Microsoft Word Kurzweil, Ray The Singularity Is Near doc

in three dimensions
, which I discuss in the next chapter. 
We might ask whether there are inherent limits to the capacity of matter and energy to support computational 
processes. This is an important issue, but as we will see in the next chapter, we won't approach those limits until late in 
this century. It is important to distinguish between the S-curve that is characteristic of any specific technological 
paradigm and the continuing exponential growth that is characteristic of the ongoing evolutionary process within a 
broad area of technology, such as computation. Specific paradigms, such as Moore's Law, do ultimately reach levels at 

---

which exponential growth is no longer feasible. But the growth of computation supersedes any of its underlying 
paradigms and is for present purposes an ongoing exponential. 
In accordance with the law of accelerating returns, paradigm shift (also called innovation) turns the S-curve of any 
specific paradigm into a continuing exponential. A new paradigm, such as three-dimensional circuits, takes over when 
the old paradigm approaches its natural limit, which has already happened at least four times in the history of 
computation. In such nonhuman species as apes, the mastery of a tool-making or -using skill by each animal is 
characterized by an S-shaped learning curve that ends abruptly; human-created technology, in contrast, has followed 
an exponential pattern of growth and acceleration since its inception. 
DNA Sequencing, Memory, Communications, the Internet, and Miniaturization 
Civilization advances by extending the number of important operations which we can perform 
without thinking about them. 
—A
LFRED 
N
ORTH 
W
HITEHEAD
,
1911
42
Things are more like they are now than they ever were before. 
—D
WIGHT 
D.
E
ISENHOWER
The law of accelerating returns applies to all of technology, indeed to any evolutionary process. It can be charted with 
remarkable precision in information-based technologies because we have well-defined indexes (for example, 
calculations per second per dollar, or calculations per second per gram) to measure them. There are a great many 
examples of the exponential growth A implied by the law of accelerating returns, in areas as varied as electronics of all 
kinds, DNA sequencing, communications, brain scanning, brain reverse engineering, the size and scope of human 
knowledge, and the rapidly shrinking size of technology. The latter trend is directly related to the emergence of 
nanotechnology. 
The future GNR (Genetics, Nanotechnology, Robotics) age (see chapter 5) will come about not from the 
exponential explosion of computation alone but rather from the interplay and myriad synergies that will result from 
multiple intertwined technological advances. As every point on the exponential-growth curves underlying this panoply 
of technologies represents an intense human drama of innovation and competition, we must consider it remarkable that 
these chaotic processes result in such smooth and predictable exponential trends. This is not a coincidence but is an 
inherent feature of evolutionary processes. 
When the human-genome scan got under way in 1990 critics pointed out that given the speed with which the 
genome could then be scanned, it would take thousands of years to finish the project. Yet the fifteen-year project was 
completed slightly ahead of schedule, with a first draft in 2003.
43
The cost of DNA sequencing came down from about 
ten dollars per base pair in 1990 to a couple of pennies in 2004 and is rapidly continuing to fall (see the figure 
below).
44 

---

There has been smooth exponential growth in the amount of DNA sequence data that has been collected (see the 
figure below).
45
A dramatic recent example of this improving capacity was the sequencing of the SARS virus, which 
took only thirty-one days from the identification of the virus, compared to more than fifteen years for the HIV virus.
46

---

Of course, we expect to see exponential growth in electronic memories such as RAM. But note how the trend on 
this logarithmic graph (below) proceeds smoothly through different technology paradigms: vacuum tube to discrete 
transistor to integrated circuit.
47 

---

However, growth in the price-performance of magnetic (disk-drive) memory is not a result of Moore's Law. This 
exponential trend reflects the squeezing of data onto a magnetic substrate, rather than transistors onto an integrated 
circuit, a completely different technical challenge pursued by different engineers and different companies.
48 

---

Exponential growth in communications technology (measures for communicating information; see the figure 
below) has for many years been even more explosive than in processing or memory measures of computation and is no 
less significant in its implications. Again, this progression involves far more than just shrinking transistors on an 
integrated circuit but includes accelerating advances in fiber optics, optical switching, electromagnetic technologies, 
and other factors.
49 
We are currently moving away from the tangle of wires in our cities and in our daily lives through wireless 
communication, the power of which is doubling every ten to eleven months (see the figure below). 

---

The figures below show the overall growth of the Internet based on the number of hosts (Web-server computers). 
These two charts plot the same data, but one is on a logarithmic axis and the other is linear. As has been discussed, 
while technology progresses exponentially, we experience it in the linear domain. From the perspective of most 
observers, nothing was happening in this area until the mid-1990s, when seemingly out of nowhere the World Wide 
Web and e-mail exploded into view. But the emergence of the Internet into a worldwide phenomenon was readily 
predictable by examining exponential trend data in the early 1980s from the ARPANET, predecessor to the Intemet.
50 

---

This figure shows the same data on a linear scale.
51 

---

In addition to servers, the actual data traffic on the Internet has also doubled every year.
52

---

To accommodate this exponential growth, the data transmission speed of the Internet backbone (as represented by 
the fastest announced backbone communication channels actually used for the Internet) has itself grown exponentially. 
Note that in the figure "Internet Backbone Bandwidth" below, we can actually see the progression of S-curves: the 
acceleration fostered by a new paradigm, followed by a leveling off as the paradigm runs out of steam, followed by 
renewed acceleration through paradigm shift.
53

---

Another trend that will have profound implications for the twenty-first century is the pervasive movement toward 
miniaturization. The key feature sizes of a broad range of technologies, both electronic and mechanical, are 
decreasing, and at an exponential rate. At present, we are shrinking technology by a factor of about four per linear 
dimension per decade. This miniaturization is a driving force behind Moore's Law, but it's also reflected in the size of 
all electronic systems—for example, magnetic storage. We also see this decrease in the size of mechanical devices, as 
the figure on the size of mechanical devices illustrates.
54 

---

As the salient feature size of a wide range of technologies moves inexorably closer to the multi-nanometer range 
(less than one hundred nanometers—billionths of a meter), it has been accompanied by a rapidly growing interest in 
nanotechnology. Nanotechnology science citations have been increasing significantly over the past decade, as noted in 
the figure below.
55

---

We see the same phenomenon in nanotechnology-related patents (below).
56 

---

As we will explore in chapter 5, the genetics (or biotechnology) revolution is bringing the information revolution, 
with its exponentially increasing capacity and price-performance, to the field of biology. Similarly, the 
nanotechnology revolution will bring the rapidly increasing mastery of information to materials and mechanical 
systems. The robotics (or "strong AI") revolution involves the reverse engineering of the human brain, which means 
coming to understand human intelligence in information terms and then combining the resulting insights with 
increasingly powerful computational platforms. Thus, all three of the overlapping transformations—genetics, 
nanotechnology, and robotics—that will dominate the first half of this century represent different facets of the 
information revolution. 

---