Trends: Pattern errors & extrapolation errors

It strikes me that there are two types of errors in trend forecasting.

The first are errors in imagining that a trend exists. These pattern errors mistake a smattering of signals for an identifiable trend. Pattern errors join the dots to draw a shape that isn’t there.

The second type of trend forecasting errors arise from misinterpreting the shape of the growth. These extrapolation errors mistake the shape of the growth of usage, adoption or development.

Perspectives on trends most commonly get lumped into two categories: linear growth or exponential growth.

Early in the lifecycle linear growth is likely. If a trend has picked up any kind of steam, linear extrapolation greatly underestimates its growth.

If a trend has reached that tipping point, exponential extrapolation makes sense. But looking at a new technology trend and immediately inferring exponential growth over-estimates the trends potential.

These two patterns are reconciled in the s-curve. The real difficulty is identifying where on the s-curve the trend currently exists.