Updated: see also #88
Interesting ideas about “think-tools” are offered by the authors of Deepamehta (caution, German start page in java takes minutes to load). I was not able to install their java behemoth, and the online demo contained empty explanation widgets and crashed after a few clicks. But I could read this presentation (caution: 30 MB PDF).
The talk argues that today’s GUIs still bother the user too much with their system-centered entities like programs, files, and windows, while they should rather let the user directly manipulate and view the semantic concepts.
- The focus shift away from files to the semantic content contained in them (or from document to data), is now commonplace and is important for the semantic web to make its agents understand the file’s content without human help.
- Of course, I also agree with the principle of direct manipulation underlying the quoted WIMP paradigm to handle larger /aggregate /virtual entities, like documents, just by dragging their icons or other tangible handles (deputies, stand-in’s, placeholders), like shortcuts or hyperlinks.
- But I doubt that I am comfortable with totally abandoning the awareness that these are still tools and, instead, thinking of them as the semantic entities themselves that are being manipulated. My reluctance against this radical approach is not due to Ockhamist nominalism but simply because it would force me to prematurely formalize my thoughts: While I would no longer need to think about boxes in slides or paragraphs in notepad, I would have to even decide what role they play before I assign them a semantic icon. And vague relationships (whose meaning is only emerging, that are expressed by tools like coloring or dotted connectors), would have to become explicit to show their semantics.
Once again, this approach seems as if someone generalized a specific cognitive preference/ style to become a one-size-fits-all solution. The authors of Deepamehta seem to hate overlapping windows and love very large-sized windows showing many panes and nodes simultaneously. They also justify this very plausibly and agreeably as auxiliary storage for our limited short-term memory. Additionally, the large displays may be justified by plausible considerations about the “sense of place” where an ideal navigation system should deliver items to instead of forcing the user to switch context.
However, to put lots of details in a large display, is just one way of gaining overview, but it is not everyone’s way, and may just confuse others. Strangely, the authors refer to their approach as akin to mind maps and concept maps, which is difficult to understand since these two are very different (see my #81).
Finally, they seem to fight a somewhat outdated version of the quoted WIMP GUI: their screenshots do not show taskbar buttons (that serve as handles/stand-in’s for currently hidden context chunks), and they describe the desktop metaphor primarily as a system of drawers and in/out baskets, rather than the desktop in the narrow sense that is a special folder’s view and serves as exactly the demanded short-term memory storage and as the context place where navigation should deliver items to.