Working in as a Seattle IT Consultant, I am often asked to predict the future for my clients. I often wonder, am I really an expert in technology or an expert in the future. When we don’t predict the future correctly, our client suffers. When we do a good job of predicting the future our clients succeed. According to Moore’s law, the number of transistors is about 18 months. Each time a transistor is added to a chip, the power of the chip increases. Double the number of transistors and we double the processing power of the chip. The question though is, can Moore’s law continue?
I’m intrigued by the different viewpoints of the universe. Newton and Einstein created laws that predict the universe we can see. That includes the motion of planets as well as cars and airplanes inside our daily world. Yet there is another set of laws that also predicts the way our world works. Quantum physics predicts the way things function at the sub-atomic level. When things get really small, the rules that Einstein gave us, don’t work anymore. We can’t predict where an electron will be at a given moment. We can’t measure the sub atomic world the same way as the rest of the world. If we lived at the sub-atomic level our perceptions of the world would be completely wrong. Fortunately that world is so small that only theoretical physicists have to worry about such thoughts. Well at least when I was studying physics in High School that was true. For those studying physics in High School today there’s are real problem and it has to do with Moor’s Law.
Each 18 months that we double the number of transistors on a single chip. In order to do this, we can actually measure the number of molecules required to create the transistor. The trick is etching transistors using the wave length of very small like waves. With each doubling the light wave cycle needs to be smaller and smaller. As transistors begin to approach 5 molecules in size, the laws of Quantum physics begin to affect the transistors. It seems that in quantum mechanics we can’t guarantee where an electron will be for a given molecule. When we have millions of molecules, statistically we have current. This is because statistically a certain number of molecules will have an electron available to send through the circuit. When we get down to only a couple molecules, we can’t guarantee that there will be an electron in the valence of the molecule that can be used for creating current. It’s actually a mind blowing concept.
I think of it like this… if we ask a million people to come to a party. We can predict that some will have money in their pocket. Some will have keys in their pocket. Some will have lint in their pocket. Some will have a pencil in their pocket. In fact with a million people in the room you might find some strange and expected things in some people’s pockets. You just can’t predict it. So if you checked, there would probably be lots of change in 1 million people’s pockets. If you needed $1000 dollars in change, you could probably find that change amongst the million party visitors.
Yet you couldn’t guarantee that any one person had change in their pocket. In fact you couldn’t guarantee what was in any given person’s pocket. Yet as long as you have a million people at the party, a lot of people will have change. Yet if there is only 5 people at the party, you can’t guarantee anything. So too with molecules…
You can’t guarantee that at the instant you need an electron; there will be an electron in its valence. Because guarantees and accuracy are what computers are about, as transistors get smaller we can’t guarantee the accuracy of the transistor.
Which means that yes; Moore’s Law will eventually end. The predictions are that sometimes around 2020 or so, we’ll hit that end. How does this affect your planning for your clients? I am planning this into my thinking about my Seattle IT Services clients. Most of us are only thinking one or two years from now, but if we think about how our clients will be growing, we can’t depend on the rapid growth of Silicon chips to produce more and more computing power. This is still 8 to 15 years from now. Yet, as we see the world of IT changing so rapidly, we can also predict that this infinite 18 month growth cycle will affect anyone’s career that is dependent on this type of growth. Could that be you?