12 September 2008

Measuring Progress Toward the Technological Singularity

Another problem I'm having with the writings about the technological singularity is the concept that measuring the growth rate in world GDP is how you measure progress toward the singularity. (See Wikipedia's article on the technological singularity which references Hanson's IEEE article.)

GDP basically measures the amount of a human's labor (physical or intellectual) that is traded to other humans. We can see that GDP measures the amount of a human's labor (or time) by considering computers. Despite rapid improvements in functionality, the selling price of computers is roughly constant or even falling when we take inflation into account. This happens because the amount of human time it takes to design and build a computer is rising much more slowly than the amount of functionality available in the computer.

We can see that GDP measures trade by remembering that mothers who stay at home and take care of their own kids don't contribute to GDP, but if the mothers switch houses and pay each other to take care of each others kids, then GDP increases.

Historically, growth in GDP has been driven by two forces: an increase in the number of humans, and an increase in the amount of time a human spends producing goods for others instead of producing goods for self consumption.

Hopefully population growth rates are slowing and stablizing. And you can't trade more than 100% of a human's time. So GDP should stop growing at some point. No doubt a large portion of the world is trading a small portion of their time, and quite a bit of GDP growth can come from improving this. But in a hypothetical world where GDP doubles every quarter, this source of growth should be used up in a year or two.

Instead of GDP, progress toward the singularity should be measured via some measure of choice or available functionality. This is far harder to measure, but feels more useful. In the stone age, there were a limited number of materials one could use for producing goods: stone, wood, hides, twisted plant fiber, ... Gradually new materials were introduced: ceramics, bronze, copper, iron, steel. Plastic, kevlar, platinum, carbon fiber, asprin, yttrium-oxide, nano-tubes, ... The rate at which new materials are introduced increases, and measuring the rate at which the number of new materials increases is likely to be a good proxy for measuring progress toward the singularity.

This basically argues that you measure technological progress by looking at the number of patents granted. Which, at least in the U.S., doesn't seem to be increasing at an exponential rate.