The Trouble with Moore's Law Determinism

You've probably heard of Moore's Law, which states that the number of transistors that can fit on a computer chip doubles every 18 months to two years, give or take. It has held true since Intel's Gordon Moore made the conjecture in 1965, and though it might not go on forever, the exponential rise in computing power has driven all kinds of technological change. But there's something that's been bugging me for a while in the way people reference Moore's Law, and I figured the new year was as good a time as any to get it off my chest.

The error is in the assumption that a very specific exponential curve regarding change in the power of transistors and circuits is the same as change in technological innovation, which is the same as societal change driven by technology. It isn't.

Perhaps no one is more responsible for this misconception than Ray Kurzweil, the engineer/inventor/technocelebrity who has popularized the notion of the "singularity," the moment when artificial intelligence exceeds human intelligence and, depending on your view, either an unimaginable but nevertheless utopian future is upon us, or we're doomed. I doubt many people who have thought seriously about it doubt that artificial intelligence will exceed human intelligence eventually, and what happens to us when that occurs is a fascinating and complicated question. But along the way, it's a mistake to believe that increasing processing power is the same thing as "change" in technology and particularly in our own lives.

As novelist Nick Harkaway (I highly recommend his recent book Angelmaker) says in this talk, if you just accept that Moore's Law produces "change" in some kind of linear fashion, that means that this year will be twice as changey as last year—certainly possible—but it also means that 2023 will be 1,024 times as changey as 2013, which is kind of ridiculous. Take, for instance, your phone. The smartphone was a big innovation over the ordinary cell phone. Very changey. But once you've got that phone, the fact that your next phone has twice as much memory and a a chip that's twice as fast doesn't mean that it will have twice as big an impact on your life as getting your first smartphone did. Our smartphones can get faster and hold more photos, but it's going to take an innovative leap produced by someone's creative thinking to make them something profoundly different from what they are now.

There are some questions and problems that can be solved with nothing more than brute-force calculation, and once our computers can make a sufficient number of calculations fast enough, the problem will be solved. But many problems just don't work that way, and other problems turn out not to be what we originally thought they were. In some cases, the more we know, the more complexity is revealed to us, as though we increased the power of our microscope only to find that the crystal we were looking at has a structure even more tiny and intricate than we thought, and we're going to need an even stronger microscope to grasp it all.

Because of advancing technology, the world gets more complex all the time (and to me anyway, more interesting). But not everything proceeds on an exponential curve. Is your life today changing 1,024 times as fast as it did ten years ago? Of course not. And it's going to be a little while before change occurs so rapidly that the only way we can grasp it is for the AIs that have taken over to pull us toward a post-human transcendance. Which, by the way, happens to be the title of a new film in which Johnny Depp's brain is uploaded and quickly swallows the Earth. Here's the trailer, which is one of those that gives you pretty much the entire film in two and a half minutes:

Comments

A thing to keep in mind about Moore's Law is that it relies on improvements in one dimension that have two-dimensional effect. That is, the improvements made by chip manufacturers are narrowing the width of connectors, a change in just one dimension. However when you do that, features on the chip can then be smaller in both width and height. Therefore when you cut line width in half, you get four times as many objects on the same chip area.

Very few other things in technology (or life) work that way. Moore's Law is based on a nearly-unique feature of integrated circuit manufacture.

This entire article is a great example of how difficult it is for our linear vision to see exponential growth. Your first statement of "not 1024 times as changey" is actually incorrect, in regards to information technology.
Since you used smart phones as an example, I'd like to point out that a simple web search would reveal that smart phone development has followed the exponential growth pattern reliably, and that's just the hardware side... If you take into consideration the immense expansion of the infrastructure, wireless supports, various apps and utilities that have exploded out of that market, it is quite easy to see how maintained an exponential growth rate. 10 years ago the top selling phone was a flip type with limited text capabilities and a $500 price tag, on top of the 2 year contract.
Even the search engine you use to find this information is another great example of the speed of our development. 10 years ago, a search was available, but hardly common pratice enough to have its own verbage.
Information technology is more than Moore's Law, which is precisely why this article is completely off base and misleading. We ARE 1024 times more "changey", but like watching a tree grow, it is difficult to see day by day.

You need to be logged in to comment.
(If there's one thing we know about comment trolls, it's that they're lazy)

Connect
, after login or registration your account will be connected.