I'm sure most of us are familiar with Moore's Law, the observation that the transistor density of circuits doubles every 24 months
One thing that I have always been a little curious about with Moore's Law is exactly how long it can go on for, surely we can only etch on integrated circuits down to a certain size before further miniturisation becomes impossible. One clue for how Moore's law can continue far into the future comes from this wonderful graph
in which Ray Kurzweil has expanded the standard Moore's law plot back in time by considering all sorts of different technologies. As each technology is pushed to its limits a new one replaces it, allowing more growth. In this picture Moore's law is actually the fifth in a series of exponential computer growth laws.
I have no idea how long computational power can continue to grow at its present rate using integrated circuits, or whether or not a sixth technology will be found that can replace them (see e.g. timeline of quantum computing). Perhaps we will never find another significant improvement, and be stuck with slightly improved versions of current technology forever.
As an interesting aside there was a paper on astro-ph sometime last year (here) about the theoretical maximum size of a computer, taking into account all of the information available in the Universe. Turns out that Moore's Law can hold for an absolute maximum of 600 more years.
late edit: interesting note about quantum computing in the comments
One thing that I have always been a little curious about with Moore's Law is exactly how long it can go on for, surely we can only etch on integrated circuits down to a certain size before further miniturisation becomes impossible. One clue for how Moore's law can continue far into the future comes from this wonderful graph
in which Ray Kurzweil has expanded the standard Moore's law plot back in time by considering all sorts of different technologies. As each technology is pushed to its limits a new one replaces it, allowing more growth. In this picture Moore's law is actually the fifth in a series of exponential computer growth laws.
I have no idea how long computational power can continue to grow at its present rate using integrated circuits, or whether or not a sixth technology will be found that can replace them (see e.g. timeline of quantum computing). Perhaps we will never find another significant improvement, and be stuck with slightly improved versions of current technology forever.
As an interesting aside there was a paper on astro-ph sometime last year (here) about the theoretical maximum size of a computer, taking into account all of the information available in the Universe. Turns out that Moore's Law can hold for an absolute maximum of 600 more years.
late edit: interesting note about quantum computing in the comments
