Exponential Trends Cut both Ways
When an exponential trend (like the improvement in the price-to-performance ratio of computers) is going your way, everything’s wonderful. But when improvement stops being exponential and becomes merely linear (as all trends eventually do) then things get very bad.
Ray Kurzweil claims that in the next 10 years we will see 20,000 years of technological progress. Kurzweil comes to this number by taking the exponential of 10 which equals 22,026. This, of course, is nonsense, because technology, broadly measured, is not exponentially accelerating (see here). Kurzweil is referring to Moore’s Law which has accurately described the exponential acceleration of computers for some time now. If Moore’s Law (the doubling of computer power every year) continues then we will have about 20,000 years of improvement (at this year’s rate) in the next 10 years.
However…
If Moore’s Law ends, then it will take 20,000 years at today’s rate of improvement to get to where you thought you would be in 10 years.
This is why Kurzweil’s religious fantasies won’t come true. The improvement in the price-to-performance ratio of computers will plateau just like all technologies have and since computers will no longer improve exponentially; we will be thousands of years away from where the Transhumanists thought we would be in just decades.
Exponential trends cut both ways.
Are we completely ignoring the concept of "paradigm shift" and how plateaus just increase the pressure for these shifts?
Reply to this
I agree. Transistors weren't the first or last technology paradigm which experienced said exponential growth, and they won't be the last. Recently Intel released a new line of 45nm chips, with double the transistor density of their 65nm chips. They also used a new substrate for their transistors to reduce power consumption and heat. This is an advance still being made utilizing the current paradigm, silicon based and transistor-centric. Research involving Nano-scale production methods is investigating printing circuitry at the nanoscale, like with an ink-jet printer. When we do hit a wall using current production methods, we will utilize new ones, cost effective and more efficient.
Reply to this