Nearly 50 years ago, Gordon Moore suggested that the number of transistors that could be placed on a silicon chip would continue to double at regular intervals for the foreseeable future. Known as Moore’s law, the truth of that observation has made computers cheap and ubiquitous. Cellphones are so inexpensive there are now more than six billion of them—almost one for every person on the planet.
Moore’s Law has also made mass automated surveillance dirt cheap. Government surveillance that used to cost millions of dollars can now be carried out for a fraction of that.
We have yet to fully grasp the implications of cheap surveillance. The only thing that is certain is that we will be seeing a great deal more surveillance—of ordinary citizens, potential terrorists, and heads of state—and that it will have major consequences.
In the past, surveillance was labor intensive. Twice as much surveillance required twice as many people and cost twice as much. But when surveillance became automated, its cost declined exponentially.
To understand the economics of surveillance, it is worth looking more closely at Moore’s Law.
In 1965 Gordon Moore observed that the number of transistors on a single chip had doubled every year since the invention of the integrated circuit in 1958. Since that time, his Law has been modified. The increase in transistor count has slowed to around 40 percent per year. A number of similar predictions have been made about exponential rates of increase in network capacity, pixels, and magnetic storage. Many of those predictions have proven true.
These technologies are the building blocks for surveillance systems. If you combine a number of technologies that are improving at the rate of 40 percent a year in a system, you can end up with systems whose performance is increasing even faster. Consider computer systems.
Computers combine integrated circuit technology, semiconductor storage, magnetic storage, and network performance into a single system. As a result in the 1990s, while the transistor counts were increasing at a 40 percent rate, system processing power was growing at an 80 percent rate.
Something growing at the rate of 80 percent a year increases by a factor of more than 300 in ten years. If the capability of surveillance systems were to increase at this rate, in ten years a dollars’ worth of today’s surveillance could be bought for fractions of a penny. Applications that were not feasible at a dollar suddenly are practical. These types of advances made the NSA collection of metadata feasible.
And if the capability of surveillance systems continues to increase at this rate, technologies that, say, identify people’s faces when they enter a store or board a plane are suddenly practical.
To my mind, there are two broad classes of automated surveillance— participatory and involuntary, and the line that separates them is fuzzy. Participatory surveillance arrived with the widespread use of the Internet. During this period users were actively involved in exposing their information over the Internet when they provided personal information in the course of purchasing products, searching for information, or interacting on social networking sites.
People were voluntary participants in the surveillance process even if they did not fully understand its implications. When they granted companies the right to use their information, they got services of great value in return.
Consciously or not, users were monetizing their privacy. That is, they traded information about themselves and access in virtual space and got free services in exchange. Amazon captured customer information and in return provided better selection and service, like one click shopping.
Google, founded in 1998, provided valuable free search in return for serving up targeted ads to users. Facebook provided communities, timelines, and “walls” for people wanting to network. Facebook users—now numbering more than a billion—received these services free in return for allowing Facebook to use their information.