I *really* hope someone takes your suggestions and builds a customized Linux distro around them… if for no other reason than to make MS & Apple take notice.

I guess if Scooter’s trolling, then I am too. Statements like this really do detract from the rest of the article. Based on this depth of understanding, couldn’t this also be true?

“Only programmers want to live in a grey, motionless, silent world of text.”

I mean programmers just work in text, right? No, they use text *to accomplish something else.*

Maybe both professions are a little deeper than that.

Please meet some librarians and find out how much they like to read, research, think, learn technology, not learn technology, use technology, teach technology, help others research, try to make things easy for users, do things for users when it’s not easy enough, organize information, make information available, develope useful services, provide reliable services, protect user privacy, protect access to information, and above all, how different they all are.

This is all discussed in gorgeous detail in Jef Raskin’s The Humane Interface. Amazingly, most of your insights appear there too, though much more fleshed out. For example, your point about one long list being faster than separate shorted lists is due to plugging in the numbers into the equation for Hick’s Law. Were your thoughts influenced by the book? I would recommend it to anyone thinking about interface design, especially anyone who wants to rethink the whole desktop. I couldn’t recommend it more—it’s that profound. At the very least, people should watch Aza’s talks.

As for your suggestions, I think you’ve discarded the ZUI concept too quickly, due to having tested a poorly-implemented and desktop-like version. A real ZUI, a la Jef Raskin’s, would never differentiate between manipulating a proxy for a document and the document itself. All you would see is the content—no icons (or thumbnails) and separate windows. In fact, the boundaries between different “files” would disappear. All there would be is your information, arranged how you want it (either manually or automatically, however you wish). And administrative debris would not exist, because all functionality comes in the form of commands, like what we see with Ubiquity (which you mentioned). And they would work system-wide, because all formats are ultimately converted to and exposed as the one document format (which is also the essentially the filesystem). You could search absolutely anywhere, including within commands and their descriptions. And documents never ever annoyingly overlap each other. What I would love to see is a ZUI for Google Wave integrated with Ubiquity: one universal document format (HTML5), commands instead of applications, tagging, collaboration, undo everywhere, search anywhere, and super-Tuftean information design (thanks to zooming).

I eagerly await Part 2 (and to see if your Portals proposal includes any humorous Portal references).

@CodeMonkey: To clarify My name is not important’s reply, the web is already document-centric. Web pages are documents. Even “web apps” are just fancy dynamic documents, and the good ones give each of their documents different URLs, allowing one to work in a fully document-centric manner. It fits quite well into the application-less model. The “task” to which you refer is not a whole application, but a simple command: browse. This could easily be part of the command system (such as Ubiquity). And the web-page–rendering would be handled by completely separate component of the system. You will not find one task-oriented application that cannot be more humanely redesigned as a command. People want to perform their tasks, not to get into task modes.

As unlikely as you are to read this, I do think you have a point, but I think that tasks and objects can be unified.

The object for web browsing is a page (tweak the abstraction level as you desire; it may not actually be a page). The object for VOIP is a conversation, but you don’t “go” to a conversation; you have a system-wide address book, and you perform the “Call with VOIP” action/operation on them, thus creating the conversation. For web browsing, possibly a page. Possibly something else. We can’t really know without experimenting. For Twitter, I’m also not sure. But I think the model fits: objects, views (both representation and manipulation) on these objects, and actions on these objects (triggered by views) that either transform it, mutate it, or make a new object using it (and perhaps other objects, too).

If you do reply, please subscribe to this post’s comment feed, or at least check back. :)

I do disagree that Compiz reduces usability, though I understand why some think it’s just a messy amalgam of transition effects. Despite my fetish for minimalism, I honestly enjoy working on my Ubuntu box because everything’s easy to shuffle around and between–though it definitely required a fair bit of metawork to set up the movements and hotkeys that worked best for me. At its best it represents a hybrid of a literal desktop’s reassuring spatial sense of “I know where X object is” and the virtual desktop’s ability to manage a large number of active objects at once.