You run your viewer application (called a web browser) and use it to download a data structure (called HTML) that describes a document. The browser takes that structure and turns it into a styled layout (called a web page) for you to read. Pretty cool. But it gets much cooler.

You can generate that data structure (HTML) programmatically. Which is to say: its meaning can be driven by behavior and other data. And it gets even cooler than that.

The data structure, before being displayed as styled text, is converted into a tree structure (called the DOM). You can run code to inspect and modify this tree structure. This adds another place where you can intercede behavior.

Let's recap: you generated some output via one means, then you ran some code on the output to sculpt it further. In a viewer application. Incredible!

The geniuses who made Netscape 2 conceived the DOM as an interface for a programming language to use to manipulate HTML.

The DOM has some fantastic properties:

• It can be accessed from a programmable context.
• Its structure maps directly to the structure and data of the document it describes.
• It is comprehensive, exposing basically every component of the document.
• It is live. Modifications get back-ported to the document.
• It allows interactive programming. When code is run, the change is usually visible and always appears immediately.

These properties make the web programmable. They enable almost all of dynamicness that we think of as a modern web app. They enable tools to be built on top of the environment. For example: inspectors and debuggers like Firebug that present intuitive output.

Having the DOM available to develop against feels like a privilege. I love it.

Part of Mary Rose Cook’s research to build software quickly