My latest Mind and Matter column in the Wall Street Journal:
In 1965, the computer expert Gordon Moore published his famous little graph showing that the number of
“components per integrated function” on a silicon chip-a measure of
computing power-seemed to be doubling every year and a half. He had
only five data points, but Moore’s Law has settled into an almost
iron rule of innovation. Why is it so regular?
The technology guru Ray Kurzweil recently pointed out that a version of Moore’s Law has
been true since the early years of the 20th century. That is to
say, before the integrated circuit even existed, the four previous
technologies-electromechanical, relay, vacuum tube and
transistor-had all improved along the very same trajectory: The
computing power that $1,000 buys has doubled every two years for a
century.
A similar graph can be plotted for the number of radio
communications fitting into the electromagnetic spectrum. Ever
since Guglielmo Marconi’s first transmission in 1895, the number of
possible simultaneous wireless communications has doubled every 30
months. (This is now known as Cooper’s Law, after the inventor
Martin Cooper, who demonstrated the first hand-held cellphone.)
Both graphs are roughly exponential, meaning that they curve
rapidly upward (or appear as straight lines if plotted on a
logarithmic scale). Why don’t they move in lurches followed by
stagnations? And why can’t we cheat these laws by jumping
ahead?
What’s remarkable about the extension of these regularities back
in time is that they now appear to have marched imperturbably
through the upheavals of the 20th century without breaking step.
How is it possible that the Great Depression did not slow down
technological progress? Why didn’t the great infusion of technology
spending during World War II accelerate it?
There’s no certain answer. The inevitable, inexorable and
incremental march of technological improvement remains baffling, as
does the steady march of world economic growth at 2% to 5% a year
ever since the 1940s-far steadier than the progress of any
individual country.
A glimmer of explanation can be found in Reed’s Law, named after
the computer scientist David P. Reed. This states that the utility
of large networks increases exponentially with the size of the
network. That is to say, it goes up faster than the number of
participants or the number of possible pairs of participants (which
goes up by Metcalfe’s Law).
The Silicon Valley investor Steve Jurvetson thinks this may
explain the exponential shape of Moore’s and Cooper’s laws-so long
as you substitute “ideas” for participants. In other words,
technology is driving its own progress by steadily expanding its
own capacity to bring ideas together. The implication is that,
short of arresting half the planet’s people, we could not stop the
march of technology even if we wanted to.
This is mainly reassuring, because bad policies can’t prevent
improvement-but also depressing, because good policies can’t
accelerate improvement. The policies and breakthroughs to which we
attach such importance are all but irrelevant on the global scale,
though they can, of course, result in a country missing out on the
benefits or losing its natural “share” of technology or growth to
another country (ask the North Koreans).
A few years ago Mr. Jurvetson added to this menagerie of laws by coining
Rose’s Law of quantum computing, after a Canadian executive in the
field. (Such computers focus on probabilities and exploit the idea
that subatomic particles can exist in multiple states at the same
time.) The law’s prediction of an annual doubling in quantum
computing capacity-at speeds too scary to contemplate-has come to
pass. If quantum computers are actually doing anything practical in
their incomprehensible brains, then quantum computers will soon
make their conventional cousins look primitive.