What goes where? In desktop applications, you try as often as possible to keep everything in one file. It’s much easier that way. The user can move it to a different directory, he can copy it to a floppy, upload it to a network, download it from a network, whatever. It’s all in that file. But it doesn’t necessarily make sense to store everything in the file when we’re talking about HTML and the Web. You could imagine, for instance, two completely different ways of implementing change tracking for HTML documents. One is like Word, where changes are marked right there in the document stream itself, and the browser or editor has to deal with them. The other is that the server always gives you the latest copy, but there is a source control database that indicates every change, and an editor can get the contents of this database by asking the right questions.

Another important issue is file size and speed. Again, these issues are not unique to the Internet, but they are magnified by slow connection speeds and bandwidth conservation issues.

A third issue is extending HTML to represent authoring information. We’d have to store a table of authors, information on whether text was added or deleted and by whom, and so on. Or some of this could be done outside the HTML source.

Finally, security is an issue.