The manifold problems of technology forecasting Yogi Berra and others are reported to have said that "prediction is very hard, especially about the future." That is true of politics, and it is also true of technology. While the general impression that technology forecasts are far too optimistic is true, it is not universally true. Yes, we don't live in underwater cities, nor do we commute in helicopters. However, there are many technologies that have surprised not only the general public, but sometimes even their inventors and promoters, with their success. Such has been the case with wireless telephony. Before that, computers were also underestimated (with Ken Olsen of DEC, one of the great figures in the development of minicomputers, famously expressing his doubts about the need for computers in homes). And before that, cars, and before them railroads, also surprised with their wide spread. In many, even most, cases, what people do with a technology differs widely from what inventors had in mind. The main difficulty in technology prediction is the human factor. What will people want to do with the wonderful gizmo you have just invented? Perhaps they will just turn up their noses and not do anything with it. Thus it should not be surprising that forecasting in this area is difficult. However, there are some general principles that do apply much of the time. One of them, quite well known, is that the penetration of new technologies tends to follow the logistic curve, the S-shaped curve that appears in many areas. Another general principle, not as quantitative, nor as well known, contradicts the general expectation that new technologies will replace old ones. Instead, new technologies frequently serve to strengthen their predecessors. (Thus, in popular language of the last few decades, they are "sustaining" and not "disruptive.") An excellent example is that of railroads and horses. The rail was expected to kill the horse. This was a common expectation on the part of both proponents and opponents of the new offspring of the emerging metal and steam technologies. Instead, the number of horses grew. In Britain, their numbers did not peak until 1905. The issue was that while transport on rails was fast and inexpensive, the "first mile" problem of getting to the rails required horses. With the general growth of the economy that was facilitated by railroads, there was more and more to carry to the rails. And even railroads used horses extensively, for switching cars in their yards, for example. Similar phenomena occur commonly in other settings, and often lead to serious mistakes in planning. How many times have you seen predictions and promises that better communications, such as faster Internet access, will stimulate telecommuting and decrease road congestion? Such predictions are almost certainly wrong. At least they have been consistently wrong for about two centuries. Many, often very knowledgeable, observers, thought that transportation and communication were substitutes for each other. But the uniform experience to this age has been that they are complements, and grow in parallel. Yes, you may work from home, but chances are that you will make more trips to meet clients, or for family and other reasons, and in the end will travel more than before (barring major upsets, such as an stronomical rise in price of fuels). So what can we conclude? Most of all, that the future is hard to predict, so we have to prepare for the unexpected. Second, though, we should keep in mind that among the most common unexpected phenomena is the resilience of old technologies, services, and business methods, and their propensity to adopt some of the innovations that we work on.