Interface Design, UNIX and the Macintosh

August 3, 2003 11:03 AM

C. Bradley Dilger's research on ease of use. I'm reading Neal Stevenson's cryptonomicon and it got me thinking about interface design. Of course not all artists design interfaces, nor do they really want people to see their art as interactive, but for the rest of us I think this is an important topic. Mr. Dilger's Ph. D. dissertation is over 200 pages of current, well written anaylsis on the concept of "ease" in our culture, especially when it relates to technology and computer interaction. And to make it even better, his bibliography is top. I especially liked this article titled The Anti-Mac Interface.
posted by abulafia (9 comments total)

As always, this browsable hierarchical index can be used to access the web pages still here. (http://web.nwe.ufl.edu/~dilger/)

Geez, there's nothing easy about that set of poorly-named directories and files.
posted by Ayn Marx at 12:02 PM on August 3, 2003

I find interface questions utterly fascinating, though my interests are less about design and more about the cultural feedback loops. I adored Cryptonomicon for similar reasons (not to mention the bloody cool WW2 part of the book), so let me recommend a few other interesting, not necessarily technical, reads: Neal Stephenson, In the beginning was the command line - fun stuff, and a lot about the joys of BeOS; Steven Johnson's Interface Culture, and last but not least Robert Johnson's User-Centered Technology, which is the most 'academic' of the three.

Any other suggestions out there?
posted by hank_14 at 3:49 PM on August 3, 2003

hank_14: I wrote an essay about interface design and critical theories. Steven Johnson's book was helpful. I used the 2 categories of informational and experiential as the basis for analyzing interfaces, and considered issues such as immediacy / hypermediacy, authorship, art world processes. You can read it here. It's a few years old, but I believe the issues are still relevant and still debated.
posted by shortfuse at 6:22 PM on August 3, 2003

I would love an interface that combined what we have now, a command line for scripting in an concise and powerful language, and the ability to give spoken commands. However, what I would most like would be a "smart" interface, one that could parse natural language to some extent, enabling me to say "extract all zips in the current directory to individual sub directories" for example. Or to say, search the following news sources for information x, y, z and correlate the results by criteria u,v,w". Of course I can write a script to do something similar, but its tedious, and I don't want to, and it wouldn't suit every situation. However there are other times where the current metaphor based interface is perfect, for example when working with graphic design applications it just makes sense to have a color palette, a paintbrush, an airbrush, an eraser, etc. It does not, however, make sense to have to drag files from one "folder" to another, and with the incompatibilities between dos 8.3 name conventions and windows long name conventions, doing it from the command line can be frustrating if you are deep in a directory structure full of long names.

Also, who the hell thought up that stupid folder thing? What's wrong with a directory?
posted by Grod at 6:24 PM on August 3, 2003

'Folder' replaced directory to go with the whole 'desktop metaphor' idea. Cf. trash can/recycle bin.
Not that I like it any more than you do.
posted by Utilitaritron at 6:46 PM on August 3, 2003

I enjoyed the Anti-Mac interface article, but the call for a heavily language-based interface seems to have rose-colored glasses on it. You want to tell the computer "Notify me if there are any new messages from Emily"? Great. Sounds so simple. But there are literally dozens of variables lurking there to be teased out.

First, "new" messages are clearly those coming in after the command is made. But is it only the "Emily" in my address book or any spam that comes through with the name "Emily" on it? And if there's only one new message, does that qualify as "new messages?" The answers to these seem obvious to us, but may not be to the computer, depending on what phrasings it's been set up to comprehend. So fine, you engage in a "dialogue" with the computer to make sure the command is understood.

But that raises more questions. What if the computer doesn't get the way you tend to phrase things? What if you instead happened to say, "I want to know when new messages from Emily come in," or "Tell me when Emily writes" or whatever. Would the computer know to ask "do you mean you want to know the time and date of Emily's new messages, or you want to be notified as soon as they arrive in your inbox?" or would it sit there and say "OK, chief"? In other words, to some extent you would have to learn to speak in the way that the computer can understand you, rather than vice versa, because if you can't, the whole "dialogue" to straighten things out is recursively doomed.

This doesn't mean the language-based interface is a bad idea. It just means that there are a lot of small liabilities and glitches that could make it just as frustrating as the GUI or DOS interface, and I think the authors were glossing over that a bit. Still, it gives you plenty to think about.
posted by soyjoy at 7:39 PM on August 3, 2003

Having the clarification dialogue happen as a simple multiple choice question (or questions), answerable via keyboard or mouse rather than another spoken comment, could make things easier.
posted by Tlogmer at 2:56 AM on August 4, 2003

I'm not to familar with computing, though I took a class in college with David Gelertner of Unabomber fame. This Wired article details his vision of a new OS, known as LifeStreams. I believe it is being commercially produced (or at least actively developed.)
posted by pjgulliver at 12:08 PM on August 4, 2003

soyjoy, I do think that the language-based interface is a bad idea, for exactly the reasons you describe: natural language is ambiguous. I don't want the computer to have to guess what I mean, no matter how good its language-processing abilities. Even if the computer were as smart as a human being, it would still misunderstand me enough of the time that I wouldn't be able to trust it.
posted by Mars Saxman at 1:02 PM on August 4, 2003