Free The Half-Life of Facts by Samuel Arbesman Page A

often exponential curves?
    .   .   .
    THERE are those who, when confronted with regularities such as Moore’s Law, feel that these are simply self-fulfilling propositions. Once Moore quantified the doubling rate of the number of components of integrated circuits, and predicted what would happen in the coming decade, it was simply a matter of working hard to make it come to pass. And once the prediction of 1975 came true, the industry had a continued stake in trying to reach the next milestone predicted by Moore’s Law, because if any company ever fell behind this curve, it would be out of business. Since it was presumed to be possible, these companies had to make it possible; otherwise, they were out of the game.
    This is similar to the well-known Hawthorne effect, when subjects behave differently if they know they are being studied. The effect was named after what happened in a factory called Hawthorne Works outside Chicago in the 1920s and 1930s. Scientists wished to measure the effects of environmental changes, such as lighting, on the productivity of the workers. They discovered that whatever they did to change the workers’ behaviors—whether they increased the lighting or altered any other aspect of their environment—resulted in increased productivity. However, as soon as the study was completed, the productivity dropped.
    The researchers concluded that the observations themselves were affecting productivity and not the experimental changes. The Hawthorne effect was defined as “an increase in worker productivity produced by the psychological stimulus of being singled out and made to feel important.” While it has been expanded to mean any change in response to being observed and studied, the focus here on productivity is important for us: If the members of an industry know that they’re being observed and measured, especially in relationship to a predicted metric, perhaps they have an added incentive to increase productivity and meet the metric’s expectations.
    But this doesn’t quite ring true, and in fact it isn’t even possible. These doublings have been occurring in many areas of technology well before Moore formulated his law. As noted earlier, this regularity just in the realm of computing power has held true as far back as the late nineteenth and early twentieth centuries, before Gordon Moore was even born. So while Moore gave a name to something that had been happening, the phenomenon he named didn’t actually create it.
    Why else might everything be adhering to these exponential curves and growing so rapidly? A likely answer is related to the idea of cumulative knowledge. Anything new—an idea, discovery, or technological breakthrough—must be built upon what is known already. This is generally how the world works. Scientific ideas build upon one another to allow for new scientific knowledge and technologies and are the basis for new breakthroughs. When it comes to technological and scientific growth, we can bootstrap what we have learned before toward the creation of new facts. We must gain a certain amount of knowledge in order to learn something new.
    Koh and Magee argue that we should imagine that the magnitude of technological growth is proportional to the amount of knowledge that has come before it. The more preexisting methods, ideas, or anything else that is essential for making a certain technology just a little bit better, the more potential for that technology to grow.
    What I have just stated can actually be described mathematically.An equation in which something grows by an amount proportional to its current size gets exactly what we hoped for: exponential growth. What this means is that if technology is essentially bootstrapping itself, much as science does, and its growth is based on how much has come before it, then we can easily get these doublings and exponential growth rates. Numerous researchers have proposed a whole variety of mathematical models to explain this, using