- Shopping Bag ( 0 items )
Malthus on a Chip
* * *
"Cancerous growths demand food; but, as far as I know, they have never been cured by getting it." So declared Alan Gregg, a respected vice president of the Rockefeller Foundation, in 1955. The cancer he was referring to was humanity itself. Human population, Gregg argued, spreads over the surface of the Earth like a metastasizing tumor. All attempts to increase resources are ultimately useless.
Even in 1955, the thought was hardly new. Thomas Malthus supplied the original script, in a hastily written pamphlet published anonymously in 1798. Population increases geometrically, food supplies only arithmetically. Sooner or later, the growing gap between supply and demand must end in war, famine, and general misery. If there is one thing certain on this Middle Earth, it is that geometric growth cannot continue forever. The laws of mathematics forbid it. Or that, at least, has been—and remains—the most fundamental article of faith in Soft Green environmentalism. "Growth," declares the renowned Stanford biologist Paul Ehrlich, is "the creed of the cancer cell."
Hard Greens love Ehrlich. The first order of business in most Hard Green screeds is to quote Ehrlich's unfortunate 1968 bestseller, The Population Bomb. It is always the same quote. "The battle to feed all of humanity is over. In the 1970s and 1980s hundreds of millions of people will starve to death in spite of any crash programs embarked upon now." Ehrlich was flamboyantly wrong. The greenrevolution intervened. Since 1968, famine has in fact declined sharply around the globe. But it is perfectly fair to reply, as Ehrlich now does, that these issues are too important to be reduced to small matters of timing. The starving masses will be no less miserable if the mass famine comes a generation later than Ehrlich expected. Nor will the ruined planet be any less ruined by their desperate struggle to survive.
The Hard Green directs his scorn next at the Club of Rome and its 1972 study, The Limits to Growth. Humanity was scheduled—almost as precisely as a Swiss train—to run out of gold by 1981; mercury by 1985; tin by 1987; zinc by 1990; petroleum by 1992; and copper, lead, and natural gas by 1993. The Club was wrong, too: The average price of all metals and minerals in fact fell by more than 40 percent between 1970 and 1988. And oil is cheaper than ever before.
Still, as the Softs are forever reminding us, the final accounts aren't in yet. In the past century we have burned up fossil fuels that took millions of years to create. We shall burn as much again, and more, in the next. The rain forest is being leveled, the seas exhausted. Spendthrifts who inherit a great fortune are still spendthrifts, even if they discover halfway through a life of profligacy that they inherited even more than their doleful accountant had previously supposed.
Malthus on a Chip
In a modern economy, Malthusian accounting requires more than ordinary mathematics. Malthus himself made a fairly elementary mistake, after all. To grow more food, you can add more oil instead of more land. Oil, packed with (fossilized) solar energy, can be used to produce fertilizer, which makes crop yields soar. The modern, global, industrial economy allows tens of thousands of other trade-offs like that one. With so many different stocks in the portfolio, it takes more than a doleful accountant to keep track of the spendthrift. It takes a doleful computer.
Environmentalism's descent into silicon can be traced to a man who developed the first large digital computers in the 1950s to track Soviet bombers. MIT electrical engineer Jay Forrester had been investigating the use of magnetic materials to store digital information when he joined the team that was designing an early computer, called "Whirlwind," for the Navy. The initial objective was to build a machine that could analyze airplane stability and control. Forrester redirected the project to design a general purpose digital computer. As the machine evolved, the Air Force proposed a new mission for it. Forrester's computer would receive data from dozens of radar operators, store and instantly process it, and then vector defensive fighters to attack inbound Soviet bombers.
With the help of a brilliant graduate student, Ken Olsen, Forrester assembled the first truly functional digital machine. Tests conducted in May 1952 were stunningly successful. Forty-eight aircraft "attacking" Boston were accurately tracked and intercepted. IBM was brought in to transform the prototype into the nation's SAGE air defense system.
In 1955, Forrester moved to MIT's newly opened Sloan School of Management, in search of a new challenge. As he searched, IBM built increasingly powerful computers. Herbert Grosch, a one-time IBM employee, demonstrated that a computer's power would increase with the square of its size. "Grosch's Law" meant building fewer, bigger machines. The entire world, IBM estimated, would end up being served by about fifty-five big mainframes.
By the 1960s, however, both the uses and the markets for mainframes had already expanded. MIT itself owned one. At Sloan, Forrester set about using it to model industries and then cities. From a computer's perspective, the problems were not all that different from tracking bombers. Both come down to simultaneously solving large numbers of nonlinear differential equations efficiently and accurately. Forrester had mastered the altogether new skill of moving otherwise unsolvable equations into the new digital machines. In 1969, a year after Ehrlich's Population Bomb was published, Forrester published Urban Dynamics, a book about the functioning of large cities.
A year later a group of policy makers, academics, and managers calling themselves "The Club of Rome" convened in Bern, Switzerland, to discuss hunger, pollution, and other "world problems." Carroll Wilson of MIT, a member of the Club's executive committee, brought Forrester along. On his way home from the meeting, Forrester sketched out a rough model of planetary resources, which he called "World1," on the back of an envelope. He promptly transformed it into a large computer program. Release 2.0 (World2) would follow, then World3, then World Dynamics.
Forrester's models were Thomas Malthus brought back to life, torrents of gloomy electrons in the solid-state brain of a machine. The mainframe computer was critical; the model could never have been solved without it. Forrester's model consisted of some forty-five interconnected subsystems. Typical subsystem blocks were NRUR (natural-resource-usage rate), DR (death rate), POL (pollution), CID (capital investment discard), BR (birthrate), and so forth. Agricultural investment increased agricultural output, which increased birthrate but also pollution; pollution decreased agricultural output; and so on. Out of the model emerged predictions of things like total world population, total pollution, and quality of life. Quality of life, the model indicated, had peaked in 1940. With the computer script already written and filmed, the book was easy. Published in 1972, under the auspices of the Club and Forrester's acolytes Dennis and Donella Meadows, The Limits to Growth would be translated into twenty languages. It sold nine million copies.
The Limits transformed the fuzzy concept of scarcity into the holy digital writ of the Soft Green movement. The book purported to calculate what would ultimately "limit population and physical growth on this finite planet, and how the world's adjustment to its limits might be smooth and controlled rather than unexpected and violent." The predicament of mankind was not good, The Limits concluded. The world was rapidly reaching the end of its ecological tether. The model the book described predicted a series of imminent disasters: vast upswings in population punctuated with massive die-offs and steady decline in the quality of life. A "rapid, radical redressing of the present unbalanced and dangerously deteriorating world situation is the primary task facing humanity."
The book's timing could hardly have been better. The Arabs embargoed their oil a year later. The endless gas lines that ensued seemed to vindicate Forrester completely. Forrester's methods endured and grew. And thus it was that Malthus and the digital machine converged.
The Rise of the Model
The models have grown ever since, as fast as the computers on which they run. Year by year, they grow more detailed, complex, and graphic. Indeed, the computer modelers might well ask at this point a completely self-referential question: How much life can they pack onto half a square inch of sand? Experts have been pondering that question for over thirty years. The sand is silicon; the life, software; the habitat, a microprocessor. The size of a processor hasn't changed much since the first one was built in 1971. But the population within has grown geometrically, year by year, from tens of transistors in the 1960s to tens of millions today. Moore's Law, set out in 1978, predicts a doubling of the number of transistors on a chip every eighteen months. Gordon Moore, meet Thomas Malthus.
With the help of a sufficiently powerful computer, a numerical model can be built for anything: a bridge, a space shuttle, a city, a war, or a planet. Every modern designer of a car, radio circuit, or the control surfaces of a wing analyzes the system by building a mathematical model of the component pieces and then numerically analyzing how the system as a whole will behave. This is "system dynamics," a well-established, quantitative discipline for tracking flows of electric current, fluid, heat, and mechanical energy through a system, to determine how it will move, warm up, cool down, oscillate, resonate, amplify, stabilize, or die out. The basic rules are not very different from rules of accounting: Track energy and material as they move through the system, conforming with conservation laws that are familiar and well understood. The complexity lies in the number of connections, not in the basic principles.
Big computers are quite capable of keeping track of lots of connections. As Forrester's electronic progeny grew in the 1970s, it was reasonable to suppose that the machines might be adapted to track the stability or collapse of anything that is bought, sold, or driven; that freezes, boils, or melts; that vibrates, decays, or collapses. More complex systems just meant more equations, which simply required bigger computers. IBM would take care of that end of things.
So if we can model a car on a computer, why not the resources of a planet, too? The basic rules seem straightforward enough. Humanity resides on a finite surface, the planet itself. There is only so much dry land; there can be only so much oil beneath it. Supply and demand are matters of basic bookkeeping. Production and consumption must balance. Boeing designs new jets using similar models: Small quantities of basic science run through large engineering computers. If you just don't trust big computer models at all, never set foot in a jumbo jet.
Soft Greens, distrustful of high technology in all other arenas, have not hesitated to trust it here. Their computer models have become hugely influential in modern environmental discourse. Backed by a model, almost any grain of research, no matter how tiny the scientific backyard in which it takes root, can grow to full-planet size, with commensurately grand implications for public policy. It is the model that generates most of our environmental headlines today: species extinction, global warming, and almost every new cancer scare. It is the model that proves that however plentiful things may look just now, we will in due course run out: of rice, wheat, or food in general; of copper, zinc, and other raw materials; of new genetic stock for medicines and agriculture; of gas and oil; of land to live on; of dumps to contain our copious wastes. Feed a minimum of data into a maximum of computer code and then let the machine whirl off in time and space. The models let you think locally, pronounce globally.
For academics, pundits, and policy makers, the model has become as important as the "Earthrise" poster is for the rest of us. It lets us see it all, in cosmic perspective: the blue and white sphere of the Earth rising over the stark white horizon of the moon against the blackness of cosmic space. The models purport to reveal the ecological planet suspended ever so delicately, in both time and space.
Models and Markets
The Limits modelers were smart people, equipped with the finest computers of their day. These were the people and the machines we were counting on for our national defense. Forrester had a brilliant track record. His protégé Ken Olsen went on to found the Digital Equipment Corporation.
So it is surprising to find just how dismally The Limits models performed. Global economic collapse was supposed to have arrived a decade ago; what arrived instead was a global economic boom. They predicted the exhaustion of everything; we see instead glut upon glut. The people entrusted to guide our fighters to intercept Soviet bombers proved unable to guide your Buick to a fossilized tree. They predicted we would have run out of oil by now and our wheels would be up on bricks. You have in fact traded up to a monstrous Chevy Suburban, and you drive it more than ever. How did such clever people, such powerful computers, manage to get things so very wrong?
In retrospect, it is perhaps tempting to dismiss anything called a "world model" as the work of a crank. The label itself brings to mind one of those eccentrics who occasionally buys space in a newspaper, where he claims to set out in a single, very elaborate equation, the inner workings of everything, the secret of universal harmony, and the key to world peace. But Forrester's methods were not the methods of cranks. They were the methods of people who successfully design jets and coordinate our national defense.
It is equally inadequate to argue that the world is just too big to model. Modeling the flight of the world is, in some respects, easier—not harder—than modeling the flight of an airplane. Big things, and overall trends, are often easier to predict than finer details. It is easier to predict where the planet itself will be six months from now (that is, on the other side of the sun) than it is to predict just where a jet will be thirty seconds after an engine falls off. Next July will be warmer than last January; the weather next week is a tougher call. The fundamentals that Forrester set out to track—materials, energy, populations, and so forth—are all ultimately governed by the basic physical constraints of the planet, which should make the modeling easier. The Earth, its atmosphere, its oil reserves, its land mass, are all finite. Surely it is possible to crunch the numbers and arrive at some overall limits on how big a biological cake can be baked with these ingredients.
It is indeed possible; it just isn't useful. On the first cut, the cake turns out to be ridiculously huge. By global standards, life occupies only a very thin film on the surface of our planet. Moreover, atoms are indestructible: We have that on the authority of Democritus of Abdera, who so named them ("atomos," "indivisible") in 430 B.C. So one cannot "exhaust" the Earth's supply of copper; the worst one can do is disperse it. And while energy can be degraded, humanity uses only a minuscule fraction of the vast amounts of nuclear energy continuously released in the core of our own planet or directed toward us after its release in the core of the sun. Just looking at the standard things normally tracked by "system dynamic" models, there's enough matter and energy out there to sustain and replicate humanity billions of times over.
It is at this point in the standard Hard Green tract that one normally sets out some illustrative figures. Standing shoulder to shoulder, all the people of the Earth would scarcely fill Delaware. All our copper, tin, zinc, and deuterium could be culled from one-billionth of the world's oceans. All our energy requirements for two thousand years could be supplied by one-quadrillionth of the thermal energy produced by radioactive decay in the core of the Earth. Suitably stacked and compressed, all our trash would fit in a pyramid just 2 miles high in southern Connecticut. All our drinking water could be supplied by the rainfall of Oregon. Where did I get all these numbers? I made them up. I could have dug up real ones from the extant, Hard Green literature, and they would have read much like the ones I invented. But why bother? People aren't going to live shoulder to shoulder in Delaware, or build pyramids in southern Connecticut. Exxon knows where to find gasoline, not deuterium. Anaconda knows how to extract copper cheaply from a mine in Utah, but not from the sea. Forget about it.
So that is what Forrester did. He assumed the basic constraints defined by existing technology, markets, and patterns of supply and demand, as they existed in 1970. Not the speculative far future, not pie-in-the-sky, just the basic, sensible, practical present. He based his model on known technology, known mines, known reserves, known quantities of arable land, and known rates of growth in all of the same. From the get-go, Forrester had to shrink his model down to size: not down to geophysical size, but down to human size. To escape the big-Earth problem, Forrester shrank it, right down to the size of an academic's mind at MIT in 1970.
This is what practical-minded engineers always do. If you are designing a jet you assume more or less current technology, current engines, current materials, with cautiously optimistic allowance for improvement and innovation here and there. Oh yes, you make some allowance for technological innovation—whatever you can reasonably imagine coming down the pike of your discipline some time soon. You allow for modest variations in geometry. To model the flight characteristics of a caterpillar, you model the worm you see, not the butterfly concealed within.
But like the caterpillar, people change, even more than a caterpillar, in the one dimension that matters the most. To simplify the rest of the story only a bit, one might say that Forrester was not smart enough to model the future ingenuity of intellects like Ken Olsen, a protégé of Forrester himself. Olsen had completed his studies at MIT in 1952 and had gone on to found the Digital Equipment Corporation in 1957. By 1972, when The Limits was published, Digital was a $188 million a year company, with all but limitless prospects out ahead of it. Worse still for Limits modelers like Forrester, a new generation of wildly creative young Olsens, of equally unlimited talent and ingenuity, had enrolled at MIT.
Forrester counted mouths, but behind every human mouth there cogitates a brain. That was how Julian Simon would later see it. Simon, an anti-Malthusian academic at the University of Maryland, would respond to the Limits crowd with a series of books and articles declaring all neo-Malthusian predictions to be bunk. He dispensed with models, examined historical trends, and concluded that growing population makes people richer, because each new brain more than offsets each new mouth. Assume that, no more, and you turn all Malthus, all The Limits, upside down. Simon understood that mouth-brain trade-offs are a lot more subtle than that. But however simpleminded my own summary of his voluminous work may be, it is less simpleminded than the one contained in Forrester's labyrinthine computer models. Forrester made no real allowance for extra brains at all. He never imagined that pounds of sand might soon displace tons of copper. Corning Glass did, and gave us fiber optics.
Innovators have conspired to thwart Malthusians since the beginning. Food caught the interest of another cleric born in Malthus's lifetime, who spent his time cultivating peas in the garden of his monastery. Gregor Mendel thus learned the science of genetics. His intellectual descendants, the bioengineers of the green revolution, devised strains of high-yielding cereals that made nonsense of all Ehrlich's prophecies of famine. One might have supposed that the mentor of a man like Ken Olsen would make some allowance for genius in his models, but Forrester couldn't. An engineer can model a car, but he cannot model another engineer, least of all a smarter one. The collective genius of humankind cannot be contained in a model.
The closest we come to getting a grip on human ingenuity is in the marketplace, which is not a model at all; it is a process. The market is the place where innovation intersects with labor, resources, pollution, and quality of life: all the other things Forrester was trying to track. Most of the intelligence that traffics in the market falls well short of genius, but in the aggregate, it is ordinary intelligence that matters the most. Free markets elicit and distribute information and ingenuity in ways no other process can match. Forrester made no serious allowance for market forces at all.
He couldn't. If markets could be reliably modeled, as the Soviets thought they could, we wouldn't need markets at all. We wouldn't want them: Competition has many disadvantages, including many irreducible inefficiencies. We put up with markets only because, as bad as they are, they are the least bad economic option around. They discover their own future, far more efficiently than any model can predict it. Indeed, no model has ever been developed to predict tomorrow's stock prices reliably, let alone next year's. Lots of people peddle such models anyway, models as reliable as a dowser's rod or the crystal paperweight of a soothsayer. The most fundamental lesson of economics is that markets as a whole "know" things the rest of us don't. The "efficient market theory" establishes, quite conclusively, and with a wealth of empirical evidence to back it up, that the market always "knows" more than any player in it. Modeling markets is not difficult. It is impossible.
But recall that Forrester needed the market—at the outset—to deal with the big-Earth problem, to shrink the Earth down to size so as to discover some interesting limits in his models, some numbers smaller than "billions upon billions." He relied on the market to tell him what was possible in 1970, then relied on his computer to tell him what would be impossible in 1997. He assumed the market to construct his model, then assumed no market to run it, and ended up proving that the market would fail. He presumed, in other words, to predict the future of what he assumed did not exist.
The Limits of Models
Forrester taught us nothing useful about the limits to growth, but he did teach us something useful about the limits of models. To begin with, dead reckoning is impossibly hard. Forrester's Whirlwind directed the fighters to the bombers but only because it was richly fed with constantly updated information from the real world. Absent such feedback, mistakes pile up. In the real world you can generally see the drift as soon as it begins and constantly correct it. But a feedback-free model has to start right and stay right as it whirls off alone in space and time.
Forrester also taught us that the biggest computer can't make up for critical deficiencies in the model itself. Forrester left out markets and innovation, roughly like a biologist trying to model the history of life on Earth without allowing for mutation or survival of the fittest. Finally, and most fundamentally, Forrester proved once and for all that some essentials—like markets and innovation—can't be modeled at all. They are way too subtle and complex for that.
This is where even the most sensible observers often go wrong: They simply forget to take account of everything we don't yet know. In her 1992 book Costing the Earth, Frances Cairncross, the Economist's environmental editor, cogently argues that some forms of macro-economic "growth" are simply products of dishonest accounting. Rapid "growth" in the Philippines, achieved by the one-time harvesting of over 12 million acres of hardwood forests, resulted in denuded hillsides, silted rivers, and declining food production and fish catches. Macro-economic bookkeepers often neglect to keep books for "intangible" assets like environmental quality.
True enough, but those same bookkeepers invariably neglect to keep books for the intangible of intellectual property as well. Even if they remember patents, which are almost impossible to value in any event, they cannot add the sum total of what a populace has learned. But that is where an ever-growing fraction of national wealth resides. Which means that macro-economic books are of no use for all the things that really matter in the long term, environmental and intellectual alike. The most we can say is that for the last two centuries at least, intellectual gains have surpassed environmental losses on the missing ledger of intangibles.
Tellingly, the people who use computers to model everything else cannot begin to model the trajectory of the computer itself. Herbert Grosch had a model for the mainframe, and it was good for a while, but then it was knocked on its head by Moore and the silicon microprocessor. Today, IBM maintains that it is inefficient, even self-destructive, to keep packing more gates into a microprocessor. Eighty percent of a microprocessor's instructions are executed by 20 percent of the gates. A "reduced instruction set" chip is faster and more efficient than the alternative, the "complex instruction set" chip. RISC beats CISC, it is said. Control the population (of gates) and life (on the chip) will get better. Similar debates rage higher up in the cybernetic food chain, between purveyors of fully equipped desktop machines ("fat clients") and stripped down "network computers" ("thin clients"). And higher still, between fully loaded operating systems (Windows) and stripped down alternatives (Java).
Like the elephant on the Serengeti, the microprocessor shapes an entire habitat for others, the dung beetles and the Baobab trees alike. Life in the microcosm exists on a silicon plain of our own design and manufacture. We can dam and dike, fence and gate, build or raze, etch, burn, irradiate with X-rays, and lithograph with impunity. We can create new species and wipe out old ones. The EPA will not interfere. We are both the Creator and the Destroyer of this world, its only Darwinian avenger. But for all that, we have no real idea where it's all headed. The best minds, the wisest cyber-pundits, the wiliest NASDAQ stock pickers, the highest-flying venture capitalists, do not even begin to agree.
Modeling the potential resources of a planet is a lot harder than modeling the potential resources of a sliver of silicon. Much harder, for the simple reason that the resources of the planet are defined and extended by the power of silicon itself and the billions upon billions of infinitely smarter human neurons that function behind it. Smart as they were, the Limits modelers couldn't come close to modeling all the rest of human intelligence. No one ever will.
The Malthusian era can be said to have ended officially on October 11, 1990. Anti-Malthusian Julian Simon had kept on writing and writing. First, the neo-Malthusians dismissed him as a lunatic. Finally, to shut Simon up, Paul Ehrlich accepted Simon's 1980 offer to stake $10,000 on long-term futures "in any standard mineral or other extractive product you name." Ehrlich got to choose both the commodities and the 1990 day of reckoning. He ended up losing $576.07 to Simon in the Malthusian trading pits.
Free markets beat modelers every time. The inflation-adjusted price of raw materials has, with only a few, comparatively modest interruptions, been falling steadily for two centuries. For any good that is shadowed by paper on exchanges in New York or Chicago-all capital, all labor, all ordinary services like transportation or health care, everything traded or sold in normal markets—predictions of pathological disequilibrium invariably fail. People simply don't run out of things they can package as "property" and trade freely in unregulated markets. With markets in command, scarcity is always giving way to abundance. Forrester and the Limits modelers didn't get the story even half right. They assumed no market, from which they concluded the market would fail and we would all get poorer. But there is a market. It does not fail. We keep getting richer.
And yet, however wrong Forrester may have been, who can really doubt Malthus? Anything that starts growing geometrically will have to stop growing at that rate, one way or another, sooner or later. The growth can stop soft or stop hard, stop by catastrophe or by collapse; it can stop with a bang or a whimper, but it is going to stop. One can argue timing, but no more. The laws of mathematics, at least, are beyond dispute. Geometric growth ramps up toward infinity. But infinities don't happen in the real world.
So what then will impose the ultimate limit to growth? Will we choke on our own wastes? No. Will our high technology, our nukes, and genetically engineered strawberries crash down upon us? No. Can environmental salvation lie instead in making ever more efficient use of existing resources? No.
As I argue in Chapter 9, the real limits to growth lie elsewhere. It is reasonable to hope that they will save humanity from Malthus. And that they will save the rest of nature from humanity, too.