This year is the 40th anniversary of Moore’s Law. It was back in April 1965 the Gordon E. Moore, then the Director of R&D at Fairchild Semiconductor and later co-founder of Intel first published the observation that quickly became a Law and then an industry mantra.

You can lose a great deal of time picking the bones out of what Moore actually said in 1965 and in subsequent revisions, but the relavent bottom line for us today is that he forecast a doubling of performance every one to two years, with a simultaneous reduction in price of up to 50% over the same period.

Moore was specifically talking about the number of transistors that could be crammed onto a silicon microprocessor, but the principle extends far beyond the economics of silicon-land. In addition, it is almost scary to recall that when Moore was writing the world was a very different place to that in which we live now. It was a time when people were becoming aware of the life-changing power of technology – the atom bomb and even the contraceptive pill had pushed us along that path – but few knew where it would lead.

Indeed, Moore’s brilliance is that while many were responding to this exciting uncertainty by defaulting into the creation ‘science fiction’, Gordon Moore opted for a bold assertion of ‘science fact’. That he was damn near 100% accurate is astonishing.

Intel is rightly proud of its association with Moore. A visit to its website (www.intel.com) gives you access to lots of pieces of mem-Moore-abilia which there is no need to regurgitate here. Intel speaks of “breakthroughs that will enable us to drive Moore’s Law well into the future” and effectively claims to have paced its development – from 4004, 8086, 286 and 486 to Pentium and Itanium – on the back of it.

One Law for the rich?
So commonplace is the Moore’s Law mantra, go glibly is it bandied about by technologists seeking to reflect upon themselves some of its cachet by association, that it is surely time to ask again what it really means and how it works. And to what end.

The great thing about Moore’s Law, apart from its catchy simplicity, is that we can all see it being proved before our very eyes. Personal computers are the most obvious case in point (and perhaps that closest to Moore’s original thinking.

Mobile phones are not far behind. Their reduction in size has been fairly dramatic over the least 20 years, reaching the point where to become much smaller would render them impossible to use. What is more, the functionality in mobile phones continues to increase while the price remains steady or even falls.

Critics will interject here and point out that reduction in battery size is the biggest contributing factor to this trend. But that is nonsense for three reasons.

First, smaller and more powerful ICs have reduced the amount of a mobile telephone’s design ‘footprint’ occupied by them. Second, the reduced power consumption resulting from micro-integration of functions on Moore’s beloved semiconductors has helped the equation relating to battery power and life. Third, batter technology itself is subject to the tenets of Moore’s Law.

This may hold the key to the appeal – and the universality – of Moore’s Law in a broad range of technologies. Chips may lie at its heart, but they are dragging auxiliary technologies along with them.

Two examples suffice to show the undoubted strengths and the possible flaw contained within the strictures of Moore’s Law.

Exhibit A: transportable
My own favourite example of the pervasiveness of Moore’s Law pre-dates even his act of eponymous legislative activity. It’s a simple frequency meter, capable of measuring frequencies up to 30MHz. Built around valves and capacitors, it’s by no stretch of the imagination digital: it’s analogue with a well-developed sense of the logarithmic in its scale. It’s a remarkable piece of engineering. More than 60 years on, it still works.

It is also enormous. Even allowing for the fact that it was built for contemporary battlefield use, and thus heavily ruggedised, it occupies far more than a cubic metre in volume and is so heavy that it could not possibly be carried by two averagely fit adults for a distance of more than about 25 metres. Yet in the 1940s its performance and its portability represented  its ‘selling points’.

60 years on and it defies the basics of Moore’s Law. The functionality of my dear old wavemeter would occupy a microscopic area on a piece of fabricated silicon. It would weigh nothing. It would cost next to nothing. It would vastly outperfom its distant descendant, with functionality that was beyond the realms of imagination way back then.

At the risk of being repetitive, the genius of Gordon Moore was to think so far beyond what was conceivable at the time. That, and Moore’s conviction in his belief, and the manner in which the subsequent turn of events has proved him right, is staggering.

Exhibit B: Wintelism
The nexus between Intel’s microprocessors and Microsoft’s Windows software is often described as ‘Wintelism’. Moore can hardly be blamed for the latter but can hardly, as one of Intel’s co-founders, be absolved of responsibility for the former. Depending on your point of view, ‘Wintelism’ is either a term of abuse or a paeon to harmonious and virtuous circles of development.

Until you consider that ‘Wintelism’, seen through the prism of Moore’s Law, actually pits Intel and Microsoft against each other, head to head.

As mention earlier, Intel has pledged to deliver on the promise of Moore’s Law in terms of price and performance. Microsoft, meanwhile, seems to have undertaken to develop software that consumes more and more of that processing capacity, often to little effect and sometimes in a positively harmful way.

While it would be naive to think that Intel does not see the opportunity for fatter margins whenever it launches a new microprocessor, Microsoft seems to offer successive generations of products that do more, sometimes less well, often more expensively. Maybe that should be called Balmer’s Law.

Parkinson’s predicament
In fact there is no need for such new shibboleths. The fact that Windows software used to work just fine on machines built with just a fraction of the processing power and memory  available today, and seems to be more error-strewn with each new release of the basic software package, indicates that Balmer’s Law (or the strategy of Microsoft is nothing more nor less than Parkinson’s Law. This says that ‘work expands to fill the time [or in this space the processing power] available.

Intel aspires to Moore’s Law. Microsoft to Parkinson’s Law. I know which side of the ‘Wintelist’ equation I would rather live within.
Jim Chalmers