mcnamara@vixvax.mgi.com (03/22/90)
Trend extrapolation has been called a good test to determine what won't happen. One example that is in print is included in the book _Augustine's Laws_ (Norman Augustine, Penguin, 1987). It extrapolates the increasing cost of tactical aircraft versus growth in the defense budget and growth of the US GNP. The aircraft cost equals the total defense budeget in 2050, and it equals the GNP in 2120. Clearly something has to change. I also remember someone from Intel(Andy Grove?) doing the same thing with memory geometry. The result was that 20 or 30 years from now the line widths couldn't be shrunk anymore as they approached physical limits. A third example is the _Limits to Growth_ debate. The trends explored here were exponential, and indicated horrible things happening to the planet if these rates were continued. These cases have led some futurists and forecasters to state that trend extrapolation is a better indication of what won't happen than it is of what will happen. Yet clearly, corporate planning models are based on (usually) linear growth, and other life phenomena are modeled likewise. A personal example is inflation, which is assumed to increase the cost of goods linearly, and pay scales (which we *hope* matches the inflation rate). The implication is that when extrapolation indicates that a system boundary is reached, extrapolation will give incorrect results, and another paradigm must be used. Several years ago "S Shaped Curves" (initial exponential growth whose rate of change decreased as they approached a limit) were suggested. Does anyone have a reference on these? What other models or tools have been used in this area? What are some other examples? Curt McNamara "Our future world is our own growth in now, mcnamara@mgi.com not in the tomorrow of passing-time" Maurice Nicoll, in _Parabola_ Feb. 1990
rick@hanauma.stanford.edu (Richard Ottolini) (03/22/90)
Just go to the library and puruse "future" books written in the 60s and 70s. Most extrapolated trends and were wrong.