GEARED CYCLES

Both discoveries and innovations (thus defined) lie on learning curves during certain intervals; but they are bunched together, for they come in certain waves, separated by periods of relative stagnation.

During the industrial age, there have been three such complete waves, separated by lulls centered in 1802, 1857, 1921, and 1980. (We now seem poised for a take-off that should get under way in 1984 and culminate around 1993.)

The cycles (vaguely reminiscent of Kondratyev's economic cycles) have certain conspicuous similarities: the innovations appear in the same order as the discoveries that preceded them; the rate of innovations is faster than that of the discoveries, so that the time elapsing from discovery to innovation is considerably shortened during each cycle; and though the three observed cycles are very similar in shape, the time scale (like all history) accelerates.

But for us the most important aspect is the connection of each cycle with energy. Each stagnation valley is close to a primary energy peak, and the rise of the innovation wave foreshadows the rise of a new fuel, which penetrates the market at something like the rate of the discoveries that preceded the innovations. (It seems as if the innovations demanded a new energy source, a guess possibly confirmed by the fact that several sources, such as coal, had long been known before they began to penetrate the market.)

Price, the indicator of the level at which supply balances demand, is a symptom, not a cause, of economic development, and Marchetti's model does not use it; however, price does go through a sudden spike whenever a fuel peaks, so that one may expect a considerable decrease in energy prices, especially of oil, over the next few years (as Prof. Singer has been telling us for some time).

It is, of course, not difficult to find the evidence for a preconceived theory, and there are professors at Berkeley, Pittsburgh, and M.I.T. who make a living that way. It is more difficult to fit a curve to 80 years of history with no more than two constants at a time; in any case, Marchetti took the raw data from a German book on technological innovation and had no say in their selection.

If you believe in the guides of history more than in the printouts of the computer models (as this writer does), advise your son to study nuclear engineering, but warn him of the difficult times ahead around 2015, when natural gas will probably peak. Lovins and Komanoff have a point about the coming decline of nuclear power, but they are prophets well ahead of their time: such a decline should gradually begin in the early 22^nd century.