问HN：Markdown之后是什么？
Markdown最初是HTML的一种简写方式。现在，它已成为文档、笔记、知识库和人工智能上下文的默认格式。
有趣的是，Markdown在不断吸收新功能的同时，格式本身并没有改变：
- Mermaid：从代码块生成图表
- KaTeX/MathJax：通过`$...$`语法渲染数学公式
- Frontmatter：通过YAML块提供结构化元数据
- MDX：在Markdown中嵌入React组件
- Obsidian/Logseq：反向链接、画布视图、图形可视化——这一切都来自普通的.md文件
这种模式似乎是：.md文件保持人类可读的纯文本格式，但渲染器变得越来越强大。同一个文件，输出却更加丰富。
这让我思考未来的发展方向：
1. Markdown是否会通过渲染器的约定不断演变，直到成为一种事实上的交互式文档格式？（“HTML路径”——HTML几乎没有变化，但CSS/JS/浏览器使其具备了无所不能的能力。）
2. 是否会出现一种新格式，能够原生表达交互性、可折叠部分和嵌入计算？一种介于Markdown和Jupyter笔记本之间的格式？
3. 或者答案涉及一种协议/中间件层——.md文件是源文件，但某个中间系统（如文档的语言服务器）在其上添加结构、验证和交互性？
我特别好奇的是人工智能的角度。普通的.md文件是最适合人工智能的知识格式——任何大型语言模型都可以在零设置的情况下读取、写入和搜索它们。更复杂的格式可能会获得更强的表达能力，但会失去这种特性。
你怎么看？在更好的渲染器下，.md是否“永远足够好”，还是我们正朝着某种新事物发展？
查看原文
Markdown started as a shorthand for HTML. Now it's the default format for
documentation, note-taking, knowledge bases, and AI context.<p>What's interesting is how it keeps absorbing new capabilities without changing
the format itself:<p>- Mermaid: diagrams from fenced code blocks
- KaTeX/MathJax: math rendering from `$...$` syntax
- Frontmatter: structured metadata via YAML blocks
- MDX: React components embedded in markdown
- Obsidian/Logseq: backlinks, canvas views, graph visualization — all from plain .md files<p>The pattern seems to be: the .md file stays human-readable plain text, but
renderers get increasingly powerful. Same file, richer output.<p>This makes me wonder where this goes:<p>1. Does markdown keep evolving through renderer conventions until it becomes
a de facto interactive document format? (The "HTML path" — HTML barely
changed, but CSS/JS/browsers made it capable of anything.)<p>2. Does a new format emerge that can natively express interactivity,
collapsible sections, embedded computations? Something between markdown
and Jupyter notebooks?<p>3. Or does the answer involve a protocol/middleware layer — where .md files
are the source, but some intermediate system (like a language server for
documents) adds structure, validation, and interactivity on top?<p>I'm especially curious because of the AI angle. Plain .md files are
the most AI-friendly knowledge format — any LLM can read, write, and search
them with zero setup. A more complex format might gain expressiveness but
lose this property.<p>What's your take? Is .md "good enough forever" with better renderers, or are
we heading toward something new?