TechCast Project
The TechCast Project, conducted by Professor William E. Halal at George Washington University, defines itself as a “Virtual Think Tank Tracking the Technology Revolution.” TechCast has developed an improved, online version of the Delphi Method that uses empirical background data and the collective intelligence of 100 CEOs, scientists and engineers, academics, consultants, futurists, and other experts worldwide to forecast breakthroughs in all fields of science and technology. The U.S. National Academies of Science and Engineering rated it as possibly the best forecasting system in the world.
TechCast researchers scan the literature to gather background data and organize it into a 2-3 page analysis of each technology. These analyses include information on trends driving and opposing each technology, publicly available forecasts, and other relevant data. The experts are taken through these analyses online and instructed to enter their best estimate of when each is most likely to enter the mainstream, the potential size of the economic market, and their confidence in the forecast. The raw data is automatically aggregated for distribution over the site in real time.
More than snapshots in time, this is a continual tracking process that improves as technologies “arrive.” Comments from the experts and new data are used to update the analyses periodically. This method has been used for 15 years, and the average variance of all time forecasts is +/- three years. Recorded arrivals were found to occur roughly within this same error band of three years. An application of this approach conducted two parallel studies to forecast the arrival of energy technologies, one using a group of energy experts and the other using a group of general experts. The forecasts compared usually within one to two years.
It is often thought that methods like this are subjective, whereas quantitative methods are precise. However, quantitative methods also involve uncertainty because they require underlying assumptions that often are doubtful, and so they can vary widely. This approach subsumes quantitative forecasts into the background data and allows the judgement of experts to resolve the uncertainty that remains. Experts may have their own bias. If the present level of uncertainty is defined as 100 percent, this process is thought to reduce uncertainty to about 20 - 30 percent.