Underpinning the entire technological revolution – and hence a decent proportion of economic growth – over the past 50 years is Moore’s Law.
- For a long time – perhaps as far as the 2008 financial crisis – it also contributed to increasing productivity in the West, but the economic realities have changed since then and productivity has stalled.
There are downsides, too:
- rapid progress means that equipment that is only a few years old become obsolescent,
- and given the number of toxic materials used in technology manufacture, this presents pollution issues.
But on balance, most people would see Moore’s Law as a Very Good Thing.
There’s more Geddit than one way of defining the law, but the one I like best is that the computing power available per dollar doubles every 18 months to two years. Variations focus on whether the time period should be 18 months or two years, whether computing power should be substituted with transistor density on an integrated circuit, and whether the relationship to money should be included
Gordon Moore was the co-founder of Intel (and before that, of Fairchild Smionductor), and in 1965 he wrote a paper suggesting that integrated circuit density would double each year for the forthcoming decade.
- Ten years later he changed the rule to doubling every two years.
- Later still Intel executive David House said that chip performance would double every 18 months.
Moore’s law was widely adopted as a planning assumption within the semi-conductor industry, and there is speculation to this day about to what extent this made it a self-fulfilling prophecy.
To fully appreciate the progress made in computers, we need merely contrast it to the more sedate progress in other areas – cars, homes and appliances, aircraft, transportation and energy.
- Cars are certainly significantly better – safer, faster, more economical and reliable – than the ones we had in the 1960s, but not 7,500 times better, like the computer on my desk.
A time-traveller from that period would recognise much of the non-IT world as similar to his own.
- This is in stark contrast to the changes in the last decades of the nineteenth cebtury and the first half of the twentieth.
When my grandmother got married in 1918, When my grandfather returned unscathed from the First World War she lived in a stone shack with no services.
- They had candles for lighting, a well at the end of the street, and communal cess-pit toilets.
There were no cars or even buses, and she walked four miles to work in the morning, and four miles home at night.
- There was no TV or radio, and entertainment was a sing-song in the parlour.
- Holidays were spent at a man-made lake a mile up the hill, or in Blackpool if you were lucky.
When she died in 1985, we had electricty, gas, a bathroom, a car, a phone, colour TV and a video-recorder.
- And you could fly to a foreign countries and lie on a sunny beach for a week.
Moore’s law is obviously not a law, but a mere observation, but it has held remarkably true for half a century.
- That doesn’t mean that it will be true for the next half-century, though.
By 2015, Intel itself was saying that the doubling was now up to two and a half years, and was likely to increase further.
Science magazine estimated that the peak growth in global computing power was in 1998, when the annual increase was 88%.
- Since then the rate of change is thought to have slowed to around 60% annually (still pretty impressive).
We’re running into physical limitations for making chips smaller – the chips are now approaching the size of atoms.
- The amount of heat produced by running them faster means that’s not a promising direction either – manufacturers now make “multi-core” chips that run cooler but need more complicated (and less efficient) software to operate them.
It’s also the case that each new generation of chips is more expensive to initially engineer, as R&D, testing and manufacturing costs rise.
- So we probably need an unforseen breakthrough to get back on track.
Here’s hoping that we do.
Until next time.
Footnotes [ + ]
|2.||⇧||Variations focus on whether the time period should be 18 months or two years, whether computing power should be substituted with transistor density on an integrated circuit, and whether the relationship to money should be included|
|3.||⇧||When my grandfather returned unscathed from the First World War|