The Pentium 4, in spite of its somewhat lackluster performance, is likely to be remembered as the last true desktop CPU. It was the last CPU for which the requirements presented to the engineering team were to make it run as fast as possible—and damn the power consumption.
Now, when Intel and AMD advertise new CPUs, they make claims about power consumption, not raw speed.
Since the 1GHz barrier was broken, desktop CPUs have been fast enough for a lot of users. Laptop CPUs and even GPUs are fast enough for a lot of people not to use desktops (and for some desktops, like the Mac Mini, to be built from laptop components).
The very low-power components found in PDAs are catching up. What will computing look like when a mobile phone has as much processing power as a modern computer?
The Ideal Form Factor
There are a few simple rules that, baring disruptive technologies, work well for predicting the evolution of a device. The best is that every device converges to its ideal form factor. Components that are not directly related to use interface disappear.
Consider an early PC. It had a screen, a large box containing the CPU and disk(s), and a keyboard. It might have also had a mouse.
Using this rule, we can immediately see that the CPU box would eventually vanish; the user doesn’t interact with it, and so it’s not needed. This gives us a monitor and keyboard (something, in fact, not too dissimilar to the original 1984 Mac).
The monitor is roughly cubic, but only the front is important, and so we can predict that it will shrink. Now we are left with a flat-screen connected to a keyboard and mouse. The wires from the keyboard and mouse to the screen add nothing to the user experience, so they can go, too. We are left with a thin screen, and a wireless keyboard and mouse.
This prediction could have been made in the 1980s. Some of the technologies currently used to build such a machine, like Bluetooth and TFT displays, might not have been available, but the prediction is not dependent on them.
Taking this further, what is the perfect form factor for a computer? A screen that takes up the entire wall would be a good start, or even allows any flat surface to be used. I saw a system like this at Eurographics back in 2004. It had a camera attached to a projector and warped the projected image to counter the warping caused by an uneven projection surface.
While the technology was still very expensive then, it will probably be cheap enough to be mass-produced in a few years.
Many years ago, when I was learning to type, I thought speech recognition would replace keyboards in the near future. Speech recognition is now almost good enough for dictation, but still isn’t popular. Part of the reason why is because it’s very bad for entering commands and partly because learning to dictate well is almost as hard as learning to type well.
Keyboards can be made smaller for portable use by folding them up. Projection keyboards have been shown to be inferior to physical ones because they lack tactile feedback. They could be improved by some kind of haptic device, but this would be likely to encourage RSI.
In summary, keyboards are about as good as they are likely to get, and will probably not change much until someone comes up with a radically better input device.
Having removed most of the computer, we are left with something projecting images anywhere we want them in the room (maybe onto walls or desks, maybe into 3D space). The keyboard becomes wireless, as does the mouse (possibly with the addition of detecting touches to display surfaces).
This is really great while we are in the room, but what happens when we move?