Biomimicry
Page 30
Stuart Hameroff has written a book-length ode to microtubules called Ultimate Computing. He is bold in print, and his monograph is a fascinating romp. It plunges into difficult mathematics and then suddenly breaches into the stratosphere, making predictions that raise eyebrows in some scientific circles. A cytoskeletal array would be a fine medium for artificial intelligence, Hameroff contends. How quickly could it compute? Well, in the three-and-a-half-pound universe we call a brain, there are 1015 tubulin dimers, each operating at a speed of about 109 operations per second, for a total of 1024 operations per second. If you want more dimers than that, make a bigger vat! We could even, he says with a bravado that has earned him some brow bends from more cautious colleagues, send vats of this stuff into orbit around the Earth, where it can evolve artificial consciousness. Or, he says, because microtubules are biological molecules, they would be welcome in our bodies. We could capitalize on their motorlike MAPS, he says, and send them as programmed nanorobots to do specific tasks inside the cell.
Hameroff ends his book with a dare: “Microtubules and the cytoskeleton created their place in evolutionary history by being problem solvers, organelle movers, cellular organizers, and intelligence circuits. Where do they go from here?” In a Computer magazine article in 1992, Hameroff and four other authors wager a guess: “If computation occurs in microtubules and can be decoded and assessed, cytoskeletal arrays may provide ‘devices’ with substantial computing power. Perhaps such systems will someday reach cognitive capabilities comparable to and even superior to human abilities.” And then the authors seem to read our minds: “While the ideas of dynamic coding and technological intervention in the cytoskeleton may seem farfetched, are they any more radical than were the ideas of static genetic coding and intervention in DNA and RNA some years ago?”
I thought of Hameroff’s claim when I read an article called “On the Path to Computing with DNA” by David Gifford in the November 11, 1994, issue of the journal Science. Someone had to think of it sometime. If simple enzymes can compute through shape-fitting, as Conrad contends, and if Hameroff’s microtubules can assemble and disassemble to form computing arrays, then what about the most wondrous coding mechanism of all: the code of life that twines together like two circular staircases, pairing up base by base in a simple yet splendid feat of pattern recognition? It was only a matter of time before someone climbed the peak to the DNA computer.
TRAVELING SALESMAN,
CONSULT YOUR DNA
DNA is a code, a kind of language, and you can translate what you want to say into the four-letter alphabet of nucleotide bases: A (adenine), T (thymine), G (guanine), and C (cytosine). By turning your information into a chain of molecules, you’ve managed to turn it into something that can be touched, something that’s controlled by the physics of shape-fitting and sequence matching.
You’ve also turned it into something that can be duplicated, in part because of a neat rule about complementary DNA. Here’s how complementarity works: When two strands of DNA get together, their bases line up very specifically. An A sticks to T, a C to G, and so on. Since combining is the energetically favorable thing for complementary strands to do, they will always zip together into the double-spiral helix of Crick and Watson fame. You can heat them to split them apart, but let the solution cool back to body temperature and they’ll rejoin without skipping a beat. Kevin Ulmer of Genex Corporation in Rockville, Maryland (now of seQ, Ltd.), says that’s like taking a Chevy apart, sticking the parts in a large crate, shaking it up, and having it reassemble into a car you can drive away. Since DNA “processors” are a little smaller than Chevys, however, they can fraternize by the trillions in a thimbleful of water, making them ideal for parallel processing.
DNA’s propensity for automatic assembly gave Leonard M. Adleman, who holds the Henry Salvatori Chair in Computer Science at the University of Southern California School of Engineering, an idea. In 1994, with a few test tubes of synthesized DNA strands, he set out to solve one of the most difficult computing problems known. The “directed Hamiltonian path problem” (finding a sleek path through a network of points) is a benchmark for computer prowess because an efficient algorithm (means to a solution) has yet to be found. The problem is that of the traveling salesman who must fly to many cities, but who wants an itinerary that will take him through each city only once. When there are many cities, the possible itineraries become astronomical. A trillion-operations-per-second computer trying to find a Hamiltonian path through one hundred cities, for instance, would need 10135 seconds—vastly longer than the age of the universe!
Adleman used only seven cities, looking for a path that would begin in Atlanta, end in Detroit, and pass through each intervening city only once. He gave each city a DNA name, using the letters of the DNA alphabet, A, T, G, and C, and then set out to create strands of DNA that would complement these names. To create these strands, Adleman used an increasingly common piece of lab equipment called an oligio machine that strings bases together automatically. As you’ll see in the third column below, he replaced each A with a T, each T with an A, each C with a G, and each G with a C, according to the rules of complementarity.
CITY
DNA NAME
SYNTHESIZED COMPLEMENTARY DNA NAME
Atlanta
atgcga
tacgct
Baltimore
cgatcc
gctagg
Chicago
gcttag
cgaatc
Detroit
gtccgg
caggcc
(Adleman actually used seven names with twenty letters each, but we’ll keep it simple.)
Using common recombinant DNA technology, Adleman made 30 trillion copies of these complementary DNA strands and set them aside.
Adleman then gave each segment of the route a flight name—taking the last three letters of the departure city and attaching it to the first three letters of the arrival city. If he were using English, the Atlanta to Chicago flight name would be the six capital letters in the following example: atlaNTACHIcago. But since Adleman was using DNA code, the flight names looked like this:
FLIGHT
DNA NAMES
DNA FLIGHT NAMES
Atlanta-Chicago
atgcga-gcttag
cgagct
Chicago-Detroit
gcttag-gtccgg
taggtc
Chicago-Baltimore
gcttag-cgatcc
tagcga
Baltimore-Detroit
cgatcc-gtccgg
tccgtc
Using the oligio, he manufactured the DNA flight names in actual bases, then made 30 trillion copies of each. The idea was that if stirred into the same test tube, these flight names would stick to the ending of one city name and the beginning of another, thus splinting the two names together. To test this, Adleman poured the flight names into the test tube of complementary DNA city names. (So far, lab technicians assure me, it’s as easy as Hamburger Helper.) Sure enough, the flight strands acted as splints; cgagct floated over to Atlanta and Chicago, for instance, and stuck to them like so:
It didn’t take long before the test tubes were full of long strands of DNA flight names splinted together. Through a series of recombinings and screenings, Adleman was eventually able to filter out all the strings that started or ended with the wrong city, or were too long or short. He was left with only strings of DNA molecules that represented the winning itinerary.
The problem was solved through self-assembly, the kind that occurs in Conrad’s shape-based computing. As David Gifford commented in Science: “The ‘oracle’ in Adleman’s method is the immense computational capacity of a ligation reaction that produces billions of products and by brute force tries all possible solutions.”
In Adleman’s first experiment (detailed in the Science article), only seven cities were chosen, but it seems clear that almost any Hamiltonian path problem could be solved this way. But it’s not just travelers’ queries that stand to be answered. Complex
problems such as telephone network switching, automating factory tasks, and artificial intelligence require what is called simultaneous processing. While conventional computers can explore only one or two solutions at a time, trillions of DNA molecules, each acting as a processor, can generate billions of possible solutions simultaneously.
In his press release, Adleman cheers cautiously: “It is premature to judge the long-term implications of this approach to computation; however, molecular computation has certain intriguing properties that warrant further investigation. For example, while current supercomputers can execute about a trillion operations per second, molecular computers conceivably could execute more than a thousand trillion operations per second.” In fact, it has been estimated that a DNA computer could perform more operations in a few days than all the calculations ever made by all the computers ever built.
He goes on to write: “Further, molecular computers might be as much as a billion times more energy efficient than current electronic computers. Also, storing information in DNA requires about 1 trillionth the space required by existing storage media such as video tape…. For certain intrinsically complex problems…where existing electronic computers are very inefficient and where massively parallel searches can be organized to take advantage of the operations that molecular biology currently provides, it is conceivable that molecular computation might compete with electronic computation in the near term.”
A few months after the paper was published in Science, Adleman held an impromptu conference on DNA-based computing in Princeton. To his amazement, two hundred scientists packed into a standing-room-only hall. Many talks were given and plans hatched, and though Adleman contends the field is still in “an embryonic stage,” others at the conference believed we might see some practical DNA computers in as little as five years.
Chances are, silicon computers won’t be abandoned completely—like Conrad’s tactilizing processors, DNA enthusiasts see vats of DNA as souped-up peripherals for silicon computers. They’d make a tremendous storage medium, for instance. One speaker said that a liquid DNA computer one cubic meter in size could memorize more information than all the existing computers in the world. Getting down to specifics, Eric Baum of the NEC Research Institute at Princeton estimated that one thousand liters of DNA solution could contain in coded form 1020 (that is 1 followed by 20 zeros) “words” of information. Another speaker estimated that a million times more information could be stored at the bottom of a test tube of DNA than in the entire human brain.
Members of the press gasped audibly at these forecasts, given the fact that Adleman’s experiment actually worked. It prompted Steven Levy of Newsweek to write, “Such an event is the equivalent of peering out the window of a bullet train and watching in astonishment as an unfamiliar vehicle zips by you with a fearsome whoosh. As if you were standing still.”
But for Michael Conrad or Stuart Hameroff or Ann Tate, Adleman’s announcement was an inevitable event, the shape of things to come. As Adleman has since remarked, his experiments have made him realize that “being a computer is something that we externally impose on an object.” He suggests there may be a lot of other “computers,” like DNA, that we have yet to discover.
Indeed, we are just beginning to investigate all the ways nature has already found to compute and transfer information. What may be most surprising is that it has taken us this long to look over nature’s shoulder for computing ideas. Perhaps it’s because our “search image” has been wrong; we haven’t “seen” nature’s computing devices because they don’t look like ours.
Not yet, anyway.
TO UNFLATTEN BIOLOGY: THE REAL QUEST
When I ask Michael Conrad what desktop computers will look like in the era of molecular computing, he hedges. For him, the real carrot is not the device. “The last thing the world needs is another new device,” he says. “As an aesthetic thing I can understand technology, but except for some medical technologies, I don’t really see technology as a human need. Our perceived need for technology is mostly generated by the competition of countries for export. I think it’s economies, not people, that need devices in order to grow.” This man, the head of a major computer center, doesn’t drive a car, nor does he need to. He walks to work from the Victorian apartment he and his wife, Debby, have lived in for fifteen years. If he misses a phone call while he’s walking, he doesn’t know about it; he’s beeper-free.
Conrad’s agenda, and the prime directive in his vision for the future, astoundingly enough, is to offer people a new paradigm by which to understand biology—a biological rather than a mechanical paradigm. “Right now, this Mac Plus is the ultimate machine—it’s what we know. That doesn’t mean we should use it to explain the brain.”
He’s right. We have a habit of making theories about organisms and basing them on the machine of the hour. We used to say that the human body worked like a clock, but that was when the clock was the ultimate machine. There was also a time when we said it worked just like levers and pulleys and hydraulics. Then we said it was like a steam engine, with a distribution of energies. After the Second World War, when we began to devise feedback controls for our factories, we said our body worked like a self-regulating governor or servomechanism. Now, predictably, we’re convinced that the body works like a computer. We’re using theories from computer science—theories that come from the machine world—to explain how the brain works, and that disturbs Conrad.
“We are teaching biology students that our enzymes and neurons are simple switches, turning on or off. In reality, we’re nothing like a computer, nor are we like a clock, a lever, a servomechanism, or a steam engine. We’re much more subtle and complex than that.
“This view of the organism as a digital computer has flattened biology, and I’d like to unflatten it. When I build the tactilizing processor, I hope it will make people stop and consider that there is more than one way to compute. Nature’s computers don’t work the way ours do. To think that they do is very bad for society—it makes us use digital computers for tasks we ought to be asking our brains to do—tasks to which digital computers are not suited.”
I thought a lot about what Conrad is trying to accomplish, and I think it’s much more important than beating other countries to the sixth-generation computer. Conrad’s insistence on unflattening biology reflects biomimicry’s ultimate goal—to learn more respect for nature and to recapture our sense of wonder. At its best, biomimicry should take us aback, make us more humble, and put us in the learner’s chair, seeking to discover and emulate instead of invent.
In their respective books The Death of Nature and The Reenchantment of the World, Carolyn Merchant and Morris Bergman agree that only by changing our perception of nature will we change how we behave toward her. There’s a history that proves them right. In the 1700s we ignored cultural taboos about violating nature and gave scientists permission to break the natural world into pieces to study it. With the animus and mystery gone, nature was suddenly on our leash, to do with her as we pleased.
Two centuries later, having taken reductionism about as far as we can go, there are signs that a rebound is beginning. Many scientists, especially those in the ecological sciences, have become students of the whole once again. Attitudes toward nature have also come full circle, reanimating life and restoring reverence to our relationship with the natural world.
In concert with all this, the biomimics are showing us that nature is the ultimate inventor, and that there is much that we as observers do not know—perhaps cannot know. By forming alliances with her, by using biology-friendly materials and letting evolution work its magic (even without knowing how it works), we’re bound to come out ahead of where we would be with our own linear, digital, rigidly controlled logic.
Will we be able to replicate exactly what happens in our brains by using carbon-based devices like the tactilizing processor, the microtubule array, a cube of BR, or a thimbleful of DNA? Michael Conrad laughs. “Remember, I have no illusions. I come fro
m an origin of life lab and I know how fantastic life is. To emulate nature, our first challenge is to describe her in her terms. The day the metaphors start flowing the right way, I think the machine-based models will begin to lose their grip. Natural processes and designs will finally be the standard to which we aspire. On that day, I’ll feel like I’ve done my job.”
CHAPTER 7
HOW WILL WE CONDUCT BUSINESS?
CLOSING THE LOOPS IN COMMERCE: RUNNING A BUSINESS LIKE A REDWOOD FOREST
When we objectively view the recent past—and two hundred years is recent even in terms of human evolution and certainly in terms of biological evolution—one fact becomes clear: The Industrial Revolution as we now know it is not sustainable. We cannot keep using materials and resources the way we do now. But how are we to land softly?
—BRADEN R. ALLENBY, research vice president,
Technology and Environment, AT&T
Nature has evolved systems over billions of years that work in harmony with each other, that build from bare, rocky, thin soil to lush, green forests. Without human intervention the processes of nature have evolved self-regulating forces of beauty, grace, and efficiency. Our challenge is to learn how to honor them and be inspired by their truth to create new cultural values and systems.
—JAMES A. SWAN and ROBERTA SWAN, authors of
Bound to the Earth
Stewart Brand, editor of the first Whole Earth Catalog, calls himself a “lifelong purveyor of the biological metaphor.” As a collector of tools and tips for the back-to-the-land-cum-sustainable-living movements, he realized long ago that the best tools are those that nature has already invented. So naturally, when Brand heard business consultant Hardin B. Tibbs talk about remaking industry in nature’s image (at the 1992 EcoTech Conference in Monterey, California), he wanted to be part of it. He stepped up after Tibbs’s talk and offered him a job with the Global Business Network, a consulting company working toward a sustainable economy.