• If this sum is greater than the firing threshold of this neuron, then this neuron is considered to

"fire" and its output is 1. Otherwise, its output is 0 (see variations below).

Do the following for each recognition trial:

For each layer, from layers to layerM:

And for each neuron in each layer:

• Sum its weighted inputs (each weighted input = the output of the other neuron [or initial input]

that the input to this neuron is connected to, multiplied by the synaptic strength of that

connection).

• If this sum of weighted inputs is greater than the firing threshold for this neuron, set the output

of this neuron = 1, otherwise set it to 0.

To Train the Neural Net

• Run repeated recognition trials on sample problems.

• After each trial, adjust the synaptic strengths of all the interneuronal connections to improve the

performance of the neural net on this trial (see the discussion below on how to do this).

• Continue this training until the accuracy rate of the neural net is no longer improving (i.e.,

reaches an asymptote).

Key Design Decisions

In the simple schema above, the designer of this neural net algorithm needs to determine at the outset:

• What the input numbers represent.

• The number of layers of neurons.

• The number of neurons in each layer (each layer does not necessarily need to have the same

number of neurons).

• The number of inputs to each neuron, in each layer. The number of inputs (i.e., interneuronal

connections) can also vary from neuron to neuron, and from layer to layer.

• The actual "wiring" (i.e., the connections). For each neuron, in each layer, this consists of a list

of other neurons, the outputs of which constitute the inputs to this neuron. This represents a key

design area. There are a number of possible ways to do this:

(i) wire the neural net randomly; or

(ii) use an evolutionary algorithm (see next section of this Appendix) to determine an optimal

wiring; or

(iii) use the system designer's best judgment in determining the wiring.

• The initial synaptic strengths (i.e., weights) of each connection. There are a number of possible

ways to do this:

(i) set the synaptic strengths to the same value; or

(ii) set the synaptic strengths to different random values; or

(iii) use an evolutionary algorithm to determine an optimal set of initial values; or

(iv) use the system designer's best judgment in determining the initial values.

• The firing threshold of each neuron.

• Determine the output. The output can be:

(i) the outputs of layer sub M of neurons; or

(ii) the output of a single output neuron, whose inputs are the outputs of the neurons in layer

sub M;

(iii) a function of (e.g., a sum of) the outputs of the neurons in layer sub M; or

(iv) another function of neuron outputs in multiple layers.

• Determine how the synaptic strengths of all the connections are adjusted during the training of

this neural net. This is a key design decision and the subject of a great deal of neural net

research and discussion. There are a number of possible ways to do this:

(i) For each recognition trial, increment or decrement each synaptic strength by a (generally

small) fixed amount so that the neural net's output more closely matches the correct answer.

One way to do this is to try both incrementing and decrementing and see which has the more

desirable effect. This can be time consuming, so other methods exist for making local

decisions on whether to increment or decrement each synaptic strength.

(ii) Other statistical methods exist for modifying the synaptic strengths after each recognition trial

so that the performance of the neural net on that trial more closely matches the correct

answer.

Note that neural net training will work even if the answers to the training trials are not all correct. This

allows using real-world training data that may have an inherent error rate. One key to the success of a neural

net-based recognition system is the amount of data used for training. Usually a very substantial amount is

needed to obtain satisfactory results. Just like human students, the amount of time that a neural net spends

learning its lessons is a key factor in its performance.

Variations

Many variations of the above are feasible. Some variations include:

• There are different ways of determining the topology, as described above. In particular, the

interneuronal wiring can be set either randomly or using an evolutionary algorithm.

• There are different ways of setting the initial synaptic strengths, as described above.

• The inputs to the neurons in layer sub i do not necessarily need to come from the outputs of the

neurons in layer sub i minus 1. Alternatively, the inputs to the neurons in each layer can come

from any lower layer or any layer.

• There are different ways to determine the final output, as described above.

• For each neuron, the method described above compares the sum of the weighted inputs to the

threshold for that neuron. If the threshold is exceeded, the neuron fires and its output is 1.

Otherwise, its output is 0. This "all or nothing" firing is called a nonlinearity. There are other

nonlinear functions that can be used. Commonly a function is used that goes from 0 to 1 in a

rapid but more gradual fashion (than all or nothing). Also, the outputs can be numbers other

than 0 and 1.

• The different methods for adjusting the synaptic strengths during training, briefly described

above, represent a key design decision.

• The above schema describes, a "synchronous" neural net, in which each recognition trial

proceeds by computing the outputs of each layer, starting with layer sub O through layer sub M.

In a true parallel system, in which each neuron is operating independently of the others, the

neurons can operate asynchronously (i.e., independently). In an asynchronous approach, each

neuron is constantly scanning its inputs and fires (i.e., changes its output from 0 to 1) whenever

the sum of its weighted inputs exceeds its threshold (or, alternatively, using another nonlinear

output function).

Happy Adaptation!

EVOLUTIONARY ALGORITHMS

If biologists have ignored self‐organization, it is not because self‐ordering is not pervasive and profound. It is because we biologists have yet to understand how to think about systems governed simultaneously by two sources of order.

Yet who seeing the snowflake, who seeing simple lipid molecules cast adrift in water forming themselves into cell‐like hollow lipid vesicles, who seeing the potential for the crystallization of life in swarms of reacting molecules, who seeing the stunning order in networks linking tens upon tens of thousands of variables, can fail to entertain a central thought: if ever we are to attain a final theory in biology, we will surely have to understand the commingling of self-organization and selection. We will have to see that we are the natural expressions of a deeper order. Ultimately, we will discover in our creation myth that we are expected after all.

—Stuart Kauffman

As I discussed earlier, an evolutionary algorithm involves a simulated environment in which simulated software ʺcreaturesʺ compete for survival and the right to reproduce. Each software creature represents a possible solution to a problem encoded in its digital ʺDNA.ʺ

The creatures allowed to survive and reproduce into the next generation are the ones that do a better job of solving the problem. Evolutionary algorithms are considered to be part of a class of ʺemergentʺ methods because the

solutions emerge gradually and usually cannot be predicted by the designers of the system. Evolutionary algorithms

are particularly powerful when they are combined with our other paradigms. Here is a unique way of combining all

of our ʺintelligentʺ paradigms.

Combining All Three Paradigms

The human genome contains three billion rungs of base pairs, which equals six billion bits of data. With a little data compression, your genetic code will fit on a single CD‐ROM. You can store your whole family on a DVD (digital video disc). But your brain has 100 trillion ʺwires,ʺ which would require about 3,000 trillion bits to represent. How did the mere 12 billion bits of data in your chromosomes (with contemporary estimates indicating that only 3 percent of that is active) designate the wiring of your brain, which constitutes about a quarter million times more information?

Obviously the genetic code does not specify the exact wiring. I said earlier that we can wire a neural net randomly

and obtain satisfactory results. Thatʹs true, but there is a better way to do it, and that is to use evolution. I am not referring to the billions of years of evolution that produced the human brain. I am referring to the months of evolution that go on during gestation and early childhood. Early in our lives, our interneuronal connections are engaged in a fight for survival. Those that make better sense of the world survive. By late childhood, these connections become relatively fixed, which is why it is worthwhile exposing babies and young children to a stimulating environment. Otherwise, this evolutionary process runs out of real‐world chaos from which to draw inspiration.

We can do the same thing with our synthetic neural nets: use an evolutionary algorithm to determine the optimal

wiring. This is exactly what the Kyoto Advanced Telecommunications Research Labʹs ambitious brain‐building project is doing.

Now hereʹs how you can intelligently solve a challenging problem using all three paradigms. First, carefully state

your problem. This is actually the hardest step. Most people try to solve problems without bothering to understand

what the problem is all about. Next, analyze the logical contours of your problem recursively by searching through as many combinations of elements (for example, moves in a game, steps in a solution) that you and your computer have

the patience to sort through. For the terminal leaves of this recursive expansion of possible solutions,. evaluate them with a neural net. For the optimal topology of your neural net, determine this using an evolutionary algorithm. And if all of this doesnʹt work, then you have a difficult problem, indeed.

"PSEUDO CODE" FOR THE

EVOLUTIONARY ALGORITHM

Here is the basic schema for an evolutionary algorithm. Many variations are possible, and the designer of the

system needs to provide certain critical parameters and methods, detailed below.

THE EVOLUTIONARY ALGORITHM

Create N solution "creatures" Each one has:

• A genetic code—a sequence of numbers that characterizes a possible solution to the problem.

The numbers can represent critical parameters, steps to a solution, rules, etc.

For each generation of evolution, do the following:

• Do the following for each of the N solution creatures:

(i) Apply this solution creature's solution (as represented by its genetic code) to the problem, or

simulated environment.

(ii) Rate the solution.

• Pick the L solution creatures with the highest ratings to survive into the next generation.

• Eliminate the (N - L) nonsurviving solution creatures.

• Create (N - L) new solution creatures from the L surviving solution creatures by:

(i) making copies of the L surviving creatures. Introduce small random variations into each

copy; or

(ii) create additional solution creatures by combining parts of the genetic code (using "sexual"

reproduction, or otherwise combining portions of the chromosomes) from the L surviving

creatures; or

(iii) doing a combination of (i) and (ii) above.

• Determine whether or not to continue evolving:

Improvement = (highest rating in this generation) - (highest rating in the previous generation)

If improvement < Improvement Threshold, then we're done

• The Solution Creature with the highest rating from the last generation of evolution has

the best solution. Apply the solution defined by its genetic code to the problem.

Key Design Decisions

In the simple schema above, the designer of this evolutionary algorithm needs to determine at the outset:

• Key parameters:

N

L

Improvement Threshold

• What the numbers in the genetic code represent and how the solution is computed from the

genetic code.

• A method for determining the N solution creatures in the first generation. In general, these need

only be "reasonable" attempts at a solution. If these first-generation solutions are too far afield,

the evolutionary algorithm may have difficulty converging on a good solution. It is often

worthwhile to create the initial solution creatures in such a way that they are reasonably diverse.

This will help prevent the evolutionary process from just finding a "locally" optimal solution.

• How the solutions are rated.

• How the surviving solution creatures reproduce.

Variations

Many variations of the above are feasible. Some variations include:

• There does not need to be a fixed number of surviving solution creatures (i.e., "L") from each

generation. The survival rule(s) can allow for a variable number of survivors.

• There does not need to be a fixed number of new solution creatures created in each generation

(i.e., [N - L]). The procreation rules can be independent of the size of the population. Procreation

can be related to survival, thereby allowing the fittest solution creatures to procreate the most.

• The decision as to whether or not to continue evolving can be varied. It can consider more than

just the highest-rated solution creature from the most recent generations). It can also consider a

trend that goes beyond just the last two generations.

Happy Evolving!

GLOSSARY

Aaron A computerized robot (and associated software), designed by Harold Cohen, that creates original drawings and paintings.

Alexanderʹs solution A term referring to Alexander the Greatʹs slicing of the Gordian knot with his sword. A reference to solving an insoluble problem with decisive yet unexpected and indirect means.

Algorithm A sequence of rules and instructions that describes a procedure to solve a problem. A computer program expresses one or more algorithms in a manner understandable by a computer.

Alu A meaningless sequence of 300 nucleotide letters that occurs 300,000 times in the human genome.

Analog A quantity that is continuously varying, as opposed to varying in discrete steps. Most phenomena in the natural world are analog. When we measure and give them a numeric value, we digitize them. The human brain

uses both digital and analog computation.

Analytical Engine The first programmable computer, created in the 1840s by Charles Babbage and Ada Lovelace.

The Analytical Engine had a random access memory (RAM) consisting of one thousand words of fifty decimal digits each, a central processing unit, a special storage unit for software, and a printer. Although it foreshadowed modem computers, Babbageʹs invention never worked.

Angel Capital Refers to funds available for investment by networks of wealthy investors who invest in start‐up companies. A key source of capital for high‐tech start‐up companies in the United States.

Artificial intelligence (AI) The field of research that attempts to emulate human intelligence in a machine. Fields within AI include knowledge‐based systems, expert systems, pattern recognition, automatic learning, natural-language understanding, robotics, and others.

Artificial life Simulated organisms, each including a set of behavior and reproduction rules (a simulated ʺgenetic codeʺ), and a simulated environment. The simulated organisms simulate multiple generations of evolution. The term can refer to any self‐replicating pattern.

ASR See Automatic speech recognition.

Automatic speech recognition (ASR) Software that recognizes human speech. In general, ASR systems include the ability to extract high‐level patterns in speech data.

BGM See Brain‐generated music.

Big bang theory A prominent theory on the beginning of the Universe: the cosmic explosion, from a single point of infinite density, that marked the beginning of the Universe billions of years ago.

Big crunch A theory that the Universe will eventually lose momentum in expanding and contract and collapse in an event that is the opposite of the big bang.

Bioengineering The field of designing pharmaceutical drugs and strains of plant and animal life by directly modifying the genetic code. Bioengineered materials, drugs, and life‐forms are used in agriculture, medicine, and

the treatment of disease.

Biology The study of life‐forms. In evolutionary terms, the emergence of patterns of matter and energy that could survive and replicate to form future generations.

Bionic organ In 2029, artificial organs that are built using nanoengineering.

Biowarfare Agency (BWA) In the second decade of the twenty‐first century, a government agency that monitors and polices bioengineering technology applied to weapons.

Bit A contraction of the phrase ʺbinary digit.ʺ In a binary code, one of two possible values, usually zero and one. In information theory, the fundamental unit of information.

Brain‐generated music (BGM) A music technology pioneered by Neurosonics, Inc., that creates music in response to the listenerʹs brain waves. This brain‐wave biofeedback system appears to evoke the Relaxation Response by encouraging the generation of alpha waves in the brain.

BRUTUS.1 A computer program that creates fictional stories with a theme of betrayal; invented by Selmer Bringsjord, Dave Ferucci, and a team of software engineers at Rensselaer Polytechnic Institute in New York.

Buckyball A soccer‐ball‐shaped molecule formed of a large number of carbon atoms. Because of their hexagonal and pentagonal shape, the molecules were dubbed ʺbuckyballsʺ in reference to R. Buckminster Fullerʹs building designs.

Busy beaver One example of a class of noncomputational functions; an unsolvable problem in mathematics. Being a ʺTuring machine unsolvable problem,ʺ the busy beaver function cannot be computed by a Turing machine. To compute busy beaver of n, one creates all the n‐state Turing machines that do not write an infinite number of 1s on

their tape. The largest number of 1s written by the Turing machine in this set that writes the largest number of 1s is busy beaver of n.

BWA See Biowarfare Agency

Byte A contraction for ʺby eight.ʺ A group of eight bits clustered together to store one unit of information on a computer. A byte may correspond, for example, to a letter of the English alphabet.

CD‐ROM See Compact disc read‐only memory.

Chaos The amount of disorder or unpredictable behavior in a system. In reference to the Law of Time and Chaos, chaos refers to the quantity of random and unpredictable events that are relevant to a process.

Chaos theory The study of patterns and emergent behavior in complex systems comprised of many unpredictable elements (e.g., the weather).

Chemistry The composition and properties of substances comprised of molecules.

Chip A collection of related circuits that work together on a task or set of tasks, residing on a wafer of semiconductor material (typically silicon).

Closed system Interacting entities and forces not subject to outside influence (for example, the Universe). A corollary of the second law of thermodynamics is that in a closed system, entropy increases.

Cochlear implant An implant that performs frequency analyses of sound waves, similar to that performed by the inner ear.

Colossus The first electronic computer, built by the British from fifteen hundred radio tubes during World War II.

Colossus and nine similar machines running in parallel cracked increasingly complex German codes on military intelligence and contributed to the Allied forcesʹ winning of World War II.

Combinatorial explosion The rapid‐exponential‐growth in the number of possible ways of choosing distinct combinations of elements from a set as the number of elements in that set grows. In an algorithm, the rapid growth

in the number of alternatives to be explored while performing a search for a solution to a problem.

Common sense The ability to analyze a situation based on its context, using millions of integrated pieces of common knowledge. Currently, computers lack common sense. To quote Marvin Minsky: ʺDeep Blue might be able

to win at chess, but it wouldnʹt know to come in from the rain.ʺ

Compact disc read‐only memory (CD‐ROM) A laser‐read disc that contains up to a half billion bytes of information. ʺRead onlyʺ refers to the fact that information can be read, but not deleted or recorded, on the disc.

Complicated‐minded school The use of sophisticated procedures to evaluate the terminal leaves in a recursive algorithm.

Computation The process of calculating a result by use of an algorithm (e.g., a computer program) and related data.

The ability to remember and solve problems.

Computer A machine that implements an algorithm. A computer transforms data according to the specifications of an algorithm. A programmable computer allows the algorithm to be changed.

Computer language A set of rules and specifications for describing an algorithm or process on a computer.

Computing medium Computing circuitry capable of implementing one or more algorithms. Examples include human neurons and silicon chips.

Connectionism An approach to studying intelligence and to creating intelligent solutions to problems.

Connectionism is based on storing problem‐solving knowledge as a pattern of connections among a very large number of simple processing units operating in parallel.

Consciousness The ability to have subjective experience. The ability of a being, animal, or entity to have self-perception and self‐awareness. The ability to feel. A key question in the twenty‐first century is whether computers

will achieve consciousness (which their human creators are considered to have).

Continuous speech recognition (CSR) A software program that recognizes and records natural language.

Crystalline computing A system in which data is stored in a crystal as a hologram, conceived by Stanford professor Lambertus Hesselink. This three‐dimensional storage method requires a million atoms for each bit and could achieve a trillion bits of storage for each cubic centimeter. Crystalline computing also refers to the possibility of growing computers as crystals.

CSR See Continuous speech recognition.

Cybernetic artist A computer program that is able to create original artwork in poetry, visual art, or music.

Cybernetic artists will become increasingly commonplace starting in 2009.

Cybernetic chauffeur Self‐driving cars that use special sensors in the roads. Self‐driving cars are being experimented with in the late 1990s, with implementation on major highways feasible during the first decade of the

twenty‐first century, Cybernetic poet A computer program that is able to create original poetry.

Cybernetics A term coined by Norbert Wiener to describe the science of control and communication in animals and machines. Cybernetics is based on the theory that intelligent living beings adapt to their environments and accomplish objectives primarily by reacting to feedback from their surroundings.

Database The structured collection of data that is designed in connection with an information retrieval system. A database management system (DBMS) allows monitoring, updating, and interacting with the database.

Debugging The process of discovering and correcting errors in computer hardware and software. The issue of bugs or errors in a program will become increasingly important as computers are integrated into the human brain and

physiology throughout the twenty‐first century. The first ʺbugʺ was an actual moth, discovered by Grace Murray Hopper, the first programmer of the Mark I computer.

Deep Blue The computer program, created by IBM, that defeated Gary Kasparov, the worldʹs chess champion, in 1997.

Destroy‐all‐copies movement In 2099, a movement to permit an individual to terminate her mind file and to destroy all backup copies of that file.

Destructive scan The process of scanning oneʹs brain and neural system while destroying it, with a view to replacing it with electronic circuits of far greater capacity, speed, and reliability.

Digital Varying in discrete steps. The use of combinations of bits to represent data in computation. Contrasted with analog.

Digital video disc (DVD) A high‐density compact disc system that uses a more focused laser than the conventional CD‐ROM, with storage capacities of up to 9.4 gigabytes on a double‐sided disc. A DVD has sufficient capacity to hold a full‐length movie.

Direct neural pathway Direct electronic communication to the brain. In 2029, direct neural pathways, combined with wireless communication technology, will connect humans directly to the worldwide computing network (the

Web).

Diversity Variety of choices, in which evolution thrives. A key resource for an evolutionary process. The other resource for evolution is its own increasing order.

DNA Deoxyribonucleic acid; the building blocks of all organic life‐forms. In the twenty‐first century, intelligent life-forms will be based on new computational technologies and nanoengineering.

DNA computing A form of computing, pioneered by Leonard Adleman, in which DNA molecules are used to solve

complex mathematical problems. DNA computers allow trillions of computations to be performed simultaneously.

DVD See Digital video disc.

Einsteinʹs theory of relativity Refers to two of Einsteinʹs theories Einsteinʹs Special Theory of Relativity postulates the speed of light as the fastest speed at which we can transmit information. Einsteinʹs General Theory of Relativity deals with the effects of gravity on the geometry of space. Includes the formula E =mc2 (energy equals mass times

the speed of light squared), which is the basis of nuclear power.

EMI See Experiments in Musical Intelligence.

Encryption Encoding information so that only the intended recipient can understand the message by decoding it.

PGP (Pretty Good Privacy) is an example of encryption.

Entropy In thermodynamics, a measure of the chaos (unpredictable movement) of particles and unavailable energy in a physical system of many components. In other contexts, a term used to describe the extent of randomness and

disorder of a system.

Evolution A process in which diverse entities (sometimes called organisms) compete for limited resources in an environment, with the more successful organisms able to survive and reproduce (to a greater extent) into subsequent generations. Over many such generations, the organisms become better adapted at survival. Over generations, the order (suitability of information for a purpose) of the design of the organisms increases, with the purpose being survival. In an ʺevolutionary algorithmʺ (see below), the purpose may be defined to be the discovery

of a solution to a complex problem. Evolution also refers to a theory in which each life‐form on Earth has its origin in an earlier form.

Evolutionary algorithm Computer‐based problem‐solving systems that use computational models of the mechanisms of evolution as key elements in their design.

Experiments in Musical Intelligence (EMI) A computer program that composes musical scores. Created by the composer David Cope.

Expert system A computer program, based on various artificial intelligence techniques, that solves a problem using a database of expert knowledge on a topic. Also a system that enables such a database to become available to the

nonexpert user. A branch of the artificial intelligence field.

Exponential growth Characterized by growth in which size increases by a fixed multiple over time.

Exponential trend Any trend that exhibits, exponential growth (such as an exponential trend in population growth).

Femtoengineering In 2099, a proposed computing technology on the femtometer (one thousandth of a trillionth of a meter) scale. Femtoengineering requires harnessing mechanisms inside a quark. Molly discusses femtoengineering

proposals with the author in 2099.

Florence Manifesto Brigade In 2029, a neo‐Luddite group that is based on the ʺFlorence Manifestoʺ written by Theodore Kaczynski from prison. Members of the brigade protest technology primarily through nonviolent means.

Fog swarm projection In the mid‐ and late‐twenty‐first century, a technology that allows projections of physical objects and entities through the behavior of trillions of foglets. Mollyʹs physical appearance to the author in 2099 is created by a fog swarm projection. See Foglet; Utility fog.

Foglet A hypothetical robot that consists of a human‐cell‐sized device with twelve arms pointing in all directions.

At the end of the arms are grippers so that the Foglets can grasp one another to form larger structures. These nanobots are intelligent and can merge their computational capacities with one another to create a distributed intelligence. Foglets are the brainchild of J. Storrs Hall, a Rutgers University computer scientist.

Free will Purposeful behavior and decision making. Since the time of Plato, philosophers have explored the paradox of free will, particularly as it applies to machines. During the next century, a key issue will be whether machines will evolve into beings with consciousness and free will. A primary philosophical issue is how free will is possible if events are the result of the predictable—or unpredictable—interaction of particles. Considering the interaction of particles to be unpredictable does not resolve the paradox of free will because there is nothing purposeful in random behavior.

General Problem Solver (GPS) A procedure and program developed by Allen Newell, J. C. Shaw, and Herbert Simon. GPS attains an objective by using recursive search and by applying rules to generate the alternatives at each branch in the recursive expansion of possible sequences. GPS uses a procedure to measure the ʺdistanceʺ from the

goal.

Genetic algorithm A model of machine learning that derives its behavior from a metaphor of the mechanisms of evolution in nature. Within a program, a population of simulated ʺindividualsʺ are created and undergo a process

of evolution in a simulated competitive environment.

Genetic programming The method of creating a computer program using genetic or evolutionary algorithms. See Evolutionary algorithm; Genetic algorithm.

God spot A tiny locus of nerve cells in the frontal lobe of the brain that appears to be activated during religious experiences. Neuroscientists from the University of California discovered the God spot while studying epileptic patients who have intense mystical experiences during seizures.

Gödelʹs incompleteness theorem A theorem postulated by Kurt Gödel, a Czech mathematician, that states that in a mathematical system powerful enough to generate the natural numbers, there inevitably exist propositions that can

be neither proved nor disproved.

Gordian knot An intricate, practically unsolvable problem. A reference to the knot tied by Gordius, to be untied only by the future ruler of Asia. Alexander the Great circumvented the dilemma of untying the knot by slashing it

with his sword.

GPS See General Problem Solver.

Grandfather legislation As of 2099, legislation that protects the, rights of MOSHs (mostly original substrate humans) and acknowledges the roots of twenty‐first‐century beings. See MOSH.

Haptic interface In virtual reality systems, the physical actuators that provide the user with a sense of touch (including the sensing of pressure and temperature).

Haptics The development of systems that allow one to experience the sense of touch in virtual reality. See Haptic interface.

Hologram An interference pattern, often using photographic media, that is encoded by laser beams and read by means of low‐power laser beams. This interference pattern can reconstruct a three‐dimensional image. An important property of a hologram is that the information is distributed throughout the hologram. Cut a hologram

in half, and both halves will have the full picture, only at half the resolution. Scratching a hologram has no noticeable effect on the image. Human memory is regarded to be distributed in a similar way Holy Grail Any objective of a long and difficult quest. In medieval tore, the Grail refers to the plate used by Christ at the Last Supper. The Holy Grail subsequently became the object of knightsʹ quests.

Homo erectus ʺUpright man.ʺ Homo erectus emerged in Africa about 1.6 million years ago and developed fire, clothing, language, and weapon use.

Homo habilis ʺHandy human.ʺ A direct ancestor leading to Homo erectus and eventually to Homo sapiens. Homo habilis lived approximately 1.6 to 2 million years ago. Homo habilis hominids were different from previous hominids in their bigger brain size, diet of both meat and plants, and creation and use of rudimentary tools.

Homo sapiens Human species that emerged perhaps 400,000 years ago. Homo sapiens are similar to advanced primates in terms of their genetic heritage and are distinguished by their creation of technology, including art and language.

Homo sapiens neanderthal (neanderthalensis) A subspecies of Homo sapiens. Homo sapiens neanderthalensis is thought to have evolved from Homo erectus about 100,000 years ago in Europe and the Middle East. This highly intelligent subspecies cultivated an involved culture that included elaborate funeral rituals, burying their dead with ornaments, caring for the sick, and making tools for domestic use and for protection. Homo sapiens neanderthalensis disappeared about 35,000 to 40,000 years ago, in all likelihood as a result of violent conflict with Homo sapiens sapiens (the subspecies of contemporary humans).

Homo sapiens sapiens Another subspecies of Homo sapiens that emerged in Africa about 90,000 years ago.

Contemporary humans are the direct descendants of this subspecies.

Human Genome Project An international research program with the goal of gathering a resource of genomic maps and DNA sequence information that will provide detailed information about structure, organization, and

characteristics of the DNA of humans and other animals. The project began in the mid‐1980s and is expected to be

completed by around the year 2005.

Idiot Savant A system or person who is highly skilled in a narrow task area but who lacks context and is otherwise impaired in more general areas of intelligent functioning. The term is taken from psychiatry, where it refers to a person who exhibits brilliance in one very limited domain but is underdeveloped in common sense, knowledge, and competence. For example, some human idiot savants are capable of multiplying very large numbers in their heads, or memorizing a phone book. Deep Blue is an example of an idiot savant system.

Image processing The manipulation of data representing images, or pictorial representation on a screen, composed of pixels. The use of a computer program to enhance or modify an image.

Improvisor A computer program that creates original music, written by Paul Hodgson, a British jazz saxophone player. Improvisor can emulate styles ranging from Bach to jazz greats Louis Armstrong and Charlie Parker.

Industrial Revolution The period in history in the late eighteenth and nineteenth centuries marked by accelerating developments in technology that enabled the mass production of goods and materials.

Information A sequence of data that is meaningful in a process, such as the DNA code of an organism or the bits in a computer program. Information is contrasted with ʺnoise,ʺ which is a random sequence. However, neither noise

nor information is predictable. Noise is inherently unpredictable but carries no information. Information is also unpredictable; that is, we cannot predict future information from past information. If we can fully predict future data from past data, then that future data stops being information.

Information Theory A mathematical theory concerning the difference between information and noise, and the ability of a communications channel to carry information.

Intelligence The ability to use optimally limited resources—including time—to achieve a set of goals (which may include survival, communication, solving problems, recognizing patterns, performing skills). The products of intelligence may be clever, ingenious, insightful, or elegant. R. W. Young defines intelligence as ʺthat faculty of mind by which order is perceived in a situation previously considered disordered.ʺ

Intelligent agent An autonomous software program that performs a function on its own, such as searching the Web for information of interest to a person based on certain criteria.

Intelligent function A function that requires increasing intelligence to compute for increasing arguments. The busy beaver is an example of an intelligent function.

Internet computation harvesting proposal A proposal to harvest the unused computational resources of personal computers on the Internet and thereby create virtual parallel supercomputers. There are sufficient unused ʺcomputesʺ on the Internet in 1998 to create human brain capacity supercomputers, at least in terms of hardware capability Knee of the curve The period in which the exponential nature of the curve of time begins to explode.

Exponential growth lingers with no apparent growth for a long period of time and then appears to erupt suddenly.

This is now occurring in the capability of computers.

Knowledge engineering The art of designing and building expert systems. In particular, collecting knowledge and heuristic rules from human experts in their area of specialty and assembling them into a knowledge base or expert

system.

Knowledge principle A principle that emphasizes the important role played by knowledge in many forms of intelligent activity. It states that a system exhibits intelligence in part due to the specific knowledge relevant to the task that it contains.

Knowledge representation A system for organizing human knowledge in a domain into a data structure flexible enough to allow the expression of facts, rules, and relationships.

Law of Accelerating Returns As order exponentially increases, time exponentially speeds up (i.e., the time interval between salient events grows shorter as time passes).

Law of Increasing Chaos As chaos exponentially increases, time exponentially slows down (i.e., the time interval between salient events grows longer as time passes).

Law of Time and Chaos In a process, the time interval between salient events (i.e., events that change the nature of the process, or significantly affect the future of the process) expands or contracts along with the amount of chaos.

Laws of thermodynamics The laws of thermodynamics govern how and why energy is transferred.

The first law of thermodynamics (postulated by Hermann von Helmholtz in 1847), also called the Law of Conservation of Energy, states that the total amount of energy in the Universe is constant. A process may modify

the form of energy, but a closed system does not lose energy. We can use this knowledge to determine the amount

of energy in a system, the amount lost as waste heat, and the efficiency of the system.

The second law of thermodynamics (articulated by Rudolf Clausias in 1850), also known as the Law of

increasing Entropy, states that the entropy (disorder of particles) in the Universe never decreases. As the disorder in the Universe increases, the energy is transformed into less usable forms. Thus, the efficiency of any process will always be less than 100 percent.

The third law of thermodynamics (described by Walter Hermann Nernst in 1906, based on the idea of a temperature of absolute zero first articulated by Baron Kelvin in 1848), also known as the Law of Absolute Zero, tells us that all molecular movement stops at a temperature called absolute zero, or 0 Kelvin (–273°C). Since temperature is a measure of molecular movement, the temperature of absolute zero can be approached, but it can

never be reached.

Life The ability of entities (usually organisms) to reproduce into future generations. Patterns of matter and energy that can perpetuate themselves and survive.

LISP (list processing) An interpretive computer language developed in the late 1950s at MIT by John McCarthy used to manipulate symbolic strings of instructions and data. The principal data structure is the list, a finite ordered sequence of symbols. Because a program written in LISP is itself expressed as a list of lists, LISP lends itself to sophisticated recursion, symbol manipulation, and self‐modifying code. It has been widely used for AI

programming, although it is less popular today than it was in the 1970s and 1980S.

Logical positivism A twentieth‐century philosophical school of thought that was inspired by Ludwig Wittgensteinʹs Tractatus Logico‐Philosophicus. According to logical positivism, all meaningful statements may be confirmed by observation and experiment or are ʺanalyticʺ (deducible from observations).

Luddite One of a group of early‐nineteenth‐century English workmen who destroyed labor‐saving machinery in protest. The Luddites were the first organized movement to oppose the mechanized technology of the Industrial Revolution. Today, the Luddites are a symbol of opposition to technology.

Magnetic resonance imaging (MRI) A noninvasive diagnostic technique that produces computerized images of body tissues and is based on nuclear magnetic resonance of atoms within the body produced by the application of

radio waves. A person is placed in a magnetic field thirty thousand times stronger than the normal magnetic field

on Earth. The personʹs body is stimulated with radio waves, and the body responds with its own electromagnetic

transmissions. These are detected and processed by computer to generate a three‐dimensional map of high-resolution internal features such as blood vessels.

Massively parallel neural nets A neural net built from many parallel processing units. Generally, a separate, specialized computer implements each neuron model.

Microprocessor An integrated circuit built on a single chip containing the entire central processing unit (CPU) of a computer.

Millions of Instructions per Second A method of measuring the speed of a computer in terms of the number of millions of instructions performed by the computer in one second. An instruction is a single step in a computer program as represented in the computerʹs machine language.

Mind‐body problem The philosophical question: How does the nonphysical entity of the mind emerge from the physical entity of the brain? How do feelings and other subjective experiences result from the processing of the physical brain? By extension, will machines emulating the processes of the human brain have subjective experiences? Also, how does the nonphysical entity of the mind exert control over the physical reality of the body?

Mind trigger A stimulation of an area of the brain that evokes a feeling usually (i.e., otherwise) gained from actual physical or mental experience.

Minimax procedure or theorem A basic technique used in game‐playing programs. An expanding tree of possible moves and countermoves (moves from the opponent) is constructed. An evaluation of the final ʺleavesʺ of the tree

that minimizes the opponentʹs ability to win and maximizes the programʹs ability to win is then passed back down

the branches of the tree.

MIPS See Millions of Instructions per Second.

Mission critical system A software program that controls a process on which people are heavily dependent.

Examples of mission critical software include life‐support systems in hospitals, automated surgical equipment, autopilot flying and landing systems, and other software‐based systems that affect the well‐being of a person or organization.

Molecular computer A computer based on logic gates that is constructed on principles of molecular mechanics (as opposed to principles of electronics) by appropriate arrangements of molecules. Since the size of each logic gate (device that can perform a logical operation) is only one or a few molecules, the resultant computer can be microscopic in size. Limitations on molecular computers arise only from the physics of atoms. Molecular computers

can be massively parallel by having parallel computations performed by trillions of molecules simultaneously.

Molecular computers have been demonstrated using the DNA molecule.

Mooreʹs Law First postulated by former Intel CEO Gordon Moore in the mid‐1960s, Mooreʹs Law is the prediction that the size of each transistor on an integrated circuit chip will be reduced by 50 percent every twenty‐four months. The result is the exponentially growing power of integrated circuit‐based computation over time. Mooreʹs

Law doubles the number of components on a chip as well as the speed of each component. Both of these aspects double the power of computing, for an effective quadrupling of the power of computation every twenty‐four months.

MOSH In 2099, an acronym for Mostly Original Substrate Humans. in the last half of the twenty‐first century, a human being still using native carbon‐based neurons and unenhanced by neural implants is referred to as a MOSH.

In 2099, Molly refers to the author as being a MOSH.

MOSH art In 2099, art (that is usually created by enhanced humans) that a MOSH is theoretically capable of appreciating, although MOSH art is not always shared with a MOSH.

MOSH music in 2099, MOSH art in the form of music.

Moshism In 2099, an archaic term that is rooted in the MOSH way of life, before the advent of enhanced humans through neural implants and the porting of human brains to new computational substrates. An example of a Moshism: the word papers to refer to knowledge structures representing a body of intellectual work.

MRI See Magnetic resonance imaging.

MYCIN A successful expert system, developed at Stanford University in the mid‐1970s, designed to aid medical practitioners in prescribing an appropriate antibiotic by determining the exact identity of a blood infection.

Nanobot A nanorobot (robot built using nanotechnology). A self‐replicating nanobot requires mobility, intelligence, and the ability to manipulate its environment. It also needs to know when to stop its own replication. In 2029, nanobots will circulate through the bloodstream of the human body to diagnose illnesses.

Nanobot swarm In the last half of the twenty‐first century, a swarm comprised of trillions of nanobots, The nanobot swarms can rapidly take on any form. A nanobot swarm can project the visual images, sounds, and pressure contours of any set of objects, including people. The swarms of nanobots can also combine their computational abilities to emulate the intelligence of people and other intelligent entities and processes. A nanobot swarm effectively brings the ability to create virtual environments into the real environment.

Nanoengineering The design and manufacturing of products and other objects based on the manipulation of atoms and molecules; building machines atom by atom. ʺNanoʺ refers to a billionth of a meter, which is the width of five

carbon atoms. See Picoengineering; Femtoengineering.

Nanopathogen A self‐replicating nanobot that replicates excessively, possibly without limit, causing destruction to both organic and inorganic matter.

Nanopatrol In 2029, a nanobot in the bloodstream that checks the body for biological pathogens and other disease processes.

Nanotechnology A body of technology in which products and other objects are created through the manipulation of atoms and molecules. ʺNanoʺ refers to a billionth of a meter, which is the width of five carbon atoms.

Nanotubes Elongated carbon molecules that resemble long tubes and are formed of the same pentagonal patterns of carbon atoms as buckyballs. Nanotubes can perform the electronic functions of silicon‐based components.

Nanotubes are extremely small, thereby providing very high densities of computation. Nanotubes are a likely technology to continue to provide the exponential growth of computing when Mooreʹs Law on integrated circuits

dies by the year 2020. Nanotubes are also extremely strong and heat resistant, thereby permitting the creation of three‐dimensional circuits.

Natural language Language as ordinarily spoken or written by humans using a human language such as English (as contrasted with the rigid syntax of a computer language). Natural language is governed by rules and conventions

sufficiently complex and subtle for there to be frequent ambiguity in syntax and meaning.

Neanderthal See Homo sapiens neanderthal (neanderthalensis).

Neural computer A computer with hardware optimized for using the neural network paradigm. A neural computer is designed to simulate a massive number of models of human neurons.

Neural connection calculation In a neural network, a term that refers to the primary calculation of multiplying the ʺstrengthʺ of a neural connection by the input to that connection (which is either the output of another neuron or an initial input to the system) and then adding this product to the accumulated sum of such products from other connections to this neuron. This operation is highly repetitive, so neural computers are optimized for performing it.

Neural implant A brain implant that enhances oneʹs sensory ability, memory, or intelligence. Neural implants will become ubiquitous in the twenty‐first century.

Neural network A computer simulation of human neurons. A system (implemented in software or hardware) that is intended to emulate the computing structure of neurons in the human brain.

Neuron Information‐processing cell of the central nervous system. There are an estimated 100 billion neurons in the human brain.

Noise A random sequence of data. Because the sequence is random and without meaning, noise carries no information. Contrasted with information.

Objective experience The experience of an entity as observed by another entity, or measurement apparatus.

OCR See optical character recognition.

Operating system A software program that manages and provides a variety of services to application programs, including user interface facilities and management of input‐output and memory devices.

Optical character recognition (OCR) A process in which a machine scans, recognizes, and encodes printed (and possibly handwritten) characters into digital form.

Optical computer A computer that processes information encoded in patterns of light beams; different from todayʹs conventional computers, in which information is represented in electronic circuitry or encoded on magnetic surfaces. Each stream of photons can represent an independent sequence of data, thereby providing extremely massive parallel computation.

Optical imaging A brain‐imaging technique similar to MRI but potentially providing higher resolution imaging.

Optical imaging is based on the interaction between electrical activity in the neurons and blood circulation in the

capillaries feeding the neurons.

Order Information that fits a purpose. The measure of order is the measure of how well the information fits the purpose. In the evolution of life‐forms, the purpose is to survive. In an evolutionary algorithm (a computer program that simulates evolution to solve a problem), the purpose is to solve the problem. Having more information, or more complexity, does not necessarily result in a better fit. A superior solution for a purpose—

greater order—may require either more or less information, and either more or less complexity. Evolution has shown, however, that the general trend toward greater order does generally result in greater complexity.

Paradigm A pattern, model, or general approach to solving a problem.

Parallel processing Refers to computers that use multiple processors operating simultaneously as opposed to a single processing unit. (Compare with Serial computer.)

Pattern recognition Recognition of patterns with the goal of identifying, classifying, or categorizing complex inputs.

Examples of inputs include images such as printed characters and faces, and sounds such as spoken language.

Perceptron In the late 1960s and 1970s, a machine constructed from mathematical models of human neurons. Early Perceptrons were modestly successful in such pattern recognition tasks as identifying printed letters and speech sounds. The Perceptron was a forerunner of contemporary neural nets.

Personal computer A generic term for a single‐user computer using a microprocessor, and including the computing hardware and software needed for an individual to work autonomously, PGP See Pretty Good Privacy.

Picoengineering Technology on the picometer (one trillionth of a meter) scale. Picoengineering will involve engineering at the level of subatomic particles.

Picture portal In 2009, a visual display for viewing people and other real‐time images. In later years, the portals project three‐dimensional, real‐time scenes. Mollyʹs son, Jeremy, uses a picture portal to view the Stanford University campus.

Pixel An abbreviation for picture element. The smallest element on a computer screen that holds information to represent a picture. Pixels contain data giving brightness and possibly color at particular points in the picture.

Pretty Good Privacy (PGP) A system of encryption (designed by Phil Zimmerman) distributed on the Internet and widely used. PGP uses a public key that can be freely disseminated and used by anyone to encode a message and a

private key that is kept only by the intended recipient of the encoded messages. The private key is used by the recipient to decode messages encrypted using the public key. Converting the public key into a private key requires

factoring large numbers. If the number of bits in the public key is large enough, then the factors cannot be computed in a reasonable amount of time using conventional computation (and thus the encoded information remains secure). Quantum computing (with a sufficient number of qu‐bits) would destroy this type of encryption.

Price‐performance A measure of the performance of a product per unit cost.

Program A set of computer instructions that enables a computer to perform a specific task. Programs are usually written in a high‐level language such as ʺCʺ or ʺFORTRANʺ that can be understood by human programmers and then translated into machine language using a special program called a compiler. Machine language is a special set

of codes that directly controls a computer.

Punch card A rectangular card that typically records up to eighty characters of data in a binary coded format as a pattern of holes punched in it.

Quantum computing A revolutionary method of computing, based on quantum physics, that uses the ability of particles such as electrons to exist in more than one state at the same time. See Qu‐bit.

Quantum decoherence A process in which the ambiguous quantum state of a particle (such as the nuclear spin of an electron representing a qu‐bit in a quantum computer) is resolved into an unambiguous state as the result of direct or indirect observation by a conscious observer.

Quantum encryption A possible form of encryption using streams of quantum entangled particles such as photons.

See Quantum entanglement.

Quantum entanglement A relationship between two physically separated particles under special circumstances.

Two photons may be ʺquantum entangledʺ if produced by the same particle interaction and emerging in opposite

directions. The two photons remain quantum entangled with each other even when separated by very large distances (even when light‐years apart). In such a circumstance, the two quantum entangled photons, if each forced

to make a decision to choose among two equally probable pathways, will make the identical decision and will do so

at the same instant in time. Since there is no possible communication link between two quantum entangled photons, classical physics would predict that their decisions would be independent. But two quantum entangled photons make the same decision and do so at the same instant in time. Experiments have demonstrated that even if

there were an unknown communication path between them, there is not enough time for a message to travel from

one photon to the other at the speed of light.

Quantum mechanics A theory that describes the interactions of subatomic particles, combining several basic discoveries. These include Max Planckʹs 1900 observation that energy is absorbed or radiated in discrete quantities, called quanta. Also Werner Heisenbergʹs 1921 uncertainty principle stating that we cannot know both the exact position and momentum of an electron or other particle at the same time. Interpretations of quantum theory imply

that photons simultaneously take all possible paths (e.g., when bouncing off a mirror). Some paths cancel each other out. Remaining ambiguity in the path actually taken is resolved based on the conscious observation of an observer.

Qu‐bit A ʺquantum bit,ʺ used in quantum computing, that is both zero and one at the same time, until quantum decoherence (direct or indirect observation by a conscious observer) causes each quantum bit to disambiguate into

a state of zero or one. One qu‐bit stores two possible numbers (zero and one) at the same time. N qu‐bits; stores 2 to the Nth power possible numbers at the same time. Thus an N qu‐bit quantum computer would try 2 to the Nth power possible solutions to a problem simultaneously, which gives the quantum computer its enormous potential

power.

RAM See Random Access Memory.

Random Access Memory (RAM) Memory that can be both read and written with random access of memory locations. Random access means that locations can be accessed in any order and do not need to be accessed sequentially. RAM can be used as the working memory of a computer into which applications and programs can be

loaded and run.

Ray Kurzweilʹs Cybernetic Poet A computer program designed by Ray Kurzweil that uses a recursive approach to create poetry. The Cybernetic Poet analyzes word sequence patterns of poems it has ʺreadʺ using Markov models (a

mathematical cousin of neural nets) and creates new poetry based on these patterns.

Read‐Only Memory (ROM) A form of computer storage that can be read from but not written to or deleted (e.g., CD‐ROM).

Reading machine A machine that scans text and reads it aloud, initially developed for those who are visually impaired, reading machines are currently used by anyone who cannot read at their intellectual level, including reading disabled (e.g., dyslexic) persons and children first learning to read.

Recursion The process of defining or expressing a function or procedure in terms of itself. Typically, each iteration of a recursive‐solution procedure produces a simpler (or possibly smaller) version of the problem than the previous

iteration. This process continues until a subproblem whose answer is already known (or that can be readily computed without recursion) is obtained. A surprisingly large number of symbolic and numerical problems lend themselves to recursive formulations. Recursion is typically used by game‐playing programs, such as the chess-playing program Deep Blue.

Recursive formula A computer‐programming paradigm that uses recursive search to find a solution to a problem.

The recursive search is based on a precise definition of the problem (e.g., the rules of a game such as chess).

Relativity A theory based on two postulates: (1) that the speed of light in a vacuum is constant and independent of the source or the observer, and (2) that the mathematical forms of the laws of physics are invariant in all inertial systems. Implications of the theory of relativity include the equivalence of mass and energy and of change in mass,

dimension, and time with increased velocity. See also Einsteinʹs theory of relativity.

Relaxation Response A neurological mechanism discovered by Dr. Herbert Benson and other researchers at the Harvard Medical School and Bostonʹs Beth Israel Hospital. The opposite of the ʺfight or flightʺ or stress response,

the Relaxation Response is associated with reduced levels of epinephrine (adrenaline) and norepinephrine (noradrenaline), blood pressure, blood sugar, breathing, and heart rates.

Remember York movement In the second decade of the twenty‐first century, a neo‐Luddite web discussion group.

The group is named to commemorate the 1813 trial in York, England, during which a number of the Luddites who

destroyed industrial machinery were hanged, jailed, or exiled.

Reverse engineering Examining a product, program, or process to understand it and to determine its methods and algorithms. Scanning and copying a human brainʹs salient computational methods into a neural computer of sufficient capacity is a future example of reverse engineering.

RKCP See Ray Kurzweilʹs Cybernetic Poet.

Robinson The worldʹs first operational computer, constructed from telephone relays and named after a popular cartoonist who drew ʺRube Goldbergʺ machines (very ornate machinery with many interacting mechanisms).

During World War II, Robinson provided the British with a transcription of nearly all significant Nazi coded messages, until it was replaced by Colossus. See Colossus.

Robot A programmable device, linked to a computer, consisting of mechanical manipulators and sensors. A robot may perform a physical task normally done by human beings, possibly with greater speed, strength, and/or precision.

Robotics The science and technology of designing and manufacturing robots. Robotics combines artificial intelligence and mechanical engineering.

ROM See Read‐Only Memory.

Russellʹs Paradox The ambiguity created by the following question: Does a set that is defined as ʺall sets that do not include themselvesʺ include itself as a member? Russellʹs paradox motivated Bertrand Russell to create a new theory of sets.

Search A recursive procedure in which an automatic problem solver seeks a solution by iteratively exploring sequences of possible alternatives.

Second Industrial Revolution The automation of mental rather than physical tasks.

Second law of thermodynamics Also known as the Law of Increasing Entropy, this law states that the disorder (amount of random movement) of particles in the Universe may increase but never decreases. As the disorder in the

Universe increases, the energy is transformed into less usable forms. Thus, the efficiency of any process will always be less than 100 percent (hence the impossibility of perpetual motion machines).

Self‐replication A process or device that is capable of creating an additional copy of itself. Nanobots are self-replicating if they can create copies of themselves. Self‐replication is regarded as a necessary means of manufacturing nanobots due to the very large number (i.e., trillions) of such devices needed to perform useful functions.

Semiconductor A material commonly based on silicon or germanium with a conductivity midway between that of a good conductor and an insulator. Semiconductors are used to manufacture transistors. Semiconductors rely on the

phenomenon of tunneling. See Tunneling.

Sensorium In 2019, the product name for a total touch virtual reality environment, which provides an all-encompassing tactile environment.

Serial computer A computer that performs a single computation at a time. Thus two or more computations are performed one after the other, not simultaneously (even if the computations are independent). The opposite of a parallel processing computer.

Silicon Valley The area in California, south of San Francisco, that is a key center of high‐technology innovation, including the development of software, communication, integrated circuits and related technologies.

Simple‐minded school The use of simple procedures to evaluate the terminal leaves in a recursive algorithm. For example, in the context of a chess program, adding up piece values.

Simulated person A realistic, animated personality incorporating a convincing visual appearance and capable of communicating using natural language. By 2019, a simulated person can interact with real persons using visual, auditory, and tactile means in a virtual reality environment.

Simulator A program that models and represents an activity or environment on a computer system. Examples include the simulation of chemical interaction and fluid flow. Other examples include a flight simulator used to train pilots and a simulated patient to train physicians. Simulators are also often used for entertainment.

Society of mind A theory of the mind proposed by Marvin Minsky in which intelligence is seen to be the result of proper organization of a large number (a society) of other minds, which are in turn comprised of yet simpler minds.

At the bottom of this hierarchy are simple mechanisms, each of which is by itself unintelligent.

Software Information and knowledge used to perform useful functions by computers and computerized devices.

Includes computer programs and their data, but more generally also includes such knowledge products as books,

music, pictures, movies, and videos.

Software‐based evolution Software simulation of the evolutionary process. One example of software‐based evolution is Network Tierra, designed by Thomas Ray. Rayʹs ʺcreaturesʺ are software simulations of organisms in

which each ʺcellʺ has its own DNA‐like genetic code. The organisms compete with one another for the limited simulated space and energy resources of their simulated environment.

Speaker independence Refers to the ability of a speech‐recognition system to understand any speaker, regardless of whether or not the system has previously sampled that speakerʹs speech.

Stored‐program computer A computer in which the program is stored in memory along with the data to be operated on. A stored‐program capacity is an important capability for systems of artificial intelligence in that recursion and self‐modifying code are not possible without it.

Subjective experience The experience of an entity as experienced by the entity, as opposed to observations of that entity (including its internal processes) by another entity, or by a measurement apparatus.

Substrate Computing medium or circuitry. See Computing medium.

Supercomputer The fastest and most powerful computer available at any given time. Supercomputers are used for computations demanding high speed and storage (e.g., analyzing weather data).

Superconductivity The physical phenomenon whereby some materials exhibit zero electrical resistance at low temperatures. Superconductivity points to the possibility of great computational power with little or no heat dissipation (a limiting factor today). Heat dissipation is a major reason that three‐dimensional circuits are difficult to create.

Synthesizer A device that computes signals in real time. In the context of music, a (usually computer based) device that creates and generates sounds and music electronically.

Tactile virtualism By 2029, a technology that allows one to use a virtual body to enjoy virtual reality experiences without virtual reality equipment other than the use of neural implants (which include high‐bandwidth wireless communication). The neural implants create the pattern of nerve signals that corresponds to a comparable ʺrealʺ

experience.

Technology An evolving process of tool creation to shape and control the environment. Technology goes beyond the mere fashioning and use of tools. It involves a record of tool making and a progression in the sophistication of tools. It requires invention and is itself a continuation of evolution by other means. The ʺgenetic codeʺ of the evolutionary process of technology is the knowledge base maintained by the tool‐making species.

Three‐dimensional chip A chip that is constructed in three dimensions, thus allowing for hundreds or thousands of layers of circuitry. Three‐dimensional chips are currently being researched and engineered by a variety of companies.

Total touch environment In 2019, a virtual‐reality environment that provides an all‐encompassing tactile environment.

Transistor A switching and/or amplifying device using semiconductors, first created in 1948 by John Bardeen, Walter Brattain, and William Shockley of Bell Labs.

Translating telephone A telephone that provides real‐time speech translation from one human language to another.

Tunneling In quantum mechanics, the ability of electrons (negatively charged particles orbiting the nucleus of an atom) to exist in two places at once, in particular on both sides of a barrier. Tunneling allows some of the electrons to effectively move through the barrier and accounts for the ʺsemiʺ conductor properties of a transistor.

Turing machine A simple abstract model of a computing machine, designed by Alan Turing in his 1936 paper ʺOn Computable Numbers.ʺ The Turing machine is a fundamental concept in the theory of computation.

Turing Test A procedure proposed by Alan Turing in 1950 for determining whether or not a system (generally a computer) has achieved human‐level intelligence, based on whether it can deceive a human interrogator into believing that it is human. A human ʺjudgeʺ interviews the (computer) system, and one or more human ʺfoilsʺ over

terminal lines (by typing messages). Both the computer and the human foil(s) try to convince the human judge of

their humanness. If the human judge is unable to distinguish the computer from the human foil(s), then the computer is considered to have demonstrated human‐level intelligence. Turing did not specify many key details, such as the duration of the interrogation and the sophistication of the human judge and foils. By 2029, computers

are passing the test, although the validity of the test remains a point of controversy and philosophical debate.

Utility fog A space filled with Foglets. At the end of the twenty‐first century, utility fog can be used to simulate any environment, essentially providing ʺrealʺ reality with the environment‐transforming capabilities of virtual reality See Fog swarm projection; Foglet.

Vacuum tube The earliest form of an electronic switch (or amplifier) based on vacuum‐filled glass containers. Used in radios and other communication equipment and early computers; replaced by the transistor.

Venture Capital Refers to funds available for investment by organizations that have raised pools of capital specifically to invest in companies, primarily new ventures.

Virtual body In virtual reality, oneʹs own body potentially transformed to appear (and ultimately to feel) different than it does in ʺrealʺ reality.

Virtual reality A simulated environment in which you can immerse yourself. A virtual reality environment provides a convincing replacement for the visual and auditory senses, and (by 2019) the tactile sense. In later decades, the olfactory sense will be included as well. The key to a realistic visual experience in virtual reality is that when you move your head, the scene instantly repositions itself so that you are now looking at a different region of a three‐dimensional scene. The intention is to simulate what happens when you turn your real head in the real world: The images captured by your retinas rapidly change. Your brain nonetheless understands that the world has

remained stationary and that the image is sliding across your retinas only because your head is rotating. Initially, virtual reality (including crude contemporary systems) requires the use of special helmets to provide the visual and auditory environments. By 2019, virtual reality will be provided by ubiquitous contact‐lens‐based systems and implanted retinal‐imaging devices (as well as comparable devices for auditory ʺimagingʺ). Later in the twenty‐first

century, virtual reality (which will include all the senses) will be provided by direct stimulation of nerve pathways using neural implants.

Virtual reality auditory lenses In 2019, sonic devices that project high‐resolution sounds precisely placed in the three‐dimensional virtual environment. These can be built into eyeglasses, worn as body jewelry, or implanted.

Virtual reality blocking display In 2019, a display technology using virtual reality optical lenses (see below) and virtual reality auditory lenses (see above) that creates highly realistic virtual visual environments. The display blocks out the real environment, so you see and hear only the projected virtual environment.

Virtual reality head‐directed display In 2019, a display technology using virtual reality optical lenses (see below) and virtual reality auditory lenses (see above) that projects a virtual environment stationary with respect to the position and orientation of your head. When you move your head, the display moves relative to the real environment. This mode is often used to interact with virtual documents.

Virtual reality optical lenses In 2009, three‐dimensional displays built into glasses or contact lenses. These ʺdirect eyeʺ displays create highly realistic virtual visual environments overlaying the ʺrealʺ environment. This display technology projects images directly onto the human retina, exceeds the resolution of human vision, and is widely

used regardless of visual impairment. In 1998, the Microvision Virtual Retina Display provides a similar capability

for military pilots, with consumer versions anticipated.

Virtual reality overlay display In 2019, a display technology using virtual reality optical lenses (see above) and virtual reality auditory lenses (see above) that integrates real and virtual environments. The displayed images slide when you move or turn your head so that the virtual people, objects, and environment appear to remain stationary

in relation to the real environment (which you can still see). Thus if the direct eye display is displaying the image of a person (who could be a geographically remote real person engaging in a three‐dimensional visual phone call with

you, or a computer‐generated simulated person), that projected person will appear to be in a particular place relative to the real environment that you also see. When you move your head, that projected person will appear to

remain in the same place relative to the real environment.

Virtual sex Sex in virtual reality incorporating a visual, auditory, and tactile environment. The sex partner can be a real or simulated person.

Virtual tactile environment A virtual reality system that allows the user to experience a realistic and all-encompassing tactile environment.

Vision chip A silicon emulation of the human retina that captures the algorithm of early mammalian visual processing, an algorithm called center surround filtering.

World Wide Web (WWW) A highly distributed (not centralized) communications network allowing individuals and

organizations around the world to communicate with one another. Communication includes the sharing of text, images, sounds, video, software, and other forms of information. The primary user interface paradigm of the ʺwebʺ

is based on hypertext, which consists of documents (which can contain any type of data) connected by ʺlinks,ʺ

which the user selects by a pointing device such as a mouse. The Web is a system of data‐and‐message servers linked by high‐capacity communication links that can be accessed by any computer user with a ʺweb browserʺ and

Internet access. With the introduction of Windows98, access to the Web is built into the operating system. By the late twenty‐first century, the Web will provide the distributed computing medium for software‐based humans.

Y2K (year 2000 problem) Refers to anticipated difficulties caused by software (usually developed several decades prior to the year 2000) in which date fields used only two digits. Unless the software is adjusted, this will cause computer programs to behave erratically when the year becomes ʺ00.ʺ These programs will mistake the year 2000

for 1900.

NOTES

PROLOGUE: AN INEXORABLE EMERGENCE

1. My recollections of The Twilight Zone episode are essentially accurate, although the gambler is actually a small‐time crook named Rocky Valentine. Episode 28, ʺA Nice Place to Visitʺ (I learned the name of the episode after writing the prologue), aired during the first season of The Twilight Zone, on April 15, 1960.

The episode begins with a voice‐over: ʺPortrait of a man at work, the only work heʹs ever done, the only work he

knows. His name is Henry Francis Valentine, but he calls himself Rocky, because thatʹs the way his life has been—

rocky and perilous and uphill at a dead run all the way. . . .ʺ

While robbing a pawnbrokerʹs shop, Valentine is shot and killed by a policeman. When he awakens, he is met by

his afterlife guide, Pip. Pip explains that he will provide Valentine with whatever he wants. Valentine is suspicious, but he asks for and receives a million dollars and a beautiful girl. He then goes on a gambling spree, winning at the roulette table, at the slot machines, and later, at pool. He is also surrounded by beautiful women, who shower him

with attention.

Eventually Valentine tires of the gambling, the winning, and the beautiful women. He tells Pip that it is boring to

win all the time and that he doesnʹt belong in Heaven. He begs Pip to take him to ʺthe Other Place.ʺ With a malicious gleam in his eye, Pip replies, ʺThis is the Other Place!ʺ Episode synopsis adapted from Marc Scott Zicree, The Twilight Zone Companion (Toronto: Bantam Books, 1982, 113–115).

2. What were the primary political and philosophical issues of the twentieth century? One was ideological—totalitarian systems of the right (fascism) and left (communism) were confronted and largely defeated by capitalism (albeit with

a large public sector) and democracy. Another was the rise of technology, which began to be felt in the nineteenth

century and became a major force in the twentieth century. But the issue of ʺwhat constitutes a human beingʺ is not

yet a primary issue (except as it affects the abortion debate), although the past century did witness the continuation of earlier struggles to include all members of the species as deserving of certain rights.

3. For an excellent overview and technical details on neural‐network pattern recognition, see the ʺNeural Network

Frequently Asked Questionsʺ web site, edited by W S. Sarle, at <ftp://ftp.sas.com/pub/neural/FAQ.html>. In addition, an article by Charles Arthur, ʺComputers Learn to See and Smell Us,ʺ from Independent, January 16, 1996, describes the ability of neural nets to differentiate between unique characteristics.

4. As will be discussed in chapter 6, ʺBuilding New Brains,ʺ destructive scanning will be feasible early in the twenty-first century. Noninvasive scanning with sufficient resolution and bandwidth will take longer but will be feasible by the end of the first half of the twenty‐first century.

CHAPTER 1: THE LAW OF TIME AND CHAOS

1. For a comprehensive overview and detailed references on the big bang theory and the origin of the Universe, see

ʺIntroduction to Big Bang Theory, Bowdoin College Department of Physics and Astronomy at

<http://www.bowdoin.edu/dept/physics/astro.1997/astro4/bigbang.ht ml>.

Print sources on the big bang theory include: Joseph Silk, A Short History of the Universe (New York: Scientific American Library, 1994); Joseph Silk, The Big Bang (San Francisco: W H. Freeman and Company, 1980); Robert M.

Wald, Space, Time & Gravity (Chicago: The University of Chicago Press, 1977); and Stephen W. Hawking, A Brief History of Time (New York: Bantam Books, 1988).

2. The strong force holds an atomic nucleus together. It is called ʺstrongʺ because it needs to overcome the powerful repulsion between the protons in a nucleus with more than one proton.

3. The electroweak force combines electromagnetism and the weak force responsible for beta decay. In 1968, American physicist Steven Weinberg and Pakistani physicist Abdus Salam were successful in their unification of the weak

force and the electromagnetic force using a mathematical method called gauge symmetry.

4. The weak force is responsible for beta decay and her slow nuclear processes that occur gradually

5. Albert Einstein, Relativity: The Special and the General Theory (New York: Crown Publishers, 1961).

6. The laws of thermodynamics govern how and why energy is transferred.

The first law of thermodynamics (Postulated by Hermann von Helmholtz in 1847), also called the Law of

Conservation of Energy, states that the total amount of energy in the universe is constant.

The second law of thermodynamics (articulated by Rudolf Clausias in 1850), also known as the Law of Increasing

Entropy, states that entropy, or disorder, in the Universe never decreases (and, therefore, usually increases). As the disorder in the Universe increases, the energy is transformed into less usable forms. Thus the efficiency of any

process will always be less than 100 percent.

The third law of thermodynamics (described by Walter Hermann Nernst in 1906, based on the idea of a

temperature of absolute zero first articulated by Baron Kelvin in 1848), also known as the Law of Absolute Zero, tells us that all molecular movement stops at a temperature called absolute zero, or 0 Kelvin (–273°C). Since temperature

is a measure of molecular movement, the temperature of absolute zero can be approached, but it can never be

reached.

7. ʺEvolution and Behaviorʺ at <http://ccp.uchicago.edu/~jyin/evolution.html> contains an excellent collection of articles and links exploring the theories of evolution. Print sources include Edward O. Wilson , The Diversity of Life (New York: W. W. Norton & Company, 1993); and Stephen Jay Gould, The Book of Life (New York: W. W. Norton & Company, 1993).

8. Four hundred million years ago, vegetation spread from lowland swamps to create the first land‐based plants. This development permitted vertebrate herbivorous animals to step onto land, creating the first amphibians. Along with

the amphibians, arthropods also stepped onto land, some of which evolved into insects. About 200 million years ago,

dinosaurs and mammals began sharing the same environment. The dinosaurs were far more noticeable. Mostly the

mammals stayed out of the dinosaursʹ way, with many mammals being nocturnal.

9. Mammals became dominant in the niche of land‐based animals after the demise of the dinosaurs 65 million years

ago. Mammals are the more intellectual animal class, distinguished by warm blood, the nourishment of their

children with maternal milk, hairy skin, sexual reproduction, four appendages (in most cases) and, most notably, a

highly developed nervous system.

10. Primates, the most advanced mammalian order, were distinguished by forward‐facing eyes, binocular vision, large brains with a convoluted cortex, which permitted more advanced reasoning faculties, and complicated social

patterns. Primates were not the only intelligent animals, but they had one additional characteristic that would hasten the age of computation: the opposable thumb. The two qualities heeded for the subsequent emergence of technology

were now coming into place: intelligence and the ability to manipulate the environment. Itʹs no coincidence that

fingers are called digits. The origin of the word digit, as used in Modern English and appearing first in Middle

English, is from the Latin word digitus, for ʺfingerʺ or ʺtoeʺ; perhaps akin to Greek deiknynai, ʺto show.ʺ

11. About 50 million years ago, the anthropoid suborder of primates split off. Unlike their prosimian cousins, the

anthropoids underwent rapid evolution, giving rise to advanced primates such as monkeys and apes about 30

million years ago. These sophisticated primates were noted for subtle communication abilities using sounds,

gestures, and facial expressions, thereby allowing the development of intricate social groups. About 15 million years ago, the first humanoids emerged. Although they initially walked on their hind legs, they used the knuckles of their front legs for balance.

12. Although it is worth pointing out that a 2 percent change in a computer program can be very significant.

13. Homo sapiens are the only technology‐creating species on Earth today, but were not the first such species. Emerging about five million years ago was Homo habilis (i.e., ʺhandyʺ human being), known for his erect posture and large brain. He was called handy because he fashioned and used tools. Our most direct ancestor, Homo erectus, showed up in Africa about two million years ago. Homo erectus was also responsible for advancing technology, including the domestication of fire, the development of language, and the use of weapons.

14. Technology emerged from the mists of humanoid history and has accelerated ever since. Technologies invented by

other human species and subspecies included the domestication of fire, tools of stone, pottery, clothing, and other

means of providing for basic human needs. Early humanoids also initiated the development of language, visual art,

music, and other means for human communication.

About ten thousand years ago, humans began domesticating plants, and soon thereafter, animals. Nomadic

hunting tribes began settling down, allowing for more stable forms of social organization. Buildings were

constructed to protect both humans and their farming products. More effective means of transportation emerged,

facilitating the emergence of trade and large‐scale human societies.

The wheel appears to be a relatively recent innovation, with the oldest excavated wheels dating from about 5,500

years ago in Mesopotamia. Emerging around the same time in the same region were rafts, boats, and a system of

ʺcuneiformʺ inscriptions, the first form of written language that we are aware of.

These technologies enabled humans to congregate in large groups, allowing the emergence of civilization. The

first cities emerged in Mesopotamia around 6,000 years ago. Emerging about a millennium later were the ancient

Egyptian cities, including Memphis and Thebes, culminating in the reigns of the great Egyptian kings. These cities

were constructed as war machines with defensive walls protected by armies utilizing weapons drawn from the most

advanced technologies of their time, including chariots, spears, armor, and bows and arrows. Civilization in turn

allowed for human specialization of labor through a caste system and organized efforts at advancing technology. An

intellectual class including teachers, engineers, physicians, and scribes emerged. Other contributions by the early

Egyptian civilization included a paperlike material manufactured from papyrus plants, standardization of

measurement, sophisticated metalworking, water management, and a calendar.

More than 2,000 years ago, the Greeks invented elaborate machinery with multiple internal states. Archimedes,

Ptolemy, and others described levers, cams, pulleys, valves, cogs, and other intricate mechanisms that

revolutionized the measurement of time, navigation, mapmaking, and the construction of buildings and ships. The

Greeks are perhaps best known for their contributions to the arts, particularly literature, theater, and sculpture.

The Greeks were superseded by the superior military technology of the Romans. The Roman empire was so

successful that it produced the first urban civilization to experience long‐term peace and stability Roman engineers constructed tens of thousands of kilometers of roads and thousands of public constructions such as administrative

buildings, bridges, sports stadiums, baths, and sewers. The Romans made particularly notable advances in military

technology, including advanced chariots and armor, the catapult and javelin, and other effective tools of war.

The fall of the Roman empire around 500 A.D. ushered in the misnamed Dark Ages. While progress during the

next thousand years was slow by contemporary standards, the ever tightening spiral that is technological progress

continued to accelerate. Science, technology, religion, art, literature, and philosophy all continued to evolve in

Byzantine, Islamic, Chinese, and other societies. Worldwide trade enabled a cross‐fertilization in technologies. In

Europe, for example, the crossbow and gunpowder were borrowed from China. The spinning wheel was borrowed

from India. Paper and printing were developed in China about 2,000 years ago and migrated to Europe many

centuries later. Windmills emerged in several parts of the world, facilitating expertise with elaborate gearing

machines that would subsequently support the first calculating machines.

The invention in the thirteenth century of a weight‐driven clock using the cam technology perfected for

windmills and waterwheels freed society from structuring their lives around the sun. Perhaps the most significant

invention of the late Middle Ages was Johannes Gutenbergʹs invention of the movable‐type printing press, which

opened intellectual life beyond an elite controlled by church and state.

By the seventeenth century, technology had created the means for empires to span the globe. Several European

countries, including England, France, and Spain, were developing economies based on far‐flung colonies. This

colonization spawned the emergence of a merchant class, a worldwide banking system, and early forms of

intellectual property protection, including the patent.

On May 26, 1733, the English Patent Office issued a patent to John Kay for his ʺNew Engine for Opening and

Dressing Wool.ʺ

This was good news, for he had plans to manufacture his ʺflying shuttleʺ and market it to the burgeoning English

textile industry. Kayʹs invention was a quick success, but he spent all of his profits on litigation, attempting in vain to enforce his patent. He died in poverty, never realizing that his innovation in the weaving of cloth represented the launching of the Industrial Revolution.

The widespread adoption of Kayʹs innovation created pressure for a more efficient way to spin yarn, which

resulted in Sir Richard Arkwrightʹs Cotton Jenny, patented in 1770. In the 1780s, machines were invented to card and comb the wool to feed the new automated spinning machines. By the end of the eighteenth century, the English

cottage industry of textiles was replaced with increasingly efficient centralized machines. The birth of the Industrial Revolution led to the founding of the Luddite movement in the early 1800s, the first organized movement opposing

technology.

15. Primatologist Carl Van Schaik observed that the orangutans of Sumatraʹs Suaq Balimbing swamp all make and use

tools to reach insects, honey, and fruit. Though captive orangutans are easily taught to use tools, the Suaq primates are the first wild population observed using tools. The use of tools may be a result of necessity. Orangutans in other parts of the world have not been observed to use tools, basically because their food supply is more easily accessible.

Carl Zimmer, ʺTooling Through the Trees.ʺ Discover 16, no. 11 (November 1995): 46–47.

Crows fashion tools from sticks and leaves. The tools are used for different purposes, are highly predictable in

their construction, and even have hooks and other mechanisms for finding and manipulating insect prey. They often

carry these devices when flying and store them next to their nests.

Tina Adler, ʺCrows Rely on Tools to Get Their Work Done.ʺ Science News 149 no. 3 (January 20, 1996): 37.

Crocodiles canʹt grip prey, so they sometimes trap prey between rocks and/or roots. The tree root acts to anchor

the dead prey while the crocodile eats its meal. Some people have attributed the crocodiles use of stones and roots as using tools.

From the ʺAnimal Diversity Web Siteʺ at the University of Michiganʹs Museum of Zoology,

<http://www.oit.itd.umich.edu/projects/ADW/>.

16. An animal communicates for a variety of reasons: defense (to signal approaching danger to other members of its

species), food gathering (to alert other members to a food source), courtship and mating (to alert members of its

desirability and to warn potential competitors away), and maintenance of territory. The basic motivation for

communication is survival of the species. Some animals use communication not only for survival, but also to express

emotion.

There are many fascinating examples of animal communication:

• A female tree frog found in Malaysia uses its toes to tap on vegetation, alerting potential mates to her

availability. Lori Oliwenstein, Fenella Saunders, and Rachel Preiser, ʺAnimals 1995.ʺ Discover 17, no. 1

(January 1996): 54–57.

• Male meadow voles (a small rodent) groom themselves in order to produce body odors that will attract

their mates. Tina Adler, ʺVoles Appreciate the Value of Good Grooming.ʺ Science News 149, no. 16 (April

20, 1996): 247.

• Whales communicate through a series of calls and cries. Mark Higgins, ʺDeep Sea Dialogue.ʺ Nature

Canada 26, no. 3 (Summer 1997): 29–34.

• Primates, of course, vocalize to communicate a variety of messages. One group of researchers studied

capuchin monkeys, squirrel monkeys, and golden‐lion tamarins in Central and South America. Often

these animals are unable to see each other through the forest, so they developed a series of calls or trills

that would alert members to move toward food sources. Bruce Bower, ʺMonkeys Sound Off, Move Out.ʺ

Science News 149, no. 17 (April 27, 1996): 269.

17. Washoe and Koko (male and female gorillas, respectively) are credited with acquiring American Sign Language

(ASL). They are the most famous of the communicating primates. Viki, a chimpanzee, was taught to vocalize three

words (mama, papa, and cup). Lana and Kanzi (female chimpanzees) were taught to press buttons with symbols.

Steven Pinker reflects upon researchersʹ claims that apes fully comprehend sign language. In The Language

Instinct: How the Mind Creates Language (New York: Morrow, 1994), he notes that the apes learned a very crude form of ASL, not the full nuances of this language. The signs they learned were crude mimics of the ʺreal thing.ʺ In

addition, according to Pinker, the researchers often misinterpreted apesʹ hand motions as actual signs. One

researcher on Washoeʹs team who was deaf noted that other researchers would keep a log of long lists of signs,

whereas the deaf researcherʹs log was short.

18. David E. Kalish. ʺChip Makers and U.S. Unveil Project.ʺ New York Times, September 12, 1997.

19. The chart ʺThe Exponential Growth of Computing, 1900–1998ʺ is based on the following data:

Date

Device

Add Time

Calculations

Cost

Cost

CPS/$1000

(sec)

per Second (cps)

(then dollars)

1998 Dollars

1900

Analytical Engine

9.00E–00

1.11E–01

$1,000,000

$19,087,000

5.8211–06

1908

Hollerith Tabulator

5.00E+01

2.00E–02

$9,000

$154,000

1.299E–04

1911

Monroe Calculator

3.00E+01

3.33E–02

$35,000

$576,000

5.787E–05

1919

IBM Tabulator

5.00E–00

2.00E–01

$20,000

$188,000

1.064E–03

1928

National Ellis 3000

1.00E+01

1.00E–01

$15,000

$143,000

6.993E–04

1939

Zuse 2

1.00E–00

1.00E–00

$10,000

$117,000

8.547E–03

1940

Bell Calculator

3.00E–01

3.33E–00

$20,000

$233,000

1.431E–02

Model 1

1941

Zuse 3

3.00E–01

3.33E–00

$6,500

$72,000

4.630E–02

1943

Colossus

2.00E–04

5.00E+03

$100,000

$942,000

5.308E–00

1946

ENIAC

2.00E–04

5.00E+03

$750,000

$6,265,000

7.981E–01

1948

IBM SSEC

8.00E–04

1.25E+03

$500,000

$3,380,000

3.698E–01

1949

BINAC

2.86E–04

3.50E+03

$278,000

$1,903,000

1.837E–00

1949

EDSAC

1.40E–03

7.14E+02

$100,000

$684,000

1.044E–00

1951

Univac I

1.20E–04

8.33E+03

$930,000

$5,827,000

1.430E–00

1953

Univac 1103

3.00E–05

3.33E+04

$895,000

$5,461,000

6.104E–00

1953

IBM 701

6.00E–05

1.67E+04

$230,000

$1,403,000

1.188E+01

1954

EDVAC

9.00E–04

1.11E+03

$500,000

$3,028,000

3.669E–01

1955

Whirlwind

5.00E–05

2.00E+04

$200,000

$1,216,000

1.645E+01

1955

IBM 704

2.40E–05

4.17E+04

$1,994,000

$12,120,000

3.438E–00

1958

Datamatic 1000

2.50E–04

4.00E+03

$2,179,100

$12,283,000 3.257E–01

1958

Univac II

2.00E–04

5.00E+03

$970,000

$5,468,000

9.144E–01

1959

Mobidic

1.60E–05

6.25E+04

$1,340,000

$7,501,000

8.332E–00

1959

IBM 7090

4.00E–06

2.50E+05

$3,000,000

$16,794,000

1.489E+01

1960

IBM 1620

6.00E–04

1.67E+03

$200,000

$1,101,000

1.514E–00

1960

DEC PDP–1

1.00E–05

1.00E+05

$120,000

$660,000

1.515E+02

1961

DEC PDP–4

1.00E–05

1.00E+05

$65,000

$354,000

2.825E+02

1962

Univac III

9.00E–06

1.11E+05

$700,000

$3,776,000

2.943E+01

1964

CDC 6600

2.00E–07

5.00E+06

$6,000,000

$31,529,000

1.586E+02

1965

IBM 1130

8.00E–06

1.25E+05

$50,000

$259,000

4.826E+02

1965

DEC PDP–8

6.00E–06

1.67E+05

$18,000

$93,000

1.792E+03

1966

IBM 360 Model 75

8.00E–07

1.25E+06

$5,000,000

$25,139,000

4.972E+01

1968

DEC PDP–10

2.00E–06

5.00E+05

$500,000

$2,341,000

2.136E+02

1973

Intellec–8

1.56E–04

6.41E+03

$2,398

$8,798

7.286E+02

1973

Data General Nova

2.00E–05

5.00E+04

$4,000

$14,700

3.401E+03

1975

Altair 8800

1.56E–05

6.41E+04

$2,000

$6,056

1.058E+04

1976

DEC PDP–11

3.00E–06

3.33E+05

$150,000

$429,000

7.770E+02

Model 70

1977

Cray 1

1.00E–08

1.00E+08

$10,000,000

$26,881,000

3.720E+03

1977

Apple II

1.00E–05

1.00E+05

$1,300

$3,722

2.687E+04

1979

DEC VAX 11

2.00E–06

5.00E+05

$200,000

$449,000

1.114E+03

Model 780

1980

Sun–1

3.00E–06

3.33E+05

$30,000

$59,300

5.621E+03

1982

IBM PC

1.56E–06

6.41E+05

$3,000

$5,064

1.266E+05

1982

Compaq Portable

1.56E–06

6.41E+05

$3,000

$5,064

1.266E+05

1983

IBM AT–80286

1.25E–06

8.00E+05

$5,669

$9,272

8.628E+04

1984

Apple Macintosh

3.00E–06

3,33E+05

$2,500

$3,920

8.503E+04

1986

Compaq

2.50E–07

4.00E+06

$5,000

$7,432

5.382E+05

Deskpro 386

1987

Apple Mac II

1.00E–06

1.00E+06

$3,000

$4,300

2.326E+05

1993

Pentium PC

1.00E–07

1.00E+07

$2,500

$2,818

3.549E+06

1996

Pentium PC

1.00E–08

1.00E+08

$2,000

$2,080

4.808E+07

1998

Pentium II PC

5.00E–09

2.00E+08

$1,500

$1,500

1.333E+08

Cost conversions from dollars in each year to 1998 dollars are based on the ratio of the consumer price indices (CPI) for the respective years, based on CPI data as recorded by the Woodrow Federal Reserve Bank of Minneapolis. See

their web site, <http://woodrow.mpls.frb.fed.us/economy/calc/cpihome.html>.

Charles Babbage designed the Analytical Engine in the 1830s and continued to refine the concept until his death in

1871. Babbage never completed his invention. I have estimated a date of 1900 for the Analytical Engine as an

estimated date for when its mechanical technology became feasible, based on the availability of other mechanical

computing technology available in that time period.

Sources for the chart ʺThe Exponential Growth of Computing, 1900–1998ʺ include the following:

25 Years of Computer History

<http://www.compros.com/timeline.html>

BYTE Magazine ʺBirth of a Chipʺ

<http://www.byte.com/art/9612/sec6/art2.htm>

cdc.html@www.citybeach.wa.edu (Stretch)

<http://www.citybeach.wa.edu.au/lessons/history/video/sunedu/comp uter/cdc.html>

Chronology of Digital Computing Machines

<http://www.best.com/~wilson/faq/chrono.html>

Chronology of Events in the History of Microcomputers

<http://www3.islandnet.com/~kpolsson/comphist/comp1977.htm>

The Computer Museum History Center

<http://www.tcm.org/html/history/index.html>

delan at intopad.eecs.berkeley.edu

<http://infopad.eecs.berkeley.edu/CIC/summary/delan>

Electronic Computers Within the Ordnance Corps

<http://ftp.arl.mil/~mike/comphist/61ordnance/index.html>

General Processor Information

<http://infopad.eecs.berkeley.edu/CIC/summary/local/>

The History of Computing at Los Alamos

<http://bang.lanl.gov/video/sunedu/computer/comphist.html>

The Machine Room

<http://www.tardis.ed.ac.uk/~alexios/MACHINE‐ROOM/>

Mind Machine Web Museum

<http://userwww.sfsu.edu/~hl/mmm.html>

Hans Moravec at Carnegie Mellon University: Computer Data

<http://www.frc.ri.cmu.edu/~hpm/book97/ch3/processor.list>

PC Magazine Online: Fifteen Years of PC Magazine

<http://www.zdnet.com/pcmag/special/anniversary/>

PC Museum

<http://www.microtec.net/~dlessard/index.html>

PDP‐8 Emulation

<http://csbh.mhv.net/~mgraffam/emu/pdp8.html>

Silicon Graphics Webpage press release

<http://www.pathfinder.com/money/latest/press/PW/1998jun16/270.ht ml >

Stan Augarten, Bit by Bit: An Illustrated History of Computers (New York: Ticknor & Fields, 1984).

International Association of Electrical and Electronics Engineers (IEEE), ʺAnnals of the History of the Computer,ʺ

vol. 9, no. 2, pp. 150–153 (1987). IEEE, vol. 16, no. 3, p. 20 (1994).

Hans Moravec, Mind Children: The Future of Robot and Human Intelligence (Cambridge, MA: Harvard University Press, 1988).

René Moreau, The Computer Comes of Age (Cambridge, MA: MIT Press, 1984).

20. For additional views on the future of computer capacity, see: Hans Moravec , Mind Children: The Future of Robot and Human Intelligence (Cambridge, MA: Harvard University Press, 1988); and ʺAn Interview with David Waltz, Vice President, Computer Science Research, NEC Research Instituteʺ at Think Questʹs web page

<http://tqd.advanced.org/2705/waltz.html>. I also discuss this subject in my book The Age of Intelligent Machines (Cambridge, MA: MIT Press, 1990), 401–419. These three sources discuss the exponential growth of computing.

21. A mathematical theory concerning the difference between information and noise and the ability of a

communications channel to carry information.

22. The Santa Fe institute has played a pioneering role in developing concepts and technology related to complexity and emergent systems. One of the principal developers of paradigms associated with chaos and complexity has been

Stuart Kauffman. Kauffmanʹs At Home in the Universe: The Search for the Laws of Self‐Organization and Complexity (Oxford: Oxford University Press, 1995) looks ʺat the forces for order that lie at the edge of chaosʺ (from the card catalog description).

In his book Evolution of Complexity by Means of Natural Selection (Princeton, NJ: Princeton University Press, 1988), John Tyler Bonner asks the question: ʺHow is it that an egg turns into an elaborate adult? How is it that a bacterium, given many millions of years, could have evolved into an elephant?ʹ

John Holland is another leading thinker from the Sante Fe Institute in the emerging field of complexity. His book

Hidden Order: How Adaptation Builds Complexity (Reading, MA: Addison‐Wesley, 1996) presents a series of lectures that Holland presented at the Santa Fe Institute in 1994.

23. Also see John H. Holland, Emergence: From Chaos to Order (Reading, MA: Addison‐Wesley, 1998) and M. Mitchell Waldrop, Complexity: The Emerging Science at the Edge of Order and Chaos (New York: Simon and Schuster, 1992).

CHAPTER 2: THE INTELLIGENCE OF EVOLUTION

1. In the early 1950s, the chemical composition of DNA was already known. At that time, the important questions

were: How is the DNA molecule constructed? How does DNA accomplish its work? These questions would be

answered in 1953 by James D. Watson and Francis H. C. Crick.

Watson and Crick wrote ʺThe Molecular Structure of Nucleic Acid: A Structure for Deoxyribose Nucleic Acidʺ

published in the April 25, 1953 issue of Nature. For more information on the race by various research groups to discover the molecular structure of DNA, read Watsonʹs book, The Double Helix (New York: Atheneum Publishers, 1968).

2. Translation starts by unwinding a region of DNA to expose its code. A strand of messenger RNA (mRNA) is created

by copying the exposed DNA base‐pair codes. The appropriately named messenger RNA records a copy of a portion

of the DNA letter sequence and travels out of the nucleus into the cell body. There the mRNA encounters a ribosome

molecule, which reads the letters encoded in the mRNA molecules and then, using another set of molecules called

transfer RNA (tRNA), actually builds protein chains one amino acid at a time. These proteins are the worker

molecules that perform the cellʹs functions. For example, hemoglobin, which is responsible for carrying oxygen in

the blood from the lungs to the bodyʹs tissues, is a sequence of 500 amino acids. With each amino acid requiring

three nucleotide letters, the coding for hemoglobin requires 1,500 positions on the DNA molecule. Molecules of

hemoglobin, incidentally, are created 500 trillion times a second in the human body, so the machinery is quite

efficient.

3. The goal of the Human Genome Project is to construct detailed genetic sequence maps of the 50,000 to 100,000 genes in the human genome, and to provide information about the overall structure and sequence of the DNA of humans

and of other animals. The project began in the mid‐1980s. The web site of the Human Genome Project,

<http://www.nhgri.nih.gov/HGP/>, contains ʺinformation on the background of the project, current and future goals, and detailed explanations on the structure of DNA.

4. Thomas Rayʹs work is described in an article by Joe Flower, ʺA Life in Silicon.ʺ New Scientist 150, no. 2034 (June 15, 1996): 32–36. Dr. Ray also has a web site with updates on his software‐based evolution at

<http://www.hip.atr.co.jp/~ray/>.

5. A selection of books exploring the nature of intelligence includes: H. Gardner, Frames of Mind (New York: Basic Books, 1983); Stephen Jay Gould, The Mismeasure of Man (New York: Basic Books, 1983); R. J. Herrnstein and C.

Murray, The Bell Curve (New York: The Free Press, 1994); R. Jacoby and N. Glaubennan, eds., The Bell Curve Debate (New York: Times Books, 1995).

6. To further explore the theories of expansion and contraction of the Universe, see: Stephen W. Hawking, A Brief History of Time (New York: Bantam Books, 1988); and Eric L. Lerner, The Big Bang Never Happened (New York: Random House, 1991). For the latest updates, see the International Astronomical Union (IAU) web site at

<http://www.intastun.org/>, as well as the above noted ʺIntroduction to Big Bang Theoryʺ at

<http://www.bowdoin.edu/dept/physics/astro.1997/astro4/bigbang.ht ml >.

7. See chapter 3, ʺOf Mind and Machines,ʺ including the box ʺThe View from Quantum Mechanics.ʺ

8. Peter Lewis, ʺCan Intelligent Life Be Found? Gorilla Will Go Looking.ʺ New York Times, April 16, 1998.

9. Voice Xpress Plus from the Dictation Division of Lernout & Hauspie Speech Products (formerly Kurzweil Applied Intelligence) allows users to give ʺnatural languageʺ commands to Microsoft Word. It also provides large‐vocabulary

continuous‐speech dictation. The program is ʺmode‐less,ʺ so users do not need to indicate when they are giving

commands. For example, if the user says: ʺI enjoyed my trip to Belgium last week. Make this paragraph four points

bigger. Change its font to Arial. I hope to go back to Belgium soon.ʺ Voice Xpress Plus automatically determines that the second and third sentences are commands and will carry them out (rather than transcribing them). It also

determines that the first and fourth sentences are not commands, and will transcribe them into the document.

CHAPTER 3: OF MIND AND MACHINES

1. To learn more about the current state of brain‐scanning research, the article ʺBrains at Work: Researchers Use New Techniques to Study How Humans Thinkʺ by Vincent Kiernan is a good place to begin. This article, in the Chronicle of Higher Education (January 23, 1998, vol. 44, no. 20, pp. A16–17), discusses uses of MRI to map brain activity during complex thinking processes.

ʺVisualizing the Mindʺ by Marcus E. Raichle in the April 1994 Scientific American provides background on various brain‐imaging technologies: MRI, positron emission tomography (PET), magnetoencephalography (MEG), and

electroencephalography (EEG).

ʺUnlocking the Secrets of the Brainʺ by Tabitha M. Powledge is a two‐part article in the July–August issue of

Bioscience 47 (pp. 330–334 and 403–409), 1997.

2. Blood‐forming cells of the bone marrow and certain layers of the skin grow and reproduce frequently, replenishing themselves in a period of months. In contrast, muscle cells do not reproduce for several years. Neurons have not

been considered to reproduce at all after oneʹs birth, but recent findings indicate the possibility of primate neuron reproduction. Dr. Elizabeth Gould of Princeton University and Dr. Bruce S. McEwen of Rockefeller University in

New York found that adult marmoset monkeys are able to manufacture brain cells in the hippocampus, a brain

region that is connected to learning and memory. Conversely, when the animals are under stress, the ability to

manufacture new brain cells in the hippocampus diminishes. This research is described in an article by Gina Kolata,

ʺStudies Find Brain Grows New Cells,ʺ The New York Times, March 17, 1998.

Other types of cells will grow and reproduce if necessary. For example, if seven‐eighths of the liver cells are

removed, the remaining cells will grow and reproduce until most of the cells are replenished. Arthur Guyton,

Physiology of the Human Body, fifth edition (Phila., PA: W B. Saunders, 1979): 42–43.

3. Oppression of human races, nationalities, and other groups has often been justified in the same way.

4. Platoʹs works are available in Greek and English in the Loeb Classical Library editions. A detailed account of Platoʹs philosophy is presented in J. N. Findlay, Plato and Platonism: An Introduction. On the dialogues as Platoʹs chosen form, see D. Hylandʹs ʺWhy Plato Wrote Dialogues.ʺ Philosophy and Rhetoric 1 (1968): 38–50.

5. A brief history of logical positivism can be found in A. J. Ayer, Logical Positivism (New York: Macmillan, 1959): 3–28.

6. David J. Chalmers distinguishes ʺbetween the easy problems and the hard problem of consciousness,ʺ and argues

that ʺthe hard problem eludes conventional methods of explanation entirelyʺ in an essay entitled ʺFacing Up to the

Problem of Consciousness.ʺ Stuart R. Hameroff, ed., Toward a Science of Consciousness: The First Tucson Discussions and Debates (Complex Adaptive Systems) (Cambridge, MA: MIT Press, 1996).

7. This objective view was systematically defined early in the twentieth century by Ludwig Wittgenstein in an analysis of language called logical positivism. This philosophical school, which would subsequently influence the emergence

of computational theory and linguistics, drew its inspiration from Wittgensteinʹs first major work, the Tractatus Logico‐Philosophicus. The book was not an immediate hit and it took the influence of his former instructor, Bertrand Russell, to secure a publisher.

In a foreshadowing of early computer‐programming languages, Wittgenstein numbered all of the statements in

his Tractatus indicating their position in the hierarchy of his thinking. He starts out with statement 1: ʺThe world is all that is the case,ʺ indicating his ambitious agenda for the book. A typical statement is number 4.0.0.3. 1: ʺAll

philosophy is a critique of language.ʺ His last statement, number 7, is ʺWhat we cannot speak about we must pass

over in silence.ʺ Those who trace their philosophical roots to the early Wittgenstein still regard this short work as the most influential work of philosophy of the past century. Ludwig Wittgenstein, Tractatus Logico‐Philosophicus, translated by D. E Pears and B. E McGuiness, Germany, 1921.

8. In the preface to Philosophical Investigations, translated by G. E. M. Anscombe, Wittgenstein ʺacknowledgesʺ that he made ʺgrave mistakesʺ in his earlier work, the Tractatus.

9. For a useful overview of Descartesʹs life and work, see The Dictionary of Scientific Biography, vol. 4, pp. 55–65. Also, Jonathan Réeʹs Descartes presents a unified view of Descartesʹs philosophy and its relation to other systems of thought.

10. Quoted from Douglas R. Hofstadter, Gödel, Escher Bach: An Eternal Golden Braid (New York: Basic Books, 1979).

11. ʺComputing Machinery and Intelligence,ʺ Mind 59 (1950); 433–460, reprinted in E. Feigenbaum and J. Feldman, eds., Computers and Thought (New York: McGraw‐Hill, 1963).

12. For a description of quantum mechanics, read George Johnson, ʺQuantum Theorists Try to Surpass Digital

Computing,ʺ New York Times, February 18, 1997.

CHAPTER 4: A NEW FORM OF INTELLIGENCE ON EARTH

1. Simple calculating devices had been perfected almost two centuries before Babbage, starting with Pascalʹs Pascaline in 1642, which could add numbers, and a multiplying machine developed by Gottfried Wilhelm Leibniz a couple of

decades later. But automating the computing of logarithms was far more ambitious than anything that had been

previously attempted.

Babbage didnʹt get very far—he exhausted his financial resources, got into a dispute with the British government

over ownership, had problems getting the unusual precision parts fabricated, and saw his chief engineer fire all of

his workmen and then quit himself. He was also beset with personal tragedies, including the death of his father, his wife, and two of his children.

The only obvious thing to do now, Babbage figured, was to abandon his ʺDifference Engineʺ and embark on

something yet more ambitious: the worldʹs first fully programmable computer. Babbageʹs new conception—the

ʺAnalytical Engineʺ—could be programmed to solve any possible logical or computational problem.

The Analytical Engine had a random‐access memory (RAM) consisting of 1,000 ʺwordsʺ of 50 decimal digits each,

equivalent to about 175,000 bits. A number could be retrieved from any location, modified, and stored in any other

location. It had a punched‐card reader and even included a printer, even though it would be another half century

before either typesetting machines or typewriters were to be invented. It had a central processing unit (CPU) that

could perform the types of logical and arithmetic operations that CPUS do today. Most important, it had a special

storage unit for the software with a machine language very similar to those of todayʹs computers. One decimal field

specified the type of operation and another specified the address in memory of the operand. Stan Augarten, Bit by Bit: An Illustrated History of Computers (New York: Ticknor and Fields, 1984): 63–64.

Babbage describes the features of his machine in ʺOn the Mathematical Powers of the Calculating Engine,ʺ

written in 1837 and reprinted as appendix B in Anthony Hymans Charles Babbage: Pioneer of the Computer (Oxford: Oxford University Press, 1982). For biographical information on Charles Babbage and Ada Lovelace, see Hymanʹs

biography and Dorothy Steinʹs book Ada: A Life and a Legacy (Cambridge, MA: MIT Press, 1985).

2. Stan Augarten, Bit by Bit, 63–64. Babbageʹs description of the Analytical Engine in ʺOn the Mathematical Powers of the Calculating Engine,ʺ written in 1837, is reprinted as appendix B in Anthony Hymanʹs Charles Babbage: Pioneer of the Computer (Oxford: Oxford University Press, 1982).

3. Joel Shurkin, in Engines of the Mind, p. 104, describes Aikenʹs machine as ʺan electro‐mechanical Analytical Engine with IBM card handling.ʺ For a concise history of the development of the Mark I, see Augartenʹs Bit by Bit, 103–107.

1. Bernard Cohen provides a new perspective on Aikenʹs relation to Babbage in his article ʺBabbage and Aiken,ʺ

Annals of the History of Computing 10 (1988): 171–193.

4. The idea of the punched card, which Babbage borrowed from the Jacquard looms (automatic weaving machines

controlled by punched metal cards), also survived and formed the basis for automating the increasingly popular

calculators of the nineteenth century. This culminated in the 1890 U.S. census, which was the first time that

electricity was used for a major data‐processing project. The punched card itself survived as a mainstay of

computing until the 1970s.

5. Turingʹs Robinson was not a programmable computer. It didnʹt have to be—it had only one job to do. The first

programmable computer was developed by the Germans. Konrad Zuse, a German civil engineer and tinkerer, was

motivated to ease what he later called those ʺawful calculations required of civil engineers.ʺ Like Babbageʹs, his first device, the Z‐1, was entirely mechanical—built from an erector set in his parentsʹ living room. The Z‐2 used

electromechanical relays and was capable of solving complex simultaneous equations. It was his third version—the

Z‐3—that is the most historic. It stands as the worldʹs first programmable computer. As one would retroactively predict from the Law of Accelerating Returns as applied to computation, Zuseʹs Z‐3 was rather slow—a

multiplication took more than three seconds.

While Zuse received some incidental support from the German government and his machines played a minor

military role, there was little, if any, awareness of computation and its military significance by the German

leadership. This explains their apparent confidence in the security of their Enigma code. Instead the German military gave immensely high priority to several other advanced technologies, such as rocketry and atomic weapons.

It would be Zuseʹs fate that no one would pay much attention to him or his inventions; even the Allies ignored

him after the end of the war. Credit for the worldʹs first programmable computer is often given to Howard Aiken,

despite the fact that his Mark I was not operational until nearly three years after the Z‐3. When Zuseʹs funding was withdrawn in the middle of the war by the Third Reich, a German officer explained to him that ʺthe German aircraft

is the best in the world. I cannot see what we could possibly calculate to improve on.ʺ

Zuseʹs claim to having built the worldʹs first operational fully programmable digital computer is supported by

the patent application he filed. See, for instance, K. Zuse, ʺVerfahren zur Selbst Atigen Durchfurung von

Rechnungen mit Hilfe von Rechenmaschinen,ʺ German Patent Application Z23624, April 11, 1936. Translated

extracts, titled ʺMethods for Automatic Execution of Calculations with the Aid of Computers,ʺ appear in Brian

Randell, ed., The Origins of Digital Computers, pp. 159–166.

6. ʺComputing Machinery and Intelligence,ʺ Mind 59 (1950): 433–460, reprinted in E. Feigenbaum and J. Feldman, eds., Computers and Thought (New York: McGraw‐Hill, 1963).

7. See A. Newell, J. C. Shaw, and H. A. Simon, ʺProgramming the Logic Theory Machine,ʺ Proceedings of the Western Joint Computer Conference, 1957, pp. 230–240:

8. Russell and Whiteheadʹs Principia Mathematica (see reference at the end of this endnote), first published in 1910–1913, was a seminal work that reformulated mathematics based on Russellʹs new conception of set theory. Russellʹs

breakthrough in set theory set the stage for Turingʹs subsequent development of computational theory based on the

Turing machine (see note below). Following is my version of ʺRussellʹs paradox,ʺ which stimulated Russellʹs

discovery:

Before ending up in ʺthe Other Place,ʺ our friend the gambler had lived a rough life. He was short of

temper and not fond of losing. In our story, he is also a bit of a logician. This time he has picked the wrong

man to dispatch. If only he had known that the fellow was the judgeʹs nephew.

Known anyway as a hanging judge, the magistrate is furious and wishes to mete out the most severe

sentence he can think of. So he tells the gambler that not only is he sentenced to die but the sentence is to

be carried out in a unique way. ʺFirst off, weʹre gonna dispense with you quickly, just like you done with

the victim. This punishment must be carried out no later than Saturday. Furthermore, I donʹt want you

preparing yourself for the judgment day. On the morning of your execution, you wonʹt know for certain

that the day is at hand. When we come for you, itʹll be a surprise.ʺ

To which the gambler replies, ʺWell, thatʹs great, judge, I am greatly relieved.ʺ

To which the judge exclaims, ʺI donʹt understand, how can you be relieved? I have condemned you to

be executed. I have ordered that the sentence be carried out soon, but youʹll be unable to prepare yourself

because on the morning that we carry it out, you wonʹt know for certain that youʹll be dying that day.ʺ

ʺWell, Your Honor,ʺ the gambler points out, ʺin order for your sentence to be carried out, I cannot be

executed on Saturday.ʺ

ʺWhy is that?ʺ asks the judge.

ʺBecause since the sentence must be carried out by Saturday, if we actually get to Saturday, I will know

for certain that I am to be executed on that day, and thus it would not be a surprise.ʺ

ʺI suppose you are right,ʺ replies the judge. ʺYou cannot be executed on Saturday But I still donʹt see

why youʹre relieved.ʺ

ʺWell, if we have definitely ruled out Saturday, then I canʹt be executed on Friday either.ʺ

ʺWhy is that?ʺ asks the judge, being a little slow.

ʺWe have agreed that I canʹt be executed on Saturday therefore Friday is the last day I can be executed.

But if Friday rolls around, I will definitely know that I am to be executed on that day and therefore it

would not be a surprise. So I canʹt be executed on Friday.ʺ

ʺI see,ʺ says the judge.

ʺThus the last day I can be executed would be Thursday But if Thursday rolls around, I would know I

had to be executed on that day, and thus it would not be a surprise. So Thursday is out. By the same

reasoning, we can eliminate Wednesday, Tuesday, Monday, and today.ʺ

The judge scratches his head as the confident gambler is led back to his prison cell.

There is an epilogue to the story. On Thursday, the gambler is taken to be executed. And he is very surprised. So

the judgeʹs orders are successfully carried out.

This is my version of what has become known as ʺRussellʹs paradoxʺ after Bertrand Russell, perhaps the last

person to secure major achievements in both mathematics and philosophy. If we analyze this story, we see that the

conditions that the judge has set up result in a conclusion that none of the days comply, because, as the prisoner so adroitly points out, each one of them in turn would not be a surprise. But the conclusion itself changes the situation, and now surprise is possible again. This brings us back to the original situation in which the prisoner could (in

theory) demonstrate that each day in turn would be impossible, and so on, ad infinitum. The judge applies

ʺAlexanderʹs solutionʺ in which King Alexander slashed the hopelessly tied Gordian knot.

A simpler example, and the one that Russell actually struggled with, is the following question about sets. A set is

a mathematical construct that, as its name implies, is a collection of things. A set may include chairs, books, authors, gamblers, numbers, other sets, themselves, whatever. Now consider set A, which is defined to contain all sets that

are not members of themselves. Does set A contain itself?

As we consider this famous problem, we realize there are only two possible answers: Yes and No. We can,

therefore, try them all (this is not the case for most problems in mathematics). So letʹs consider Yes. If the answer is Yes, then set A does contain itself. But if set A contains itself, then according to its defining condition, set A would not belong to set A, and thus it does not belong to itself. Since the answer of Yes led to a contradiction, it must be wrong.

So letʹs try No. If the answer is No, then set A does not contain itself. But again according to the defining

condition, if set A does not belong to itself, then it would belong to set A, another contradiction. As with the story about the prisoner, we have incompatible propositions that imply one another. Yes implies No, which yields Yes,

and so on.

This may not seem like a big deal, but to Russell it threatened the foundation of mathematics. Mathematics is

based on the concept of sets, and the issue of inclusion (i.e., what belongs to a set) is fundamental to the idea. The definition of set A appears to be a reasonable one. The question of whether set A belongs to itself also appears

reasonable. Yet we have difficulty coming up with a reasonable answer to this reasonable question. Mathematics

was in big trouble.

Russell pondered this dilemma for more than a decade, nearly exhausting himself and wrecking at least one

marriage. But he came up with an answer. To do so, he invented the equivalent of a theoretical computer (although

not by name). Russellʹs ʺComputerʺ is a logic machine and it implements one logical transformation at a time, each

one requiring a quantum of time—so things donʹt happen all at once. Our question about set A is examined in an

orderly fashion. Russell turns on his theoretical computer (which, lacking a real computer, ran only in his head) and the logical operations are ʺexecutedʺ in turn. So at one point, our answer is Yes, but the program keeps running, and a few quantums of time later the answer becomes No. The program runs in an infinite loop, constantly alternating

between Yes and No.

But the answer is never Yes and No at the same time!

Impressed? Well Russell was very pleased. Eliminating the possibility of the answer being Yes and No at the same time was enough to save mathematics. With the help of his friend and former tutor Alfred North Whitehead, Russell recast all of mathematics in terms of his new theory of sets and logic, which they published in their Principia

Mathematica in 1910–1913. It is worth pointing out that the concept of a computer, theoretical or otherwise, was not widely understood at the time. The nineteenth‐century efforts of Charles Babbage, which are discussed in chapter 4,

were largely unknown at the time. It is not clear if Russell was aware of Babbageʹs efforts. Russellʹs highly influential and revolutionary work invented a logical theory of computation and recast mathematics as one of its branches.

Mathematics was now part of computation.

Russell and Whitehead did not explicitly talk about computers but cast their ideas in the mathematical

terminology of set theory. It was left to Alan Turing to create the first theoretical computer in 1936, in his Turing machine (see note 16 below).

Alfred N. Whitehead and Bertrand Russell, Principia Mathematica, 3 vols., second edition (Cambridge: Cambridge University Press, 1925–1927). (The first edition was published in 1910, 1912, and 1913.)

Russellʹs paradox was first introduced in Bertrand Russell, Principles of Mathematics (Reprint, New York: W. W.

Norton & Company, 1996), 2nd ed., 79–81. Russellʹs paradox is a subtle variant of the Liar Paradox. See E. W. Beth, Foundations of Mathematics (Amsterdam: North Holland, 1959), p. 485.

9. ʺHeuristic Problem Solving: The Next Advance in Operations Research,ʺ Journal of the Operations Research Society of America 6, no. 1 (1958), reprinted in Herbert Simon, Models of Bounded Rationality, vol. 1, Economic Analysis and Public Policy (Cambridge, MA: MIT Press, 1982).

10. ʺA Mean Chess‐Playing Computer Tears at the Meaning of Thought,ʺ New York Times, February 19, 1996, contains the reactions of Gary Kasparov and a number of noted thinkers concerning the ramifications of Deep Blue beating

the world chess champion.

11. Daniel Bobrow, ʺNatural Language Input for a Computer Problem Solving System,ʺ in Marvin Minsky, Semantic

Information Processing, pp. 146–226.

12. Thomas Evans, ʺA Program for the Solution of Geometric‐Analogy Intelligence Test Questions,ʺ in Marvin Minsky,

ed., Semantic Information Processing (Cambridge, MA: MIT Press, 1968), pp. 271–353.

13. Robert Lindsay, Bruce Buchanan, Edward Feigenbaum, and Joshua Lederberg describe DENDRAL in Applications of Artificial Intelligence for Chemical Inference: The DENDRAL Project (New York: McGraw‐Hill, 1980). For a brief and clear explanation of the essential mechanisms behind DENDRAL, see Patrick Winston, Artificial Intelligence (1984), pp. 1637164, 195–197.

14. For many years SHRDLU was cited as a prominent accomplishment of artificial intelligence. Winograd describes his research in his thesis Understanding Natural Language (New York: Academic Press, 1972). A brief version appears as ʺA Procedural Model of Thought and Language,ʺ in Roger Schank and Kenneth Colby, eds., Computer Models of

Thought and Language (San Francisco: W. H. Freeman, 1973).

15. Haneef A. Fatmi and R. W. Young, ʺA Definition of Intelligence,ʺ Nature 228 (1970): 91.

16. Alan Turing showed that the essential basis of computation could be modeled with a very simple theoretical

machine. He created the first theoretical computer in 1936 (first introduced in Alan M. Turing, ʺOn Computable

Numbers with an Application to the Entschemungs Problem,ʺ Proc. London Math. Soc. 42 [1936]: 230–265) in an eponymous conception called the Turing machine. As with a number of Turingʹs breakthroughs, he would have both

the first and last word. The Turing machine represented the founding of modern computational theory. It has also

persisted as out primary theoretical model of a computer because of its combination of simplicity and power.

The Turing machine is one example of the simplicity of the foundations of intelligence. A Turing machine

consists of two primary (theoretical) units: a tape drive and a computation unit. The tape drive has a tape of infinite length on which it can write, and (subsequently) read, a series of two symbols: zero and one. The computation unit

contains a program consisting of a sequence of commands, drawing from only seven possible operations:

• Read the tape

• Move the tape left one symbol

• Move the tape right one symbol

• Write 0 on the tape

• Write 1 on the tape

• Jump to another command

• Halt

Turing was able to show that this extremely simple machine can compute anything that any machine can

compute, no matter how complex. If a problem cannot be solved by a Turing machine, then it cannot be solved by

any machine. Occasionally there are challenges to this position, but in large measure it has stood the test of time.

In the same paper, Turing reports another unexpected discovery, that of unsolvable problems. These are

problems that are well defined with unique answers that can be shown to exist, but that we can also prove can never

be computed by a Turing machine—that is to say by any machine, yet another reversal of what had been a

nineteenth‐century confidence that problems that could be defined would ultimately be solved. Turing showed that

there are as many unsolvable problems as solvable ones.

Turing and Alonzo Church, his former professor, went on to assert what has become known as the Church‐

Turing thesis: If a problem that can be presented to a Turing machine is not solvable by one, then it is also not

solvable by human thought. ʺStrongʺ interpretations of the Church‐Turing thesis propose an essential equivalence

between what a human can think or know and what is computable by a machine. The Church‐Turing thesis can be

viewed as a restatement in mathematical terms of one of Wittgensteinʹs primary theses in his Tractatus. The basic

idea is that the human brain is subject to natural law, and thus its information‐processing ability cannot exceed that of a machine. We are thus left with the perplexing situation of being able to define a problem, to prove that a unique answer exists, and yet know that the answer can never be known.

Perhaps the most interesting unsolvable problem is called the Busy Beaver, which may be stated as follows: Each

Turing machine has a certain number of commands in its program. Given a positive integer n, we construct all of the

Turing machines that have n states (i.e., n commands). Next we eliminate those n‐state Turing machines that get into an infinite loop (i.e., never halt). Finally, we select the machine (one that halts) that writes the largest number of 1s on its tape. The number of Is that this Turing machine writes is called busy beaver of n.

Tibor Rado, a mathematician and admirer of Turing, showed that there is no algorithm. (that is, no Turing

machine) that can compute the busy beaver function for all nʹs. The crux of the problem is sorting out those n‐state Turing machines that get into infinite loops. If we program a Turing machine to generate and simulate every

possible n‐state Turing machine, this simulator itself goes into an infinite loop when it attempts to simulate one of the n‐state Turing machines that gets into an infinite loop. Busy beaver can be computed for some ns, and

interestingly it is also an unsolvable problem to separate those ns for which we can determine busy beaver of n from those for which we cannot.

Busy beaver is an ʺintelligent function.ʺ More precisely stated, it is a function that requires increasing intelligence to compute for increasing arguments. As we increase n, the complexity of the processes needed to compute busy

beaver of n increases.

With n = 6, we are dealing with addition and busy beaver of 6 equals 35. In other words, addition is the most

complex operation that a Turing machine with only 6 steps in its program is capable of performing. At 7, busy

beaver learns to multiply and busy beaver of 7 equals 22,961. At 8, busy beaver can exponentiate, and the number of

1s that our eighth busy beaver writes on its tape is approximately 1043. Note that this is even faster growth than

Mooreʹs Law. By the time we get to 10 we need an exotic notation in which we have a stack of exponents (10 to the

10 to the 10, etc.), the height of which is determined by another stack of exponents, the height of which is determined by another stack of exponents, and so on. For the twelfth busy beaver we need an even more exotic notation. Human

intelligence (in terms of the complexity of mathematical operations that we can understand) is surpassed well before the busy beaver gets to 100. The computers of the twenty‐first century will do a bit better.

The busy beaver problem is one example of a large class of noncomputable functions, as one can see from Tibor

Rado, ʺOn Noncomputable Functions,ʺ Bell System Technical Journal 41, no. 3 (1962): 877–884.

17. Raymond Kurzweil, The Age of Intelligent Machines (Cambridge, MA: MIT Press, 1990), pp. 132–133.

18. H. J. Berliner, ʺBackgammon Computer Program Beats World Champion,ʺ Artificial Intelligence 14, no. 1 (1980). Also see Hans Berliner, ʺComputer Backgammon,ʺ Scientific American, June 1980.

19. To download Ray Kurzweilʹs Cybernetic Poet (RKCP), go to: <http://www.kurzweiltech.com>. RKCP is further

discussed in the section The Creative Machine in chapter 8, ʺ1999.ʺ

20. See the discussion on these music composition programs in the section The Creative Machine in chapter 8, ʺ1999.ʺ

21. See W. S. Sarle, ed., ʺNeural Network Frequently Asked Questions,ʺ <ftp://ftp.sas.com/pub/neural/FAQ.html>. This web site has, numerous resources on past and current research on neural nets. G. E. Hintonʹs ʺHow Neural Networks

Learn from Experience,ʺ in the September 1992 issue of Scientific American (144–151), also provides a good

introduction to neural networks.

22. Researchers at the Productivity from Information Technology (PROFIT) Initiative at MIT have studied the

effectiveness of neural networks in understanding handwriting.

The PROFIT Initiative is based at MITʹs Sloan School of Management. The mission of the initiative is to study

how the private and public sectors use information technology. Abstracts of working papers on this and other

research on neural networks and data mining can be found at <http://scanner‐group.mit.edu/papers.html>.

23. ʺMiros, Inc. is located in Wellesley, Massachusetts, and specializes in providing face recognition software. Miros I products include Trueface PC, the first face recognition solution for computer, network and data security; and

Trueface Gatewatch, a complete hardware/software security solution that allows or denies access to buildings and

rooms by automatically recognizing a personʹs face taken by a video camera.ʺ From Miros Company Information at

<http://www.miros.com/About_Miros.htm>.

24. For more information on Brainmakerʹs aptitude to diagnose illnesses, and to predict the Standard and Poor 500 for LBS Management, see California Scientificʹs home page at <http://www.calsci.com/>.

25. The reset time stated here is an estimated average for neural connection calculations. For example, Vadim

Gerasimov estimates the peak firing frequency of neurons (which significantly exceeds the average rate) to be 250–

2,000 Hz (0.5–4 ms intervals) in ʺInformation Processing in the Human Bodyʺ at

<http://vadim.www.media.mit.edu/MAS862/Project.html>. The firing time is affected by a number of variables,

including, for example, the level and duration of a sound, as discussed in Jos. J. Eggermont, ʺFiring Rate and Firing Synchrony Distinguish Dynamic from Steady State Sound,ʺ Neuroreport 8, issue 12, 2709–2713.

26. Hugo de Garis maintains a web site on his research for ATRS Brain Builder Group at

<http://www.hip.atr.co.jp/~degaris/>.

27. For an intriguing account of this research, read Carver Mead, Analog VSLI and Neural Systems (Reading, MA: Addison‐Wesley, 1989), 257–278. Synaptics is briefly highlighted in Carol Levin, ʺHereʹs Looking at You,ʺ PC

Magazine (December 20, 1994): 31. Carver Meadʹs web site also provides detailed information on this research at the ʺPhysics of Computation–Carver Meadʹs Groupʺ at <http://www.pcmp.caltech.edu/>.

28. The SETI (Search for Extraterrestrial Intelligence) Institute conducts research on other signs of life in the Universe, its primary goal being the search for extraterrestrial intelligence. The institute is a nonprofit research organization, funded by government agencies, private foundations, and individuals, which in turn provides funding for several

dozen projects. For more information, see the SETI Institute web site, <http://www.seti.org>.

29. The author is dictating portions of this book to his computer through the continuous speech recognition program called Voice Xpress Plus from the dictation division of Lernout & Hauspie (formerly Kurzweil Applied Intelligence).

See note 9 on Voice Xpress Plus in chapter 2 for more information.

30. To find out more on State Street Global Advisorʹs purchase in a majority stake in Advanced Investment Technology, read Frank Byrt, ʺState Street Global Invests in Artificial Intelligence.ʺ Dow Jones Newswires, October 29, 1997. The genetic algorithm system used by the AIT Vision mutual fund is described in S. Mahfoud and G. Mani, ʺFinancial

Forecasting Using Genetic Algorithms.ʺ Applied Artificial Intelligence 10 (1996): 543–565. The AIT Vision mutual fund opened at the beginning of 1996 and has publicly available performance numbers. In its first full calendar year

(1996), the mutual fund increased 27.2 percent in net asset value, compared to 21.2 percent for its benchmark, the

Russell 3000 index.

It should be noted that outperforming its benchmark index does not in itself prove a superior level of decision

making. The algorithm may have been making higher‐risk investments (on average) than the average in the index.

31. There are many online resources on evolutionary computation and evolutionary and genetic algorithms. One of the best is ʺThe Hitchhikerʹs Guide to Evolutionary Computation: A List of Frequently Asked Questions (FAQ),ʺ edited

by Jorg Heitkotter and David Beasley at <http://www.cs.purdue.edu/coast/archive/clife/FAQ/www/>. This guide

includes everything from a glossary to links to various research groups.

Another helpful online resource is the web site for the Santa Fe Institute. The instituteʹs web site can be accessed at <http://www.santafe.edu>.

For an offline introduction to genetic algorithms, read John Hollandʹs article ʺGenetic Algorithms,ʺ Scientific

American 267, no. 1 (1992): 66–72. As mentioned in note 22 in chapter 1, Holland and his colleagues at the University of Michigan developed genetic algorithms in the 1970s.

For more information on the use of genetic algorithm technology to manage the development and manufacturing

of Volvo trucks, read Srikumar S. Rao, ʺEvolution at Warp Speed,ʺ Forbes 161, no. 1 (January 12, 1998): 82–83.

See also note 22 on complexity in chapter 1.

32. See ʺInformation Processing in the Human Body,ʺ by Vadim Gerasimov, at

<http://vadim.www.media.mit.edu/MAS862/Project.html>.

33. See ʺInformation Processing in the Human Body,ʺ by Vadim Gerasimov, at

<http://vadim.www.media.mit.edu/MAS862/Project.html>.

34. I founded Kurzweil Applied Intelligence (Kurzweil AI) in 1982. The company is now a subsidiary of Lernout & Hauspie Speech Products (L&H), an international leader in the development of speech and language technologies

and related applications and products. For more information about these speech recognition products, see

<http://www.lhs.com/dictation/>.

CHAPTER 5: CONTEXT AND KNOWLEDGE

1. Victor L. Yu, Lawrence M. Fagan, S. M. Wraith, William Clancey, A. Carlisle Scott, John Hannigan, Robert Blum,

Bruce Buchanan, and Stanley Cohen, ʺAntimicrobial Selection by Computer: A Blinded Evaluation by Infectious

Disease Experts,ʺ Journal of the American Medical Association 242, no. 12 (1979): 1279–1282.

2. For an introduction to the development of expert systems and their use in various companies, read: Edward

Feigenbaum, Pamela McCorduck, and Penny Nii, The Rise of the Expert Company (Reading, MA: Addison‐Wesley,

1983).

3. William Martin, Kenneth Church, and Ramesh Patil, ʺPreliminary Analysis of a Breadth‐First Parsing Algorithm:

Theoretical and Experiential Results.ʺ MIT Laboratory for Computer Science, Cambridge MA, 1981. In this

document, Church cites the synthetic sentence:

ʺIt was the number of products of products of products of products of products of products of products of

products?ʺ as having 1,430 syntactically correct interpretations.

He cites the following sentence:

ʺWhat number of products of products of products of products of products of products of products of

products was the number products of products of products of products of products of products of

products of products?ʺ as having 1,430 U 1,430 = 2,044,900 interpretations.

4. These and other theoretical aspects of computational linguistics are covered in Mary D. Harris, Introduction to Natural Language Processing (Reston, VA: Reston Publishing Co., 1985).

CHAPTER 6: BUILDING NEW BRAINS . . .

1. Hans Moravec is likely to make this argument in his 1998 book Robot: Mere Machine to Transcendent Mind (Oxford University Press; not yet available as of this writing).

2. One hundred fifty million calculations per second for a 1998 personal computer doubling twenty‐seven times by the year 2025 (this assumes doubling both the number of components, and the speed of each component every two

years) equals about 20 million billion calculations per second. In 1998, it takes multiple calculations on a

conventional personal computer to simulate a neural‐connection calculation. However, computers by 2020 will be

optimized for the neural‐connection calculation (and other highly repetitive calculations needed to simulate neuron

functions). Note that neural‐connection calculations are simpler and more regular than the general‐purpose

calculations of a personal computer.

3. Five billion bits per $1,000 in 1998 will be doubled seventeen times by 2023, which is about a million billion bits for $1,000 in 2023.

4. NECʹs goals to build a supercomputer with a maximum performance of more than 32 teraflops is chronicled in ʺNEC

Begins Designing Worldʹs Fastest Computer,ʺ Newsbytes News Network, January 21, 1998, located online at

<http://www.nb‐pacifica.com/headline/necbeginsdesigningwo_1208.shtml>.

In 1998, IBM was one of four companies chosen to participate in Pathforward, an initiative from the Department

of Energy to develop supercomputers for the twenty‐first century. Other companies involved in the project are

Digital Equipment Corporation; Sun Microsystems, Inc.; and Silicon Graphics/Cray Computer Systems (SGI/Cray).

Pathforward is part of the Accelerated Strategic Computing Initiative (ASCI). For more information on this initiative, see <http://www.llnl.gov/asci>.

5. By harnessing the accelerating improvement in both density of components and speed of components, computer

power will double every twelve months, or a factor of one thousand every ten years. Based on the projection of

$1,000 of computing being equal to the estimated processing power of the human brain (20 million billion

calculations per second) by the year 2020, we get a projection of $1,000 of computing being equal to a million human brains in 2040, a billion human brains in 2050, and a trillion human brains in 2060.

6. By 2099, $1,000 of computing will equal 1024 times the processing power of the human brain. Based on an estimate of 10 billion persons, that is 1014 times the processing power of all human brains. Thus one penny of computing will

equal 109 (one billion) times the processing power of all human brains.

7. In the Punctuated Equilibrium theories, evolution is seen to progress in sudden leaps followed by periods of relative stability. Interestingly, we often see similar behavior in the performance of evolutionary algorithms (see chapter 4).

8. Dean Takahashi, ʺSmall Firms Jockeying for Position in 3D Chip Market,ʺ Knight‐Ridder/Tribune News Service, September 21, 1994, p. 0921K4365.

9. The entire February 1998 issue of Computer (vol. 31, no. 2) explores the status of optical computing and optical storage methods.

Sunny Bains writes of companies using optical computing for fingerprint recognition and other applications in

ʺSmall, Hybrid Digital/Electronic Optical Correlators Ready to Power Commercial Products: Optical Computing