It's an old problem: if a technology is standardized too early in the process, it won't have the time to evolve naturally into something stable and future operatable. If it's standardized too late, it's already split up into so many manufacturer proprietary sub-standards that it's becoming harder for content providers to actually serve the user.
If you connect computers all around the world, it's simply impossible to standardize hardware, like screen resolutions. In the world wide web, you also don't know the user, nor the medium. For document exchange it becomes a must to find a basic exchange format; it needs to be downward compatible and simple, but also by being simple powerful enough to be upwards compatible. The simple idea behind XHTML for example is to only provide a very basic set of structural properties in a document, plus general mechanisms for media inclusions (like images). In order to manage a lot of information, it's becoming very important to follow strict rules as to what language is actually spoken to organize the data. If a computer misinterprets the data, information is lost.