Charles Arthur On Technology

Travels in the third dimension

Ten years ago I saw a stunning demonstration by Steve Benford of Nottingham University, who was trying to answer the question he could see people would be facing in the decade ahead. The problem, he perceived, was that we would have lots of data but very limited ways to examine it, principally two-dimensional graphs.

Ten years ago I saw a stunning demonstration by Steve Benford of Nottingham University, who was trying to answer the question he could see people would be facing in the decade ahead. The problem, he perceived, was that we would have lots of data but very limited ways to examine it, principally two-dimensional graphs.

What Dr Benford - as he was then - had developed was a visualisation system that could take a huge raft of data and produce a three-dimensional view. What had been a selection of files with interrelated data suddenly became a vortex of pages, like a frozen view of a tornado: how high up the "twister" a particular file or page was indicated one aspect of the data, and then it could be close to or far away from the centre depending on another criterion, and then its position on the circle relative to the "front" meant something else. Suddenly you got a completely different idea about how the pieces of data fitted together. Moreover, the visualisation allowed you to "fly" through and examine each piece individually, and look at what other pieces it related to. Start to draw lines between the data, then realign your axes to focus on a new set of criteria, and pretty soon you have an entirely new way to consider the information you've collected.

What happened to that project? Steve Benford is still at Nottingham (though now deservedly a professor), working on the problem of "advanced interaction and communication technologies for rich and dynamic social interaction". But the 3D twister of data seems to have disappeared into some computing cupboard.

A question arises: why are we still struggling with boring hierarchical file structures to view stuff on our hard drives and on the web? Why don't we have three-dimensional desktops, with windows that you can spin around and write on the back of, for example? Where are the advances in technology that the advances in processing power should have afforded?

Some of it is here, though in very limited amounts. Apple's OSX operating system can make windows of varying translucency, which give some sense of depth to the screen; drop shadows on the windows "behind" others also give a sense of depth.

However, as Apple and many programmers have discovered, translucent windows are very hard to work with usefully; text on windows behind them tends to show through, making it hard to concentrate on what you're working on. As OSX has evolved, the amount of translucency in normal settings (such as drop-down menus) has been reduced dramatically. Translucency turns out to be one of those attributes that looks great until you come to use it.

One wonders if the three-dimensional desktop will turn out to be the same. But that hasn't stopped many people from attempting to create it so that you can have a try. One such is Sphere, at www.hamar.sk/sphere/.

The concept of Sphere, according to its author, is that "the [user] environment is... represented by a sphere. The user is exactly in the middle of it. All objects are situated around the user. He can easily turn around and manipulate the objects. All the objects that users are used to having on their desktop are now integrated in a three-dimensional environment." It uses the standard two-button mouse for navigation.

Yet one of the problems with arranging your computer stuff in a theoretical sphere that extends around and behind your head is that we aren't used to working in such an environment. While it's true that the popularity of shoot-'em-ups such as Quake and Doom means that there's a whole generation that is completely used to the idea of a virtual space that exists and changes "behind" them, the fact remains that even in those games you focus on what's ahead. Humans are very front-oriented animals: notice where your eyes and ears are. I think Sphere is interesting, but I'm not sure it's going to take over the world.

The other approach to three dimensions is to accept that people use computers in a two-dimensional screen-based fashion, but to give the objects inside that screen a three-dimensional capability. This is the thinking behind Sun's Project Looking Glass ( www.sun.com/software/looking_glass/index.html), which will be a Java-based system initially aimed at the Linux operating system and Sun's own Solaris (though one can imagine the Java element being ported very rapidly to Windows).

Project Looking Glass would let windows twirl so you could stick notes on their backs; let you create a "gallery" of related documents; and pan "around" the edge of the screen to other documents. Although Sun has no set date for introducing Looking Glass, the expectation is that an early release will try to steal some of Microsoft's thunder for the release of the new Longhorn version of Windows at the end of 2006.

But the question one keeps returning to is: are 3D interfaces actually better? The evidence is conflicting. A study at the University of Leeds by Nikos Drakos in 1996 suggested that some people could quite like a 3D interface. But a more rigorous test in 2001 by the University of Maryland's Student Human-Computer Interaction Online Research Experiments (Shore) put the interfaces against each other. As reported at www.otal.umd.edu/SHORE2001/winDesktop/, they set one group using Windows Explorer (the common way of navigating files on Windows) against another using ClockWise Win3D, a 3D interface to Windows that has "rooms" in which the user can turn around.

Given 12 tasks to complete (such as finding a particular spreadsheet), the users of the two-dimensional Explorer were significantly faster than those using ClockWise. This was for a group ranging from novices to experts. Arguably, the difference came because the subjects were unused to the 3D environment. So, at the end, the authors, who were by now expert on the 3D system, tested themselves - and found a similar difference: they were still often quicker with the 2D Explorer than the 3D ClockWise.

Some of that must be due to familiarity; as the researchers point out, one comes across so many 2D interfaces. Partly, too, the tools we use limit us; given a joystick, navigating around 3D spaces becomes much simpler. But with a billion PCs sold and the next billion chugging out of factories, we seem in danger of never realising any of the potential of 3D. Which is a pity, because I'm still haunted by that twister of data conjured up all those years ago. And it would be a good way to keep my desk clear.

Steve Benford's work: www.crg.cs.nott.ac.uk/~sdb/

www.charlesarthur.com/blog