You can see the computer age everywhere but in the productivity statistics.Nobel Prize in Economics winner Robert Solow, in 1987 (quoted everywhere but it’s hard to find an exact source)
Paul A. David demonstrated that the Solow Paradox was nothing new by drawing parallels between the spread of computing and the spread of electrification at the turn of the 20th century.
New technologies just take some time – longer than we might expect – to impact (measured) productivity. But why?
One reason is that technology just takes longer than we think to really permeate societies. It may have felt like computers were ubiquitous in 1987, but in hindsight it was still the early days of the Computational Revolution.
The Clothesline Paradox
Another argument is that many of the benefits of technology aren’t captured in productivity statistics:
- It’s very difficult to measure the value of the new, previously impossible (or radically transformed), things that technology enables, especially when those things are free or nearly free (crowd-sourced data, online collaboration, social networking, machine learning… Even blogs.)
- Improved products cost roughly the same, but offer a combination of more functionality, greater reliability, and a different (better) quality of experience. This gain doesn’t show up clearly in the data.
- In some areas (e.g. data processing, video entertainment), prices fall fast enough that they obscure the gains made.
- Other things become effectively free at point of use. They have a big impact but go unmeasured (this is the clothesline paradox).
All of these are true. But a final (and apparently major) part of the “missing” productivity is missing because of a kind of switching cost: implementing a new technology takes investment in equipment, people and processes. At the beginning it’s actually harder to use the new technology, and the gains only come with time as skills and processes improve in parallel with improvements to the tech itself (think The Print Shop vs MS Publisher vs Quark vs InDesign vs Affinity).
More on this – including what I call an actual graph – tomorrow.