I've worked with a ton of phenomenally talented user interface and interaction designers, and we are always struggling against the boundaries of the interface. In doing so, we've zigzagged our way through a lot of crazy territory. Along the way, we came to some hard truths about what does and doesn't work when structuring a Web experience.
I'm not talking about the programming necessary to build the appropriate Web site. I'm describing the three forms that Web pages most closely cleave to, from the information architecture and user experience onwards.
Much like in the film Powers of Ten by Charles and Ray Eames, how the world systematically and naturally organizes itself is echoed in how we architect Web experiences from the ground up:
Web site as book: "Page model"
- Organization echoes book/magazine content patterns formed by humans from natural ratios
- Ideal for long-form content presentation
- Fulfills user expectations through consistency and legibility
- Most common site organization by a billion to one
- Blogs, content sites, transactional sites, and most Web applications conform to this model
- Information model hasn't changed much since Gutenberg
Web site as garden - "Physical world model"
- Organization limited to how we treat objects in physical space
- Somewhat flexible, but can only contain a limited amount of content in shaped zones
- Fulfills user expectations through mimicry of real-world restrictions (gravity, reactivity)
- Commonly applied to entertainment/marketing sites and Natural User Interfaces
- Information model hasn't changed much since apes roamed the Earth
Web site as petri dish - "Molecular model"
- Organized to support data in an exploratory/non-hierarchical way
- Engineered for short-term exploratory usage patterns
- Fulfills user expectations through novelty, reactivity, and ability to be reshaped
- Most popularly used through data visualization, starting to bleed into marketing experiences
- Information model hasn't changed much since the Big Bang
There is a constant tension and balance between these three types of systemic order. And they can mingle, but in a very constrained manner. Humans just aren't very adept at perceiving fundamental separations between text (words), physical objects (proxies for content), and molecules (metadata). If you go and play on WeFeelFine.org, you can see why you can't click on little flying molecules while also reading a page from a book by Dostoyevsky. If you try to mush two or three of these Web site types together, chances are your user experience and page design are going to fail.