[amluto]

My take is that they had more limited tooling and resources, so they designed a set of controls, made them visually distinct, and used them almost everywhere. It was easy to make radio buttons that looked like radio buttons and worked like radio buttons. It was substantially more difficult to do anything else.

Also, the default interface wasn’t composited, so nothing was translucent, so no one layered their controls and content. Sure, it looks kind of pretty in Material Design when a round button with nice antialiased edges sits on top of the content, but it’s terrible UX. In Win2K, if you wanted to do this, you either used “layered” windows (which were no amazingly slow that no one wanted to use them) or you had to go outside the Win32 library entirely and render the whole mess yourself. So designers mostly didn’t do this.