How much longer can the PC continue to grow by acquiring extensions and modifications piled on top of its original design? At what will point the weight of these additional layers, dictated by compromise between the past and the future, finally overcome the concern for backward compatibility? And if today's design is eventually replaced, what will its successor look like?

I don’t claim to know the final answer to such broad questions. I remember too well past predictions of mine that were, in retrospect, either laughable or, in the best cases, embarrassing, or even, in the worst cases, costly. But the question remains [and deserves to be examined].

Let's start by examining the notion that there may one day be a clean break between older designs and technological advancements.

The old and the new have always peacefully co-existed on the PC since its introduction in 1981. Look at the back of a PC today and you'll see the past (serial and parallel ports) alongside the present (USB connectors) and the future (FireWire, also known under the sweet moniker IE 1394). Open the case, and you'll find good ol' 16 bit ISA slots living in harmony with PCI slots for cards that require faster throughput and 32 bit addressing. (Earlier 8 bit ISA connectors are more rare, but 8 bit cards often work in 16 bit slots).

The same applies to processors. The Pentium II processor is an incredibly complex piece of machinery. Its genetic makeup contains the traces of noble ancestors that are appeared over twenty years ago- the 8008, 8080, 80286, 386 and 486… Just like the PC it helps come to life, the Pentium is the result of successive and clever compromises between past and future. The price of these compromises is more apparent when comparing the Pentium and the PowerPC. At equal circuit miniaturization (.25 micron) and complexity (that is, at equal number of transistors), the PowerPC offers more processing power. Or, in other terms, at equal processing power, the PowerPC is smaller and consumes less energy. Introduced in the early 90s, the PowerPC's design is more modern and does not have to accommodate earlier technologies. Of course, economies of scale due to production volumes well in excess of PowerPC levels, and the huge installed base of x86 software, help the Pentium make up part of its handicap.

Windows itself is the result of a similar evolution, one that parallels that of the processor, as the last four digits of Microsoft's phone number, 8080, suggest. Microsoft claims that there will be no evolutionary successor to Windows 98. The official party line predicts that the Windows 9x and NT lines will merge into a consumer version of NT 5.0. NT 5.0 beta itself is a chubby newborn and already weighs in at 35-50 million lines of source code. DOS weighed in at several tens thousand of lines, and Windows, at its beginnings, at several hundred thousand lines. In other words, at the height of its power, the personal computer has become a [sprawling, heavy-duty] industrial processing plant.

Any structure inherently holds a built-in limitation that makes it impossible for it to grow over a certain size. Atomic nuclei spontaneously decompose when reaching a certain stage of the periodic table (somewhere around the uranides, as I remember). The skeletal framework of the human body can only support a limited load. It is loosely related to a person's height (the load capacity of a human bone is coupled to the cube of the person's height). An unrealistically large femur bone would be required to support a 9ft human. Even the size of ants, contrary to what sci-fi novels may want to make you believe, is limited by the structure and therefore the load capacity of their chitin shell.

PCs also have structural limitations. The load capacity of a computer can be determined, for example, by the instruction set of its processor. 16 bit architectures can address up to 64K (16 x 1024). 32 bit processors were introduced to handle the larger amounts of memory made possible by technological advances and falling costs. But these 32 bit processors keep traces, sometimes more cumbersome than a coccyx or an appendix, of earlier 8 or 16 bit architectures. This scenario is replicated for each organ in the computer. Some will say that our genetic code is itself holds plenty of bugs and unnecessary sequences. This has not stopped human beings from walking upright and developing a thumb and a cortex. Therefore the increased complexity of microprocessors and chipsets, legacy designs and unavoidable bugs should not hinder the PC's evolution either. In both cases, ecological or commercial survival is determined by random selection ad necessity. On the other hand, even though semi-conductor processing power doubles every 18 months, this seems hardly comparable to the time frame (eons) biological evolution is measured in.

But the biological metaphor is tempting in other regards. The PC is a life-form. The user is a creator, a deus ex machina, the agent behind both the selective randomness and necessity that drives its evolution. Randomness is a result of the haphazard way we use computers. Necessity is determined by the requirement that PCs continuously generate more with less. In this environment, increased productivity is the basis for survival.

So what does this tell us about the future of the PC?

Well, if you look at the PC as nothing more than a machine to run Microsoft Office, then evolution can lead to lower prices, and that would be fine.

But the future of the PC may lie in real-time simulation and knowledge acquisition, as a machine of ever finer and faster performance that will take us beyond our immediate understanding, on a voyage to the center of the atom or to the frontiers of the galaxy. And this requires new architectures and life-forms vastly different from those we now know.

Let's remember that mini-computers were architecturally and structurally different from the larger mainframes whose experience they nevertheless inherited. The same is true of the PC. It started with modest aims. Certainly, history is rarely linear and without dead-ends. But it is tempting to think that today's game console or PDA may take us into after-tomorrow's cyberspace.

The question then is, when?