Monday, March 29, 2010

Why the SIngularity is Probably Nonsense

I've been catching up on my futurist web-reading as of recent, just to keep me in the loop. Prolongevists, transhumanists, Singularity adherents, doomsday sayers, Earth savers, and so on.

Don't know what prompted it; I guess it must have been the unread list-emails relating to such subjects, and especially to the issue of human longevity, that have been accumulating unread in my various dedicated mailboxes.

Anyway, I came across the whole 'Singularity' thing again. I don't know if you're familiar with the subject, but it boils down to this (taken from here, to save me typing):

"Imagine a curve that represents the technological progress of human beings throughout history. Most people would agree that we have come a long way in only a century; we've invented new technology, learned new things and developed as a race. Thus the curve slopes upward. We can examine this curve and extrapolate to create a function that describes the level of progress for any given point in time.

Many have done this, including Ray Kurzweil, the renowned inventor of synthesisers and text-to-speech machines and also author of The Age of Intelligent Machines, a book that won the Association of American Publishers' Award for the Most Outstanding Computer Science Book of 1990. Kurzweil has shown that the doubling period of the speed of computers is diminishing, i.e. it used to take us three years to double the speed and memory capacity of computers in the beginning of the 20th Century, and now the same kind of progress is achieved in only one year. He claims that these trends will continue, and that computers will be able to emulate human brains in the year 2020.


When and if computers become intelligent, computers themselves could construct new computers, causing a massive acceleration in technological progress. One way to visualise this acceleration is to consider the following rather trivial function.


f(t) = -1 / t


Readers with working knowledge of mathematics can see that this function gives us very large values for very small negative values of t. As t approaches zero (from the negative side), the value of the function approaches infinity. This is called hyperbolic growth. A hyperbolic function grows much faster than an exponential function, as the reader can easily see in the following table. Note that t takes on negative values; think of it as a countdown to the Singularity.
"

Yeah, very mathematical and scary for some—but what it boils down to is that we're going to have some point in history, probably soon-ish, where technological progress will run away from us and either completely annihilate us or else solve all our problems. It's represented by the function labeled 'hyperbolic' on this picture:

Those promoting the concept of a Singularity tend to get very enthusiastic about it, as fervents do—and especially about topics they know nothing about, which is anything related to what the future will bring. And, being woolly thinkers, they tend to forget one of the most fundamental lessons from physics: there's always friction.

That's like one of the most fundamental laws of the universe. That, and Newton's 2nd and 3rd Laws, of course, are what's putting the kibosh on the whole Singularity nonsense. The thing is, you might notice, that in the initial parts of the three curves in the picture above, they are hard to tell apart—and just how far that 'initial part' goes is a question of the exact equations governing the processes described. Right now there's nothing to suggest right now that we're talking about a hyperbola, instead of an exponential and maybe even a sigmoid function. Nothing, that is, except people's overheated imaginations.

Indeed, it seems to me that what you might call 'progress friction'—which expresses itself in all sorts of ways, from the purely physical to the conceptually achievable—is already showing its effects, and will do so increasingly as we continue moving along the curve.

And, yes, I may be wrong about that, and the Singularity Believers may be right. But at least I remember the importance of friction; one of the first and most important topics even in the most elementary of physics classes. For friction is the nemesis of singularities.

No comments: