The frozen UI

Periodically someone on K5 or in other social circles will ask why the graphical user interface is changing so little nowadays (that is, on the desktops of the world at large). I have two unoriginal reasons why, and a little rant on why the freezing of the GUI is not necessarily a bad thing.

The question of why the computer interface isn't changing is regularly asked on IRC, often on Kuro5hin, and over on Edge.Org, Jaron "Virtual Reality" Lanier, practically despairs over it.

The first reason I am asserting is that the physical devices used for the human-computer interface sharply limit the range of motifs, metaphores, and ideas that will work well with it. The computer interface nowadays is constrained to a pointing device, a keyboard, a pair of speakers, and a two-dimensional, usually flat, multicolored screen. Throw in a microphone if you feel like it.

The pointing device comes in several versions, but the rolling mouse seems to be the only version most people will tolerate for long periods of time. Some prefer a trackball, but the remaining options are even less likely to be adopted. The nipple is approaching extinction, and the finger pad is for limited use only. The stakes here are no longer academic; they are ergonomic. That leaves not much space for physical creativity.

The same goes for the keyboard. The layout will be tweaked for years to come to maximize the tolerance people show for the keyboard, and there is an ongoing redesign of the layout of speed keys, but the interface is defined and frozen.

Other input devices are unlikely to make much headway. Speech recognition's only use is to let you keep working when you take that carpal tunnel-mandated break. Prolonged talking into the computer is a pitiful way to wreck your vocal cords. Foot pedals are similarly unattractive. We may learn to like enhanced mice, like the roller-for-a-middle-button type that Logitech sells, or maybe mice with tactile feedback, but these changes would not be revolutionary.

As for the output devices, the speakers are limited in the interactivity they give. We use them mostly for music, partly because canned sounds as computer messages can get very tiresome after a while, and partly for privacy. The screen itself may get more versatile as it becomes lighter, or as LCD technology allows us to look more closely into it, but once again, these changes will not be earthshaking.

This physical configuration allows for huge amounts of creativity, but creativity doesn't quite pay well for the concept I am hawking here, which is mundane computing (MC). MC is what is requiring optimization in the graphical user interface and limiting innovation. MC does not reward clever ideas. It rewards clever ideas that people will use 8 hours a day, and vive le difference. As such, one of the requirements for mundane computing is stricter than otherwise adherence to Kernighan's principle of least surprise. This is a realm of computing that demands the opposite of what visionaries like to do, which is to explore the limits of human imagination. I want the term 'mundane computing' to catch on as an anti-hype agent. By speaking of certain applications as belonging in MC, we liberate them from the neophilic fetishism that infects the industry.

Jaron Lanier's work pioneering virtual reality shows that he is a creative genius. But his ideas simply do not work on the interface front because of the limits of the devices involved. People do not like headmounted displays. People do not like motion sickness. And when it comes to organizing the information they use in conjunction with computers, (unless they do CAD/CAM), people do not need fancy 3d viewing programs.

So, with tweaks here and there, the 'desktop' metaphores for the GUI will prevail (though they already are moving outside what was once obviously a 'desktop' motif). This is not a bad thing. The user interface for the automobile is not changing much nowadays, either. However, there is much to be done in the optimization arena, and for the same reasons.

Because computing has established its place in the modern office, there is a realm of computing that should be segregated from the rest of the field, which I am calling "mundane computing." MC is the second reason why the GUI is not about to change much, nor should it. By Mundane, I do not mean 'boring.' But first of all, I mean 'no longer novel'. MC is every use of computing which has been in the public view long enough, which people employ for long parts of their day, and in which the computer and software are obviously a means to an end, and not an end in themselves. Word processing, spreadsheets, graphic design, email, and many other uses of computers should fall in this realm.

In MC, a user interface must work well, and this 'works well' attribute is enhanced by a previous familiarity on part of the user. So, in MC, there is a barrier to entry by new ideas, (which is of course the other reason) and progress is largely of the form of evolution, not revolution. Hackers should rejoice, however, because dammitall, evolution remains necessary in the graphical user interface. Enough evolution has already happened to show my point. The desktop motif is no longer so close to a real desktop. Remember when you discovered that you could mark a rectangle on the background of a Macintosh and then move all the icons within it in unison? Remember when it was that you realized that was an utterly useless feature? I forgot when I recalled that last one.

From the limited range of hardware used in mundane computing comes a limited range of software ideas represented in the user interface. The first and obvious one to mention is the text area, which is anything in which a piece of text is displayed and can be edited as ASCII or as something more elaborate (HTML, COM, et cetera), and also of course the command line interface. This widget dominates the desktop because it dominates everything two dimensional in our daily lives. Iconographic interfacees are too limiting, and therefore limited in how much they are used on the desktop, and any machine interface in the physical world that is more elaborate tends to have a three dimensional form, and thus cannot easily be brought up on the CRT.

There are other ideas (motifs, widgets, et cetera) that have sprung up in various programs, which are creative but unlikely to be used in mainstream programs. A nifty thing I discovered (and tried to use) is Open Data Explorer. Take the pipe motif from the Unix shells, use simple widgets to represent it graphically, and then make the widgets perform Matlabesque tasks. Nice idea, but I'll stick to Matlab and Octave. This is just one example out of many things that are creative, clever, but unusable for long periods (for most people, anyway). The creative programmer is going to find resistance against his ideas entering into mundane computing.

Since the need for optimal usability for the GUI will drive it to change in the coming years, let's take a look at where it will go: the bone-rattler example for a GUI was something I encountered in a physiology class. It was a simulation written with Xlib in which there were quite a few labeled buttons to 'press'. The target for each button, however, was a 6 by 6 pixel bullet next to the label, rather than a rectangle surrounding both. Mousing needs to be minimized, and therefore targets enlarged. The new Mac trick of treating the mouse cursor like a magnifying glass when appropriate is a step exactly in that direction. Another is the ability to have sloppy focus in most X window managers. (Note that Microsoft is looking at what X window managers do well.) What will be ultimately a limit to changes in the GUI is the need to use only those widgets that represent two-dimensional physical objects. Three dimensionals won't work on a CRT.

But with a slowdown in the evolution of most software, (knock wood, cross fingers, throw salt over left shoulder) the average user will come to expect software to be more stable, more reliable, faster, and contain fewer nasty security surprises. Wouldn't maturity in the software industry be nice?