Kurzweil @ Microsoft

Ray Kurzweil stopped by Microsoft to give a talk on "The Singularity is Near". Obviously good stuff. He spent too long pointing out progression on logarithmic graphs, though. I think the MSFT audience got the picture.

He mentioned early on in the talk that people don't often see exponential growth until a "knee point".

In the attached "linear-exponential1" graph, this occurs at about point 9 on the x-axis - when everyone wakes up and says "hey, this wasn't linear after all".

One thing he didn't mention (although he hinted at it when talking about boom-bust cycles) is how the nature of the exponential growth curve can breed skepticism in its very existence. Take a look at the second graph, "linear-exponential2" which is the same data, just looking at a shorter time scale.

Here's the scenario: a new technology emerges, and the proponents predict growth. They may be aggressive and predict exponential growth, but are more likely conservative and predict linear growth. They point out at t=6 or t=7, or maybe t=8 if they're feeling lucky. Everyone subconsciously expects linear growth, so they're picturing the blue curve. Instead, they see sub-linear performance, since exponential growth actually starts out more slowly than linear growth! (In visual terms, the pink curve is below the blue curve). So the technology is declared "dead' and "not returning on its promise".

You can see this occurring over and over again. The key take away - be skeptical of linear trends on both sides of the "knee point"!