Darwin Among the Machines (41 page)

There are two approaches to embodying intelligence, whether as a brain or as a brazen head. The alternatives correspond to two different approaches to building a boat. To build a kayak, you assemble a skeleton and then give it a skin that allows it to float, just as the architectural framework of a computer is fitted, by evolution or by design, with an envelope of code. To build a dugout, you grow a tree
and then remove everything, one chip at a time, except the boat. This is how nature creates her intelligences, by spawning an overwhelming surplus of neurons and then selectively pruning them to leave a network that, if all goes well, becomes a mind. As computers are replicated by the millions, they are aggregating into structures whose design bears nature's signature in addition to our own.

Within this computational matrix, whether viewed as code subsisting on processors or processors subsisting on code, organization is arrived at as much by chance as by design. Most of the connections make no sense, and few make any money except in circuitous ways. Critics say that the World Wide Web is a passing stage (right) and doomed to failure because it is so pervaded by junk (wrong). Yes, most links will be relegated to oblivion, but this wasteful process, like many of nature's profligacies, will leave behind a structure that could not otherwise have taken form. “I could not discover the mechanism of this steady internal evolution,” wrote Stapledon in pondering the mentality of the nebulae. “But one point seemed to me certain. Natural selection played an important part within each nebula, favoring some experiments in vital organization and destroying others.”
⁹
The World Wide Web, a primitive metabolism nourished by the substance of the Internet, will be succeeded by higher forms of organization feeding upon the substance of the World Wide Web.

The prevailing approach to artificial biology views computers as terraria in which digital creatures are evolved. This assumption fits conveniently within the laboratory, but as a model of the digital universe it provides only a silhouette. Robert Davidge, of the University of Sussex, reversed the perspective in a paper titled “Processors as Organisms,” published in 1992. He suggested that we “shift our view from the programs formed in the memory to the processor itself. . . . Consider the actual processor to be the organism and the memory of instructions to be its environment for exploration. . . . The static computer processor requesting instructions down the memory bus can become an organism moving through the memory. It is the same procedure, but it entirely changes the way we regard the results.”
¹⁰
Both perspectives are essential to understanding the origins of artificial life.

“If we choose to think in the frame of mind of a biologist then we can begin to see biological types of process in the artificial creations that surround us,” explained Davidge. But a computer's movement through its memory is a one-dimensional process. “To be of any interest biologically,” Davidge noted, “the memory must be 2-dimensional not 1-dimensional as in all standard stored-program
computers. . . . If we get rid of the idea that the processor exists to execute our program then we can let it move in a 2-D or 3-D space of instructions and the motional behaviour will become a continuous track through space.”
¹¹
Shortly after Davidge published his speculations this two- or three-dimensional memory and instruction space was realized, suddenly, in the form of the World Wide Web. The Web allows code to move freely through the visible universe of processors, and it allows processors to move freely through the visible universe of code. The result is more than the sum of the parts. “Life,” as Samuel Butler observed in 1887, “is two and two making five.”
¹²

Coding and processing, like matter and energy in conventional physics, are related manifestations of an underlying field. This computational field is observed and measured in bits. A bit is the fundamental unit of information—the difference between two discernible alternatives, perceived as change or choice. The computational universe and the universe of time and space in which we live intersect by means of two kinds of bits: bits that represent a difference between two things at the same time; and bits that represent a difference between one thing at two different times.

The power of computers—whether the strictly regulated machinations of a Turing machine or the amorphous intelligence residing in our heads—derives from their ability to form maps between sequence, arranged in time, and structure, arranged in space. Memory and recall, no matter what their form, are translations between these two species of bits. “Memory locations,” according to Danny Hillis, “are just wires turned sideways in time.”
¹³
In this correspondence between sequence and structure lies the basis not only of computation and memory, but also of organic life, based on the translation of sequences (nucleotides) into structures (proteins), with natural selection the mechanism for debugging the source code and translating improvements back from the structure of the organism to the sequence of its genes. Computers are speeding the process up.

In Alan Turing's minimal example, translation between structure and sequence is executed one bit at a time. The Turing machine scans one square on its tape, reads one bit of information, makes a corresponding change in its state of mind, and, in accordance with its instructions, writes or erases one bit of information on its tape. When the next moment arrives, it goes through this process again. The Turing machine and its visible universe cross paths one symbol at a time.

As bandwidth measures the capacity to communicate information from one place to another, so it is possible to assign a magnitude to the amount of information that a Turing machine, or other organism,
is able to scan as it moves from one moment to the next. Extending Turing's terminology, we may call this the machine's depth of mental field. Going one step further, it is possible to quantify how much information can be scanned at a single moment, multiplied by the number of consecutive moments that can be comprehended within the machine's state of mind. This scale of mental capacity allows us to make comparisons between minds that may be operating at completely alien velocities in time.

In the four-dimensional universe of space and time, we are confined to a three-dimensional surface. Only one moment exists in our reality; the existence of other moments is evident only through the constructs of our mind. We breathe one lifetime at a time within a thin atmosphere condensed out of the surrounding sequence of events. All our devices for translating between sequence and structure serve to extend this atmospheric depth—the history of life on earth is compressed within sequential strands of DNA; culture is accumulated in the form of language; somehow our brains preserve the sequence of our lives from one moment to the next. As far as we know mind and intelligence exist on an open-ended scale. Perhaps mind is a lucky accident that exists only at our particular depth of field, like some alpine flower that blooms between ten thousand and twelve thousand feet. Or perhaps there is mind at elevations both above and below our own.

“It may be that the cells of which we are built up . . . each one of them with a life and memory of its own . . . reckon time in a manner inconceivable to us,” suggested Samuel Butler in 1877. “If, in like manner, we were to allow our imagination to conceive the existence of a being as much in need of a microscope for our time and affairs as we for those of our own component cells, the years would be to such a being but as the winkings or the twinklings of an eye.” Writing in
Life and Habit
on the eve of his estrangement from Charles Darwin, Butler sought to encompass Darwinism—from the life of germs to the germination of species—within a framework of all-pervasive mind. “What I wish is, to make the same sort of step in an upward direction which has already been taken in a downward one, and to show reason for thinking that we are only component atoms of a single compound creature,
LIFE
, which has probably a distinct conception of its own personality though none whatever of ours.”
¹⁴

Olaf Stapledon believed that the mind of the individual and the mind of the species need not remain estranged. “Our experience was enlarged not only spatially but temporally in a very strange manner,” explains the narrator of
Last and First Men
, concerning the composite
mind toward which our species had inexorably evolved. “In respect of temporal perception, of course, minds may differ in two ways, in the length of the span which they can comprehend as ‘now,' and in the minuteness of the successive events which they can discriminate within the ‘now.' As individuals we can hold within one ‘now' a duration equal to the old terrestrial day; and within that duration, we can if we will, discriminate rapid pulsations such as commonly we hear together as a high musical tone. As the race-mind we perceived as ‘now' the whole period since the birth of the oldest living individuals, and the whole past of the species appeared as personal memory, stretching back into the mist of infancy. Yet we could, if we willed, discriminate within the ‘now' one light-vibration from the next.”
¹⁵

Both Samuel Butler and Olaf Stapledon saw that mind, once given a taste of time, would never rest until eternity lay within its grasp. Thus we pursue those relations between sequence and structure that allow us to escape time's surface, venturing into that ocean that separates eternity from the instant in which we exist. Mathematics and music are two of the vehicles that assist us in our escape. Mathematics is available to a few; music is available to all. Mathematics allows us to assemble mental structures by which we comprehend entire sequences of logical implication at once. Music allows us to assemble temporal sequences into mental scaffolding that transcends the thinness of time in which we live. Through music, we are able to share four-dimensional structures we are otherwise only able to observe in slices, one moment at a time.

When computers began to multiply in the 1950s, artificial intelligence was believed to be just around the corner. Artificial music was surely almost as close at hand. Forty years later, artificial intelligence is still ahead. Electronics has enlarged our repertoire of instruments but has failed to produce anything more than a sympathetic resonance with the musical nature of our minds. Artificial music, so far, has the same relation to real music that animation has to life. Computers have perfect pitch, perfect timing, and perfect memory—achievements only the best musicians attain—yet something about music leaves them cold. The gap between the natural language of human beings and the higher-level languages and formalizations used by machines is slowly being bridged. But our music remains a foreign language to our machines.

How did music first evolve, and how might it develop further or evolve anew in different form? Neurophysiologist William Calvin has suggested that music is a by-product of the need to store complex sequences of motor instructions in serial buffers in the brain. “Movement
command buffers, essential for planning ballistic movements (so fast that sensory feedback arrives too late to effect corrections), were surely under selection for throwing,” Calvin explained. “For organisms that need to be both large (meters of conduction distance) and fast, one often needs the neural equivalent of an old-fashioned roll for a player-piano. We carefully plan during ‘get-set' to order to act without feedback. If those buffers are capable of sequencing other things when not needed for throwing-hammering-clubbing-kicking, then one might expect augmentation of such sequential abilities as stringing words together into sentences, or concepts into scenarios.”
¹⁶
Calvin suggested that parallel command buffers—he used the analogy of trains being assembled on the parallel tracks of a freight yard—may have stored alternative sequences at the same time. Fitter or more attractive sequences would be selected to survive, reproduce, and recombine.

Nature loves dual, or multiple, functions. Once motor-control sequences were stored in the same buffers as sequences of sound—a plausible scenario given the layout of mammalian brains—a cascade of developments was under way. The slow, evolutionary arms race between physically distinct individuals was accelerated by evolutionary processes acting within individual brains. To hit a target at even a modest distance, a thrown weapon must be released at exactly the right moment; the required submillisecond timing can be achieved only through parallel arrays of neurons that collectively smooth out the temporal jitter characteristic of a few neurons operating on their own. Precise relations between timing and frequency—the raw material of music—would be facilitated by the huge, parallel buffers of the expanding human brain, which, as we know but have yet to explain, tripled in volume over a short span of evolutionary time.

Larger brains allowed abstract four-dimensional processes to take root and grow. Individuals better able to plein, compare, rehearse, remember, and execute complex sequences of movements were better able to survive, and so these abilities were reinforced. But neither mind nor music, as far as we know, can be sparked in one brain alone. On one level, music may have evolved as a way to exercise and communicate these abilities; on another level, music may have evolved because these mental structures, able to reproduce themselves across time and distance, developed a life of their own.

To me, Calvin's hypothesis strikes a chord. My grandfather George Dyson (1883–1964) was a professional musician who first secured fame and lasting fortune by his mastery of the art of throwing bombs. He wrote dozens of well-received works of music and three
successful books, but none of them ever matched the sales of his instructions for hurling a kilogram of iron and high explosive thirty meters through the air. When war broke out in August 1914, my grandfather, unlike Olaf Stapledon or Lewis Richardson, had no second thoughts—he immediately signed up. He was commissioned as a lieutenant in the Ninety-ninth Infantry Brigade of the Royal Fusiliers, stationed at Tidworth on Salisbury Plain, and given the job of training infantry in the use of grenades before they went to France.