Online Entertainment Reaches 50 Percent Adoption
Online Entertainment Reaches 50 Percent Adoption
An Exercise in Forecasting Accuracy
TechCast tracks technologies though their normal life cycle, starting from Commercial Introduction (0 percent) to Take-Off (15 percent), Mainstream (30 percent), Mid-Point (50 percent) and Maturity (70 percent). Unlike prediction markets, which focus on a Yes/No choice to forecast a precise short-term event, the TechCast system forecasts continuous variables, such as the year a technology is likely to reach a specified adoption level. This more ambitious goal requires strong methods, but it also provides more valuable forecasts for planners and decision-makers,
The arrival of our forecast for Online Entertainment at the 50 percent level offers a timely opportunity to explore the challenge of producing accurate forecasts.
Online Entertainment Is Reaching the 50 Percent Adoption Level
TechCast began forecasting Online Entertainment about 20 years ago, and marked its arrival at the 30-percent adoption level in 2013. We then moved the target up to the 50-percent level, and now note that online music, video, and related industries dominate entertainment in global markets of roughly US$200 billion. Our forecast shows this is likely to grow to US$500â??US$700 billion at market saturation about 2020. The field is passing its 50-percent adoption level, so growth may start to slow. The US passed 51 percent in 2015.
Members can access the Entertainment forecast by clicking here.
Technical Forecasting Issues
While this forecast is good enough for most purposes, it raises technical questions. The TechCast experts call for reaching 50 percent adoption by 2018 +/- 2 years, but the S-curve suggests that arrival is likely in 2016. This S-curve fits well in a cluster of data, so it carries great weight. It is tempting to split the difference and call the arrival at 2017, but our experts' caution may be justified. Adoption cycles do not follow a smooth path but take a jagged course through the various stages of development. See our study of typical S-curves.
Experts Can be Wrong
Experts sometimes do not provide diligent estimates, as we have noted in earlier studies. (Halal, Forecasting the Technology Revolution, Technological Forecasting & Social Change, 80 (2013) 1635â??1643) We find that they often fail to read background material, make mistakes, and generally treat the survey too causally. Whereas Philip Tetlock's Good Judgment Project finds that "superforecasters" who are more accurate spend lots of time studying the issue, searching for information, and making careful judgments.
We counter these problems with a heavy dose of supervision and transparency to highlight the issues and encourage care. All forecast variables are displayed in frequency distributions to make variances clear and the S-curves chart the life cycle of a technology based on empirical data. (Note that these two common features of our work form the TechCast logo.) A pessimism/optimist index is calculated for each expert and displayed on their profile to note their proclivity to deviate from the norm. Comments supporting estimates are displayed for the two tails of the forecast distribution.
Most important, TechCast is one of the few foresight systems that conduct annual validation studies. Our format itself helps by providing the data in convenient bullet points so experts can scan meaningfully. They can also devote time to our extensive research or reading background material, hopefully providing a common starting point for making their assessments.
Power of Collective Intelligence
This illustrates the power of our method of collective intelligence. Most crowd-sourcing methods, like prediction markets, simply state an event to be forecast but provide no background data, leaving respondents to search for themselves. TechCast avoids this vacuum by scanning all available sources to provide the best possible background data: trends driving the technology and obstacles opposing it, adoption data, examples of new ventures, technology breakthroughs, other forecasts, and anything else that would be relevant. This too helps avoid the problems seen in many other surveys of experts.
Collective intelligence offers a superior method because it combines all these sources to estimate the "best possible forecast." Even the expert consensus may be considered in doubt if it disagrees with strong background data, as noted in this case. And because we conduct validity studies of our results every year, TechCast forecasts are authoritative and one of the best foresight systems available. Our validation studies show the error of our forecasts averages about +1/-3 years at ten years out. As I like to say, good enough to get decision-makers into the right ballpark.
All Public Comments
© 2013-2017 TechCast Global Inc Printed By: Dec 10, 2018 For personal use only