How to think about technology predictions: hype cycles and why forecasts miss

TUREFF• •0

Flying cars, paperless offices, fully self-driving cars by the end of the decade: the history of technology is full of confident predictions that did not arrive on time, alongside changes that arrived faster and more completely than almost anyone expected. Learning to read predictions critically is one of the most useful skills for anyone interested in the future.

The hype cycle

The research firm Gartner popularised a model called the hype cycle, which describes how attention to a new technology often moves through five phases:

Innovation trigger. A breakthrough or demonstration gets people excited.
Peak of inflated expectations. Media coverage and investment surge, and claims outrun reality.
Trough of disillusionment. Early products disappoint, projects fail and attention moves on.
Slope of enlightenment. Practical uses become clearer and the technology improves quietly.
Plateau of productivity. The technology becomes mainstream and unremarkable.

It is a useful lens, not a law. Some technologies never leave the trough, some skip phases, and studies have found that many do not follow the curve neatly. Its value is reminding you that today's excitement, or today's disappointment, is rarely the final word.

Amara's law

A related idea, often attributed to the American researcher Roy Amara, says that we tend to overestimate the effect of a technology in the short run and underestimate it in the long run. The internet is a classic example: the late-1990s boom predicted much that later happened, but on a slower timetable than investors assumed.

Why forecasts miss

Demonstrations are not products. A working prototype in ideal conditions is a long way from something reliable, affordable and safe for everyday use. The last few per cent of reliability often take the longest.
Bottlenecks are elsewhere. Regulation, infrastructure, supply chains, skills and public trust frequently matter more than the core technology.
Costs follow their own curves. Some technologies get dramatically cheaper as production scales; others stay stubbornly expensive.
People behave unexpectedly. Users adopt things in ways inventors never imagined, and ignore what was supposed to be the killer application.
Incentives shape claims. Companies seeking investment, researchers seeking funding and media seeking clicks all have reasons to sound confident.
Straight-line thinking. Extending a recent trend indefinitely ignores saturation, competition and diminishing returns.

What good forecasters do

Research on forecasting, notably the work of the psychologist Philip Tetlock on what he called superforecasters, suggests some habits that improve accuracy:

Think in probabilities, not certainties. Say how likely something is, not simply that it will happen.
Start with the base rate. How often have similar technologies hit similar targets in the past?
Break the question down into smaller parts that can each be estimated.
Update often as new evidence arrives, and be willing to change your mind.
Keep score. Write predictions down with dates attached and check them later.

Questions to ask about any prediction

When you read that a technology will transform something within a few years, ask:

Who is making the claim, and what do they gain if people believe it?
Is there a working product in real-world use, or only a demonstration?
What needs to happen besides the technology improving: laws, infrastructure, cost cuts, behaviour change?
What would have to go wrong for the prediction to fail?
Has this same prediction been made before, and what happened then?
Is a specific, checkable claim being made, or something vague enough never to be wrong?

Avoiding the opposite mistake

Scepticism can be overdone. Dismissing every new technology as hype means missing genuine shifts, and many transformative inventions were mocked at first. The aim is calibration: neither believing every headline nor assuming nothing will change. Which technology prediction do you think was most wrong, or most surprisingly right? Share your examples in the comments.