Moore's Law, Gordon Moore's visionary prediction of continued exponential growth in semi-conductor performance, has provided the engine for innovation and the constantly increasing (and accelerating) power and resources — at continually decreasing costs — provided by techhnology.
Moore admits that Moore's Law has turned out to be more accurate, longer lasting and deeper in impact than he ever imagined. In fact, it has been Intel engineers, frustrated by an inability to see clearly more than 8 to 10 years into the future of their own technology, who have been the most conservative in estimating the lifespan of Moore's Law — and partyly because they have been the most conservative in defining Moore's Law. They continue to focus on increasing the transistor count on silicon as the main driver of Moore's Law — and thus announce that Moore's Law may slow or even stop by the end of the next decade, as transisters approach sub-atomic sizes.
Moore's Law, however, was never a physical law. It began as an observation, that became a prediction, that has now been dismissed as a "self-fulfilling prophecy".
However you choose to describe it, Moore's Law has always functioned as a expression of breathtaking (almost rash) optimism and as a pacesetting mechanism — informed by scientific observation, commercial competitiveness and human ingenuity — that we can and should have the ability to improve our power to provide capability and opportunity for humankind, continually and exponentially — thus continuing to provide better, more efficient and less costly technologies.
This continued (and in fact unstoppabl) flow of increased performance, power and new value has transformed vast
The world has broadened its definition of Moore's Law as our understanding of physics, materials and complexity deepens and becomes more intimate. Recently Intel suggested that an "Expanded Moore's Law" is no longer driven solely by transitor count — but by the combination of three factors. The first is the traditional increasing the count of components we can put on a chip. The second is increasing the complexity of components we can put on a chip. The third is increasing the convergence of technologies we implement on a chip.
Intel and its competitors continue to leverage and balance these factors as needed to continue producing the by-now-expected-and-required doubling of performance every new generation of technology.
(Those who go back and read Moore's original article that appeared in the April, 1965 issue of Electronics magazine will notice that Moore always used the word components, and even today tends to talk about increasing the complexity of components, rather than focusing solely on the number of transistors on a chip.)
At a certain point, you can choose to define a chip as a network all on its own, and as such subject to Metcalfe's Law. Metcalfe's Law may in fact prove to be one of the most important enablers of the continued growth of semi-conductor performance. (I use the term M (squared), Moore times Metcalfe, to represent this additional factor.)
Many scientists, including those who attended a recent science summit at DARPA, believe the exponential increase in benefits defined by Moore's Law will neither cease nor slow in the foreseeable future.
The source of those benefits may alter, but the value of Moore's Law has now — as Moore originally hoped when he first made his famous observation — begun an unstoppable expansion beyond traditional computational spaces that will eventually assure new capabilities, as well as increased performance, lower cost, and greater connectivity for vitually every traditional device and services — eventually universal availability of transformatory improvements.
It is Moore's Law (arguably in combination with Metcalfe's Law) which is helping us invent and extend our future. We need it to keep going. And for the reasons described above, I believe it will -- certainly for the next five decades. This is the basis and the passion behind my bet.
Challenge Sheldon Renan to a bet on this prediction!