concepts: one-drag vs. shelf, piles, association by context, filter views, deep context switch vs. shallow, unifying disparate UI's, Is simple keyboard navigation forgotten on the mac? why I think users want something different, but not too different. Why are we still only using two dimensions?
When you're moving your icons around, sometimes you have to pause what you're doing to set up the place you want your icon to land. You may start a drag, then realize that your destination is not exposed, perhaps behind a window, perhaps it's in a window that's not open, or worse, in an application that is not open. Some browsers offer a neat halfway point for drags called a "shelf", where you can sort of park your drag while you re-arrange windows. This is neat, as a stop gap. You could park as many drags as you want and resume them later. DragThing can accomplish this quite handily.
What I'd like to propose is a way to not have to stop your drag in order to set up your destination. Wouldn't it be neat if, DURING a drag, without letting go, you could switch applications, even launch applications, switch windows, open windows, or otherwise continue to navigate toward your destination? The Finder offers part of this with the notion of "spring loaded folders". During a drag, if you hover over a folder icon for a set time, it will actually OPEN that folder, presenting a new window with it's host of potential drag destinations. But it stops short of launching applications or letting you tell a window to go away by hovering over it's close box. I'd like the ability to take a drag right into the Window menu and onto a window name, and if you hover there, bring that window to the front.
DragThing is SO close to being able to do part of this. IMHO it would be SO powerful to have spring loaded application switching/launching. It irks me to no end that frickin' WINDOWS lets you do this. You can start with all running applications minimized and take a desktop icon and drag it over a tile in the bar at the bottom, if you hover for a second, the OS actually switches to that application. Why doesn't DragThing offer this? I say, go further and allow application launching too. The point of One-Drag is to be able to accomplish a whole task with one click, drag, let go gesture.
Back in the Good-Ole-Days of Macintosh System 7.6, we had the most amazing integration of networking ever seen in any OS ever. To send your friend a file was one-drag. The "Address Book" was not a separate application, it was a folder. Duh. So you could pick up a file/folder/whatever, and drag it onto your friend's address entry icon, and pshing! off it went. Dam i miss that.
Piles & Context
You know what a pile is? See this: http://homepage.mac.com/rdas7/piles.html
Now you know. This is, IMHO a neat concept. It shows that a group of documents are related without enclosing them in a folder. Then again, does this give you more power than a folder? Piles is an implementation of "association by context". Context in this case is proximity with the other documents in the "pile". I'm not totally clear on the concept here but it seems like all the items in the pile are within the same folder. There may be other items in that same folder that are NOT part of the pile, but I *think* the pile is inside a folder. But i'd like there to be a way to make piles that live outside the whole notion of folders, like a freeform database file system. That way, just because two documents lie next to each other, or even associated by pile does not mean they come from the same folder, or even the same file system, for that matter. I imagine, one day, a 3D view into your database of files. But it's not about folders and hierarchies, it's about relationships between files. For example, you could say "show me all the documents relating to my web site" and you'd see a 3D "tree view" of your site, but the "tree view" isn't due to how the documents are stored on your hard drive, it's due to the branching nature of the web site. Sure, today, i set up my sites such that the branches on the site mirror the folder hierarchy on my disk drive, but that's an artificial mnemonic convenience.
To view into a database, you run a query. This query specifies the types of results you wish to see, and can list them out in a traditional "details" view. It would be great if file systems could be viewed this way. In fact they can, in a way, that's what the "search" window is for. But that view is rather grafted on top of the traditional hierarchic file system, it's not a true database. I'd like to have way more metadata than just "name, size, created, modified" etc. I'd like to be able to add my own arbitrary metadata such as "part of the rose web site" or "photo from the oregon trip", so i can filter on these concepts later, and retrieve a "folder" full of just the items i want. Then I can copy that folder in one fell swoop, or anything I drop into that folder inherits the metadata too. Which brings up another idea about association by context. If you have a 3D tree view of some system, be it a true hierarchic file system, a web site, or some database, the act of moving a file near another file (in the sense of piles) should be able to create a link or association. Even if that file is from another view, file system, database, or universe, you can still move it near the first file, since you're in 3d space, without dissociating it from it's previous connections (ie: without removing it from it's original parent folder or file system). Even if your filtered view shows the first file but doesn't show the associated file, you could select the first file to view it's associations and perhaps glean something more meaningful about that file when you see it's associations pop up.
Posted: Thu - March 4, 2004 at 12:21 AM