tried this?”
• It’s flexible and it’s easy to move between applications using copy and paste without worrying about compatibility.
• You can pack a dense amount of information into a very small space
• You can be very specific about what you mean, which obviously matters when we’re talking to computers who have no sense of nuance.
• They reveal fundamental metaphors embedded in our programming languages.

落地时建议先做的 5 件事

1. 先写清这个工具在强化思考还是在替人思考。
2. 默认交互不要只会谄媚：该追问、该给反例、该要求证据。
3. 智能体界面要露出推理步骤，而不是只给一个光滑答案。
4. 知识发布用可生长的笔记，而不是一次性营销长文。
5. 改产品前先画隐喻：用户以为自己在做什么，系统实际在做什么。

和智能体产品怎么接

龙虾PRO做 OpenClaw 落地时，最该从这类笔记里拿走的是「别把聊天框当唯一界面」。数字员工要能追问、能验证、能把过程摊开，而不是只负责说好话。

本文侧重全链路风控方法论。落地时请用自身业务单据做回放验证，不要把示例阈值直接当生产策略。 相关：风控体检 · 方案资源