I too have been asking myself this question ever since I read The Humane Interface. Although Jef Raskin laid out his ZoomWorld as a complete idea, the real ZUI expert is Ben Bederson. He has been building and studying zooming interfaces for decades. He also has some of his own answers to your question. He summarized, ‘manual organization of information in a zooming environment rarely makes sense’. He wrote an entire paper on the question, ‘The Promise of Zoomable User Interfaces’. From the section ‘Why ZUIs Are Challenging’ (pp 5–6):
As summarized in Table 1, the potential benefits of ZUIs are sometimes mirages. ZUIs are generally engaging (although they make some people feel physically sick) and visually rich. But the promise of simplicity falls short.
While human visual perception does make it easy to see where one is navigating, the reality is that it places a heavy load on short term memory to remember where in space you just were and where things are. And the requirement of human memory to know how space is organized means that ZUIs don’t scale up very well. ZUIs are often motivated by the physical world and how people like laying papers out on their desk. But no one wants all of their papers on their desk. It is much more common to have only a relatively small number of papers that are actually being worked with.
The visual overviews that ZUIs offer for free by zooming out may seem like a solution to the load on human memory, but in practice it doesn’t because visual overviews of any complexity require significant scanning and visual search in order to find anything. If there are just a small number of objects, then the visual search task is not hard – but of course, for a small number of objects, you don’t need a ZUI to solve your organizational problems.
Finally, the visual richness of ZUIs is a double-edged sword. It requires skill to design a complex space with documents of arbitrary size, aspect ratio and color that people can comprehend and scan. Also, people are not as good at scanning 2D designs as 1D layouts – unless the layout is highly structured. But highly structured 2D layouts don’t work well for visual objects of arbitrary aspect ratios. Designers are obligated to leave a lot of unused space, scale down the large objects so they are unreadable, or crop the large objects – thus losing much of their distinguishability.
I’d like to add a few points of my own. First, Jef Raskin’s zoomable interface makes the most sense in a document/object-centric context. Application-centric environments are conceptually at odds with such interfaces. Raskin’s ZUI is good for organizing large amounts of information in a way that makes applications obsolete; but the current computing landscape is all about “apps” (following Apple’s lead, from the original Macintosh to the iPhone). Even the iOS home screen, a kind of minor zoomable interface (especially after nested folders were added), is restricted to organizing app icons, not real content.
Second, paradigm shifts tend to take a long time. Concepts like graphical interfaces and direct manipulation were invented in the ’60s, but it took until the mid-’80s for these paradigms to be available to the general public. The Internet and hyperlinked documents took even longer. Multitouch screens go back to the ’80s, but the iPhone was the first generally-available affordable implementation. Zoomable interfaces are often more resource-intensive; by the time they were practical, traditional GUIs were quite established.
I believe a “killer app” is necessary in order to facilitate this app-centric–to–object-centric paradigm shift. As Alan Kay says, the best way to predict the future is to invent it. In other words, the best way to answer why ZUIs aren’t more popular is by negating the premise: build one! (That’s my own plan.)
Update: It took several decades since the invention of the concept, and several years since I answered this question, but there are now quite a number of zoomable interfaces, now called “infinite canvases”. There’s even an online list of them!