Why Predictions About Technology Fail

[A] The record of technological forecasting is bad in a particular way. Forecasts are rarely wrong about what a device will do; they are wrong about what people will do with it. [B] The telephone was expected to carry concerts into the home and did not; it carried gossip, which nobody had thought worth predicting. The prediction failed at the point where the machine met a use its inventors had no reason to imagine. [C] Forecasters know this and still cannot act on it, because the part of the problem that can be modelled is the part that does not matter. [D]

There is a second failure that is less often noticed, and it runs the other way. Technologies that are announced as revolutions tend to arrive late and then to matter more than anyone expected, because the change they bring is not the device but the rearrangement that follows it. Electric motors did little to factory output for thirty years. Output rose when somebody realised that a factory no longer had to be built around a single shaft, and redesigned the building. A forecast made at the moment of invention is therefore about the wrong object: it describes the machine, when what will decide the outcome is an organisation that does not exist yet.