Re: What is the singularity?

From: Eliezer S. Yudkowsky (sentience@pobox.com)
Date: Mon Jul 30 2001 - 10:55:39 MDT


>From a letter to the editor published by Dr. Dobbs' Journal, March 2000:

==
The True Name of the Singularity

Dear DDJ,

In the December 1999 issue, Swaine defines the Singularity as referring
to the rushing pace of change, the asymptotic graphs of processing power
and such, and the resulting "massive failure in the art of predicting
the future". That's not the historical meaning of the Singularity,
although nowadays it's often used that way. There's a fascinating story
about how "singularity" started by describing a mathematical function
going to infinity, then the center of a black hole, then a breakdown in
our ability to understand the future, and then infinities again.

The first usage of the term "Singularity" in the domain of futurism was
by Vernor Vinge, referring to the difficulty of understanding a future
which contains beings more intelligent than the author. "Here I had
tried a straightforward extrapolation of technology, and found myself
precipitated over an abyss. It's a problem we face every time we
consider the creation of intelligences greater than our own. When this
happens, human history will have reached a kind of singularity - a place
where extrapolation breaks down and new models must be applied - and the
world will pass beyond our understanding." (True Names and Other
Dangers, p. 47.) The analogy was to the singularity at the center of a
black hole, where our models of the laws of physics break down; in turn,
the center of a black hole is called a "singularity" because of the
asymptotically infinite gravitational forces, producing a discontinuity
in the curvature of space.

Once Vernor Vinge invented the term, others tried to calculate the
advent of the Singularity. The most famous projection matched Moore's
Law to an estimate of the raw processing power required by human
intelligence, coming up with an estimate of 2035 (now obsolete, on both
counts). Other famous graphs included the time when materials science
would reach the level of individual atoms, alleged to be in 2040 (yeah,
right! Have you read the news lately?), and the time when human
lifespan would be projected to increase at the rate of more than one
year per year.

Meanwhile, others were asking what would happen after the Singularity.
As Vernor Vinge's Hugo-winner "A Fire Upon the Deep" shows in the
opening pages, intelligence-enhanced minds are more effective at
enhancing intelligence. Some, including myself, half-seriously
speculated that the function goes to infinity. Consider: If computing
speeds double every two years, what happens when computer-based AIs are
doing the research? Two years... one year... six months... three
months...

The graph-projection and post-Singularity-infinity concepts then sort of
heterodyned, to form the concept of those futuristic graphs which happen
to go to infinity on any grounds whatsoever. World population is
hyperbolic to a much better fit than exponential, going to infinity in
2029. This has nothing to do with intelligence enhancement, but, being
a mathematical singularity, a futuristic graph projection, a
discontinuity, and an infinity, it was easily confused with the Vingean
Singularity. And thus it all came full circle. But what "Singularity"
really means is the rise of greater-than-human intelligence, and the
breakdown of our predictions as a result.

Eliezer S. Yudkowsky,
sentience@pobox.com
==

-- -- -- -- --
Eliezer S. Yudkowsky http://singinst.org/
Research Fellow, Singularity Institute for Artificial Intelligence



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:09:17 MST