Wired proposes a Moore’s Second Law: Overall net efficiency of any electronic system will double every 24 months.
I’m fine with that. There is a tech culture to be faster or smaller or cheaper – but not necessarily better. It’s easy to sustain existing technology. So it’s a fair criticism to say not enough attention is made to making technology more efficient – use less electricity, generate less heat, produce less waste. Today’s CPUs use more electricity than a heat lamp, and belch out so much heat that they need aluminum heatsinks the size of your fist. Consumers are reacting against loud, noisy fans in PCs.
The problem with Wired’s logic is that they fundamentally misunderstand what Moore’s Law is all about. It’s not actually a law; it’s merely an observation. In Gordon Moore’s “legendary paper”, “Cramming more components onto integrated circuits,” Moore makes several observations. Poignantly, it features a cartoon illustration of a vendor stand hawking “Handy Home Computers” the size of iPods. This is quite forward-thinking, considering this article was published on April 19th, 1965.
Moore simply points out that, all things being equal, the number of components per integrated circuit will double in magnitude for “at least ten years”. Manufacturing cost will decrease as well. Moore also notes that this progression can only be considered true with silicon semi-conductors with adequate cooling technology. Moore’s Law, goes out the window when considering the implications of biotech or quantum computing, for instance.
The paper continues on by discussing cost reduction. One way is to “amortize the engineering over several identical items”. Intel and other chip fabs already do this via speed binning. The second way is to introduce new design automation procedures; this has been realized with CAD.
Finally, the third suggested method is the building of “large systems out of smaller functions” and this is what Wired is suggesting should be Moore’s Second Law on efficiency. But I see it as a call for modularization, not efficiency. It is already happening now, with object oriented design and open standards.
What I do agree with is introducing a fourth axis to chip performance. The first three axes are speed, form factor and price. The fourth is overal system efficiency.