Taking advantage of the hour and a half lull of a meaningless football game (which pretty much made working late impossible, since everybody else went home to watch), I decided to hold off on my regular news review and catch up on some reading.
Oddly enough, it wasn't until I had put down The Humane Interface that I realized that Jef Raskin has gotten a US $2M grant to make it a reality. I've been re-reading it for a couple of days or so, and comparing it to my experiences with Quicksilver.
My main gripe with Archy (as it is now called) is that it doesn't take much thinking to figure out that it completely breaks down when the conceptual model of what you're trying to do can't be modeled as straightforward text, and I can't envision it being usable for the sort of tasks and commonly accepted deliverables on today's electronic office environment.
I've gradually been shifting away from big, complex Office documents to small, manageable Markdown documents that later evolve to Office documents, so it does have an appeal for me, but, in the end, humans need strong visual cues and references to read and maintain documentation, and spatial orientation and all-in-one documents simply don't cut it.
Although I have yet to go out and install Archy, I think I have a far better experience in Mac OS X using Exposé and Quicksilver to juggle snippets of text around, slicing and dicing them through Mac OS X system services such as HumaneText and sending them to Stickies or a command prompt - I get spatial orientation from Exposé and extremely powerful data manipulation from Quicksilver, and all I need to get started is a simple text editor to act as scratchpad.
I suspect that Archy will eventually look a lot like vim on steroids, with a smooth graphical view, intelligible commands and a couple of dedicated hotkys, but won't even come close to the seamless way I can juggle data across applications with Quicksilver. And yes, I understand part of the point is re-addressing the concept of what an "application" should be.
In the end, though, all of this is moot. My opinion is that Archy (at least as I understand it) is too far removed from the current accepted practice for human-machine interaction to be widely adopted, although some of its principles are likely to find their way into mainstream software.
In the end, it's not like some of them haven't been there in one form or the other - Jef's added value is to have systematically broken down and thought about how to piece them all together in a natural way.
I'll give it a spin when I can, but I'm not expecting that big a thing. And now, I'll just use Quicksilver to log in to my Wiki and post this by simply typing a couple of words that invoke the relevant bookmark...