- From: Al Gilman <asgilman@iamdigex.net>
- Date: Thu, 15 Jul 1999 11:48:59 -0400
- To: w3c-wai-gl@w3.org
[I am sending this separately to GL and to UA - Al] Two ideas: 1. I am concerned that too much attention is being focused on icons or symbols, graphical mnemonics for items or atoms. The real genius of the graphic plane is its ability to display multiplayer situations: constellations of items or atoms. In terms of our ability to understand graphic language that non-readers can comprehend, we need to be looking at the relational cues that show how the depicted items relate to one another. For example, the navigation frame, whether top right left or bottom, is always immediately within the border of the total frameset rectangle. It is there to provide contextualization glue; so it is placed topologically between the environment of the browse window and the specific content that defines the logical "here." 2. I was struck by the idea that scrollbars can be an insurmountable cognitive hurdle. Of course when they were new there was a lot of flapping caused by the confusion factor that when the control moves up, the controlled stuff moves down. Nowadays we don't give it a second thought. But that is just because is is part of our culture; the stuff we do by reflex and forget we ever had to learn. But there are people for whom this molehill is a mountain and rather than a speed bump, it is an impassable barrier. In terms of a fix, I think that maybe the UI for the nonreader should borrow from the art tool conventions. This is the cursor pallette. The user overtly goes to an agent selection pallette, picks what kind of operations they want to have associated with the cursor, and then moves the cursor through the displayed view to associate the tool/role that the cursor has picked up to objects in the scene to operate on them. For example, a 'speak' tool for controlling the voicing of text. This to me is how to apply graphics to make the interaction processes comprehensible. Use the move-it hand rather than scroll bars to reposition the viewframe. Just as with adaptation for the single-switch user, we are here dealing with the styling of user command input and its integration with the display channel, and client-side profiling of that styling from a profile-generic articulation of what is there that can be done. Al
Received on Thursday, 15 July 1999 11:42:53 UTC