Last week the chip maker Intel paid $10,000 for an original copy of a 40-year-old electronics magazine. It contained an article of faith for the company: Founder Gordon Moore's famous "Law" about computing power.
His wonderfully titled article "Cramming more components onto integrated circuits" predicted that about every year the number would double.
It's held up pretty well with computing power doubling every 18 months: In 1964 a computer chip had about 50 transistors, now they cram half a billion.
For the consumer this means that every year and a half a home computer becomes twice as powerful for the same amount of money. Also, Moore's Law applies to other computer-based devices: Hard drives and cell phones for example.
Chip makers remain sure they can do this yearly "doubling" for about five more years, but after that it's pretty dicey. No technology clearly stands out for continuing this explosive growth of computing power - cheap growth, that is. This means consumers will feel it in the pocket book if Moore's Law stalls.
They'll be stuck with the personal computer of 2010. And if they move, for example, into digital photography and advanced imaging, they'll find no cheap solutions for storing large numbers of photos beyond that age's capability.
And for the investor the center of gravity of the electronics industry will shift - in both hardware and software.
For hardware, advances in computing will come from outside the single-chip computer - the Intels of the world. The companies that thrive will be those that develop more intelligent peripherals, off-loading work from the computer's processor.
And Moore's Law gave us bad software: Instead of writing efficient code, programmers just waited until the computers got faster. I'm reminded of Blaise Pascal's comment three centuries ago. Apologizing for a verbose letters he wrote "I have only made this [letter] longer because I have not had the time to make it shorter."
Letters like software, take time to shorten and condense. So software giants like Microsoft find it better to write "just good enough" programs, that rely on advances by Intel to make them run efficiently. The current giants might be displaced by nimble companies that take the time to write efficient software. And when the age of Moore's Law ends, we may well be telling our children unbelievable tales about how in our day computing got bigger and better every year.
Copyright 2005 William S. Hammack Enterprises
