When a window isn't a window, just a pane

By David Weinberger

All my elderly friend wants to do is use a Web browser to pick up his e-mail. You'd think he'd be qualified, what with his law degree from Harvard and all. But, nope, he can't do it.

I started him with Windows XP. After failing to train him to move the mouse horizontally to make the mouse cursor move vertically, I got him a trackball. Not only did that get him past the cognitive non-alignment of the dimensions, it also meant that he didn't have to learn how to pick up the mouse and move it back to the beginning of the mouse pad when he wants to move in a continuous line.

I understand his difficulty. It really does make no sense. Airplanes, for example, don't have to go back to the beginning of the air strip if they want to move forward more than 500 feet. My friend certainly grasped the concept of how to move the mouse. It's just that his fingers refused to unlearn what 78 years of life in the real world had taught them about how to move forward.

The mouse problem was just the beginning. But we solved the next set of problems rather rapidly, for Windows XP has a set of tweaks useful for those who don't move quite as steadily as they used to. I set the desktop to single-click and increased the number of pixels that count as the effective range for a click. I boosted the font size a tad (or possible two tads) and chose a high-contrast color scheme for the icon labels.

I really thought we were just about home. I put a copy of Internet Explorer on his desktop, set it to go to my friend's e-mail service as soon as he clicks it, and told Explorer to fill in his username and password automatically. Once he gets to his inbox, he knows what to do.

Someone must have spiked my lemonade. I never realized just how complex even the simplest of computing tasks is. It's not even the sheer number of terms and movements one has to learn. Rather, it's the fact that there is no coherent framework for putting those terms and movements together.

For example, my friend just doesn't see why I call that stretch of screen real estate his "desktop." It doesn't look or act like a desktop. so he thinks of it as "the picture" because I loaded a family photo as wallpaper. To him, icons are, of course, also pictures, just smaller in size. Nor does he think of windows as windows. Windows are things you see through. Besides, does your real world desktop have windows?

It's not that he's not smart enough to "get" this. He's too smart. The way we've mapped our computer interface to real world analogues is just plain weird, and the way we have clustered our concepts in this untethered alternative universe gets in his way.

For example, he eventually became accustomed to calling windows "windows." But when I'm doing phone support for him, if I have him click on a drop down menu, he'll report that a new window has opened, obscuring the other window. Now, technically a drop down menu is a window, but those of us who have become acclimated to the PC fun house don't see them that way at all. And it makes a difference. The fact that menus are like windows but not windows disrupts the simple model I gave him: Every window, I told him, has an X in the corner that lets you close it. That's his escape hatch. But along comes these new windows—drop down menus—that don't have an X and that can only be escaped by clicking somewhere outside of them. That doesn't work with what the rest of us think of as windows. Worse, if you click outside of a menu and happen to click on a link or a button, a new window comes up. It is a remarkably complex system.

My friend has trouble with Windows because there is no natural model by which he can appropriate it. He has to learn it metaphor by metaphor, quirk by quirk. Without anything to fit the pieces into, there is little difference between learning and guessing.

There are obvious lessons here for people charged with writing product documentation and for deriving new nomenclature for novel products. But the problem becomes more serious if less obvious in discussions further up the tree of abstraction: How can an organization learn anything at all if lacks a sound model--not just codified in some official ontology and taxonomy, but expressed tacitly in every lived conversation--of what it already knows?

David Weinberger edits "The Journal of the Hyperlinked Organization", e-mail self@evident.com