From Tim (not Bill) O'Reilly: It's not exponential, it's sigmoidal
He talks about what people often see as exponential growth (just going up faster and faster) is often actually sigmoidal growth, in other words an 'S' curve, where things level off after a certain point.
This happens all over the place, especially in technology. Some examples:
Airplanes didn't even exist in 1900, but within 50 years, people were flying electronically-controlled supersonic jets. While airplane technology is still moving forward, it's a matter of refinements rather than complete changes - the 37-year-old Boeing 747 is still the standard for large passenger airliners, for example, and the 50-year-old MIG-21 jet is still flying in many air forces.
While modern cars have satellite navigation systems, more powerful engines, and cup holders, they're really not very different from those driven in the 1950s.
Moore's Law (which is really more "Moore's Industry Planning Suggestion") seems to be slowing down, at least with commercial CPUs - and we're only really seeing gains by using multiple cores in a system. My 2 year old G4 PowerBook is 1.67 GHz, while the latest Intel replacements have two 2.5 GHz cores - and yes the new machines are faster, but not that much faster. I remember the mid-90s when speeds really were doubling quite frequently, and one needed to get a new machine every year or two .
Digital Cameras are currently going through an exponential phase: what was a $10,000 pro camera last year is now a $3000 prosumer camera, and next year will be a $1000 hobby camera, and in a few years the same features will be found in a $400 'snapshot' camera. This is a tempting time to buy a digital camera, but it may be worth waiting a while until the technology advances slow down.
So next time you see something that looks like it's experiencing huge growth, keep in mind that it's likely to level off at some point. Also, don't get too excited by the prospect of 'exponential growth' - there are limits to everything.