Friday, May 29, 2009

THE END IS NEAR! For Moore's Law and Our 25-Year-Old Cyber-Finance Economy

Our Cyber-Finance Economy -- Mathematically Doomed -- Moore's Law is Dead

From the Great - Father in the White House on down, commentators see "glimmers of hope," but the fact is the GFC (Global Financial Crisis) is just half finished. That's because just half of the economy we've had for over 25 years is finished, the finance part. Now, for the other part, the cyber or "tech" part, the end is near as well. The other shoe drops for the cyberfinance economy as the guiding light of the "cyber" (computer) part, Moore's Law, dies.

Notice the graph that shows Moore's Law. Basically, it's Intel Corporation's graph of the transistor count on its microprocessors (CPUs). In 2006 I predicted its end and its limit and added the red line.

Moore's law actually ended in the period 2006-2008 when the transistor count failed to double as Moore's Law requires. In fact, it failed to increase at all.

Let me tell you, when Moore's law catches a cold, the tech sector of the economy catches pneumonia. Even Gordon Moore, the originator of the law acknowledges this. It happened in 2000 when we had the Tech Wreck or the Dotcom Bust.

Another tech wreck couldn't have come at a worse time than right now. Computers and high-speed digital communications enabled the finance economy that arose in the 1980s and crashed in 2008. The tech industry enabled and was a leading beneficiary of the finance economy. Those campuses of glass towers that they built in the 1980s -- in places like Shaumburg, IL, and Independence, OH (south of Cleveland) -- got filled up with employees that might otherwise have been working in factories in Chicago and Cleveland, and got filled with computers based on Wintel products. Those campuses could soon look forlorn as ghost towns, like the miles of falling-down factories in those and other cities.


Everyone talks about it but few really understand it (and it is simple). Since everyone talks about it, it is important, right?

It is the prediction of Gordon Moore, the cofounder of Intel, going back in one form to the 1960s, that the maximum number of transistors on an integrated-circuit chip will double every 2 years because that will produce the lowest cost transistors. (In the beginning, Moore put the doubling period at 18 months.)

The 1980s office and finance boom coincided with the rise of the monopolies of Intel and Microsoft.

Here is how the world works: Software sells hardware.

Microsoft makes bloatware for a reason, not just incompetence. When Intel introduces a new processor, Microsoft comes out with a new Windows version that stretches its processing capability. In other words, what Intel giveth, Microsoft taketh away. So Intel introduces a higher performance processor and the beat goes on.


People thought that the limits of transistor miniaturization and cramming would be technical -- physical limits to the miniaturization of charge-based components (transistors) that could be built and reliably operated, or complexity limits to testing processors for reliability.

It turns out the limits are economic. Moore's Law always had an economic component: It was about reducing transistor size for lower cost as well as more performance, functions, and products. Furthermore, Moore's Law really wasn't a law in any standard sense of usage, it was a business plan of Intel and Microsoft working in collusion to gouge businesses and consumers.

Eventually it had to be done in by its own success. It has brought about commoditized computers, which Wintel began calling "netbooks" last year and which it then took over.

Unfortunately for all concerned, the commodified computers are so cheap it is hard for anyone to make much money on them. That is why Dell is reporting sales down by one fourth and profits down by two thirds. At that rate, Dell soon could be a money-losing proposition. It is no longer profitable for Intel and Microsoft to continue their bait-and-switch game (Intel baits with more power, Microsoft creates a demand for more power than that). It is no longer profitable for Intel to continue the Moore's Law game plan and hard to justify or generate the R&D and plant-investment dollars to do so. That's the answer to who killed Moore's Law.


Moore's Law follows the graph on the left in the other graphic, illustrating exponential growth. Moore once quipped that he was seeminly credited with inventing the exponential function, which the graph represents. What he really offered was a very astute business plan. Of course Moore always knew that the number of transistors on a chip had a limit and wasn't truly an exponential function. The limit is shown by the logistic growth curve (S-curve) on the right and its formula, which is not a pure exponential. This is the way things grow in nature -- children, bacteria colonies, etc. And it describes the life-cycles of businesses and business phenomena, including Moore's Law. Businesses use logistic regression in forecasting many kinds of growth.

The straight line that you see on the graphs Intel supplies and that I show above comes from using a scrunched up vertical scale (logarithmic) to represent the number of transistors.


Energy drives economic activity. When energy growth falls significantly below the historical trend of 2% a year, we have a recession or a depression.

A third graphic is adapted from Theordore Modis, a leading practitioner of logistic-growth-curve analysis. It shows the power of this kind of growth analysis. As coal was being replaced by oil and gas as the primary energy source, energy growth became chaotic as shown and we had the Great Depression. In the more recent period oil we see oil being replaced by gas and nuclear energy, and we are in another crisis.

Key growth trends in economics tend to coincide, as with the Moore's Law and energy curves shown.