TCS Daily

Twin Peaks

By Russell Seitz - October 9, 2003 12:00 AM

A nice selection of oracles were on offer at this years Emerging Technologies Symposium at MIT. Nathan Myhrvold of Intellectual Ventures told it "No group of people have been more consistently wrong in history than the people who said Moore's Law was wrong."


It's a case of so far, so true -- Moore has proved so unreasonably true for so long that it's hard to find takers for bets against Myhrvold.


Those that are willing point to the fine historical print. It reminds us that Moore was addressing one material and one fabrication technology -- photolithographic chip making on good old silicon. Decades before the run-up to silicon's reaching its physical limits, Silicon Valley saw the eximer laser handwriting on the wall and began diversifying into new technologies and advances in materials that will extend the validity of Moore's law well beyond the end of silicon's tether. Some were utterly material like wide band gap semiconductors and others reflected leaps of theoretical faith in isotope effects and quantum computing.


While the expansion of its science and technology base perennially rejuvenates the computer hardware of the near future, the same cannot be said of the software and models of the recent past. Complex models of global systems are only as good as the data that they operate upon, and as the data is refined by research over time, old models of the future can often diverge from newer and more elegant ones. They often benefit from higher resolution as well as more objective data. In 1990, modeling a thunderstorm was a tour de force of climate modeling, but now the models do hurricanes.


While Myhrvold's bold statement is presently true, it disregards the fate of yesterday's computer models. Many were hailed as the killer apps of the decade, century, or millennium because they pointed in one direction -- towards the end of something. You had your choice of Life as We Know It, Industrial Civilization, The Rainforest, Sunlight, Ozone, Affordable Canned Tuna Fish, Energy, or The World. Yet the champions of this assortment of fates of the Earth have mostly been forgotten or denied by their posterity. They once shared Myhrvold's touching faith in Moore's law, but unlike him were afflicted with a permanent incapacity to understand that there is another, higher one, almost on a par with the laws of thermodynamics.


It is one equally tied to entropy: Garbage in, Garbage out. All those attempting to predict resource depletion, or the accretion of pollution, can become GIGO's lawful prey if they fail to remember that the data they used the last time will be worse that the data they will use the next time. Because science is perpetually replacing estimates with measurements, and our understanding of complexity moves along in the only direction that time's arrow allows, models and the predictions they give rise to are doomed to perpetual obsolescence, so being an oracle isn't what it used to be.


The Pythoness of Delphi was a sacred person. Mocking her predictions, however wrong, could earn an impious, Athenian ostracism or a cup of hemlock. At its inception, global systems modeling likewise commanded oracular awe. In air-conditioned precincts, figures of Hierophantic dignity oversaw titanic machines that devoured reams of punch cards as intelligible as cuneiform tablets. Yet somehow the vast output of these fearsome engines was condensed into words as brief, ominous and cryptic as any the Pythoness uttered. When the Club of Rome debuted it reduced the accumulated wisdom of 140 years of science to just two words: Energy Crisis. Beholding long lines at gas stations, most of the known world declared the Club the soothiest sayers seen to date.


A decade later, using the accumulated wisdom of 150 years, plus a decade's worth of Moore's law in action, cooler heads in the National Academies of the world redid the analysis with owlish indifference to the Club of Rome. Reckoning up America and the world's gargantuan coal reserves, and the economic attractions of conservation, a bigger and better informed model rechristened the Energy Crisis the "Oil Glut." Nothing lasts forever, and Moore certainly never said that his law would. One or the other will have to peak first -- petroleum reserves or silicon performance, and the former are not doubling as fast as the latter.


The latest bad news for antique model fans comes from scientists in the Geology Department of the University of Uppsala. For all the demonizing of SUV's, ELF has been tossing its Molotov cocktails in the wrong direction -- 80% of the estimate of gas and petroleum reserves that served to inform the Kyoto Convention are nowhere to be found. Not even Asbestos Hummers commissioned to shrug off generations of ELF attack could snort forth CO2 enough to fulfill UNEP's prophecy of a global meltdown. The Club of Rome merely jumped the gun by a century. On any well-run Ship of Fools, it is the purser's duty to run the ship's pool. Care to wager on when hydrogen will run out, or when the rising tide of computational power will slack and ebb? Five dollars says bets will be placed on both.

TCS Daily Archives