User flow and navigation

[zuketo]

User flow and navigation improvements as part of the docs redesign.

@debadair's previous document on Navigation success and failure

CC @Adrian-J

Edit: I've begun organizing feedback in this Navigation feedback document

[debadair]

Actually, it was @Titch990 that put that document together. :-) I completely agree with her point, though.

IMO, the biggest issues with our existing navigation are:

• Nav widgets aren't always visible. (@Adrian-J's mock-ups address this--scrolling the content is independent of the nav).
• Lack of context. When you get deep linked into the docs from search, it's not obvious where you are in the overall doc set or what the "nearby" docs are.
• There are multiple (redundant) ways to move around within a book, but no easy way to navigate between related books. (As illustrated by @Titch990's example.)
• As you move between docs (or through a procedure), you lose all context of where you came from and there's no easy way to get back.

The current architecture of our books also encourages "nav" pages that have no meaningful content. They are really just a hack to get around the fact the nav widgets don't effectively show hierarchical relationships between topics. The way chunking works has encouraged a proliferation of short topics, often by accident rather than design. And as you move through those topics, you lose the organizational benefit of the hierarchical nature of those topics. Adrian's point about it being a better user experience if some of these were "squashed" into larger topics is spot-on. (But right now, having long pages is painful because you scroll past all of the nav aids.)

We also cross-link between topics and books in ways that turn the user experience into a "treasure hunt", and make the lack of signposts and navigation that gets you back where you came from even worse.

Much of this evolved out of a desire to avoid replicating content in multiple places--which is problematic for both maintenance and SEO. Reusing content with our toolchain is possible, but complicates the authoring experience. Because we don't have authoring tooling that enables us to introspect the source/highlights where content is used, it's easy to make changes that affect other docs without realizing it or miss updating related content. And the complexity makes it hard for developers and community members to contribute to the docs, when we very much need and want them to contribute.