Moore’s Law is responsible for many of the defects in today’s quality of life. Whereas a typical U.S.1950s appliance – say, a Maytag washing machine — lasted for 30 or 40 years without major repairs, modern electronic gadgetry is lucky if it survives intact until its third birthday. With Moore’s Law slowing gently to a halt as we reach the limits of the physically possible, it’s time to impose the 1950s Maytag virtues on the tech sector.
Maytag was founded in 1893 and brought out its first fully automatic washing machine in 1949. It was at that time a family-controlled company, and remained so until the death of Frederick Maytag II in 1962, who had been President of the company since 1940. If you bought a Maytag made while he was in charge, perhaps a 1959 matched washer-dryer set with the new “Halo of Heat” technology that stopped the dryer overheating, you were likely to be using it into at least the 1980s – I had one in a house I rented from 1979-82. The “Loneliness of the Maytag repairman” advertisement series began in 1967, presumably relying on the superior durability of the company’s 1950s products, and became proverbial by the 1990s.
Of course, Maytag’s very superior product quality and durability were only possible because of operating in a truly free market. Washing machines were unregulated in the 1950s; the nonsense about “energy efficiency” came in only after the 1973 oil crisis.
Moore’s Law, propounded by Intel founder Gordon Moore in 1965, said that processor speed doubled every year (later slowed to every two years as microprocessors advanced.) This was a truly brilliant prediction, which held good in full until around 2010. Using it enabled designers and engineers to estimate accurately what processors would be capable of several years in advance, thus greatly aiding product design and marketing. It is only ceasing to be true now because of limitations imposed by the speed of light and the size of molecules. The last few iterations of it have been aided by complex design and material changes but even these are reaching a limit. Future technological advances in this sector will be possible, but primarily using massively parallel processing and increasingly carefully designed software.
Moore’s Law was such a brilliant prediction, valid for almost half a century and has been so useful in the IT business that it seems unfair to cavil at the difficulties it has caused. Nevertheless, those difficulties exist.
For one thing, since the Boskin Commission of 1996 Moore’s Law has allowed the government to distort consumer price index figures. Since that time, the index has taken account of “hedonic” benefits from new technology, measuring them in the tech sector by the Moore’s Law advances in processor speed, as though tech products doubled in capability every two years.
In reality, the computer on which I type this is at most twice as capable as the Internet-enabled Windows 95 IBM Aptiva model on which I worked in 1996 not, as Moore’s Law and the CPI calculations would have it, 1,000 times as capable. Its processor may work 1,000 times as fast, its memory may be 1,000 times as large, but the clunkiness and bloat of the software it uses mean it has nothing like the 1,000 times the functionality. Also, word processing is word processing.
Furthermore, the CPI’s inclusion of “hedonic” improvements in quality do not take account of “hedonic” deteriorations in quality, such as the failure of modern washing machines to last anything like the 30-year lifespan of a 1950s Maytag. Thus since 1996 price increases in the United States have been systematically understated and growth figures overstated, probably by close to 1% per annum. When this is adjusted for, the devastating effect of “funny money” monetary policies since 1995 in distorting resource allocation and destroying productivity growth becomes even more stark.
Moore’s Law has also had a thoroughly insidious effect on quality control and durability in the tech sector. If 2018’s product is going to be twice as good as 2016’s, there is little point in making the product last until 2020, and no point whatever in making it last until 2046. The norm has thus become products that fall apart after 2 years and are more or less completely impossible to repair (or at least, impossibly expensive to repair.)
This is not an idea that can entirely be blamed on Gordon Moore. While Henry Ford made essentially the same Model T for 19 years (though Model T’s generally did not last that long in road use) Alfred P. Sloan at General Motors pioneered the idea of the annual model change, introducing new features and new styling on General Motors models so that consumers would feel the need to stay abreast of the latest automotive gadgets and trends. The ultimate epitome of this approach was Harley Earl’s 1959 Cadillac, complete with tail fins designed to mimic the look of a spaceship, which was that generation’s ultimate fashion statement; the equivalent of the latest model iPhone. 1959 Cadillacs were not notably less durable than the Model T; they were quite well made, but they were certainly a great deal less durable than they could have been, or than contemporary Maytags.
After 1960, the annual model change ceased to work so well. Styling had become boring and hard to differentiate – it is difficult at first glance to distinguish a 1965 sedan from a 2015 sedan. The initial effect of this was an effort to cut corners and a sharp deterioration in product quality, so that mid 1970s “compact” cars were unspeakably prone to constant trips to the repair shop – needless to say, to the great profit of repair shops. (They were not however so bad as were their British equivalents, even at the top end; my father’s 1973 Jaguar was a disaster.)
However, the arrival of well-made Japanese and German competition sorted that out. Since 1980 or so Detroit’s product quality has been steadily improving; my 1998 Buick gave 15 years of service with only one major repair and I have strong hopes that my current 2010 Lincoln MKS will give me good service until 2025 or so. Detroit may seethe at the slowdown in the replacement cycle of U.S. automobile sales but consumers have immeasurably benefited – a benefit that is, unlike the Moore’s Law nonsense, not caught in official statistics, either of GDP or consumer prices. (The correction should however only modestly inflate current GDP estimates, but should sharply ding those of the 1970s.)
If the Moore’s Law benefits are drawing to a close, Silicon Valley needs to redesign its products. Once the ability to add new features is lessened, only a few consumers will be attracted by new design features like Harley Earl’s tail fins (a cellphone with tail fins would in any case be inconvenient – the fins would gouge pieces out of the user’s ear!) If 2022’s cellphone is to be only a modest improvement on 2017’s, and even 2027’s and 2032’s to have only moderately more functionality, then consumer demand will arise for Silicon Valley products that last at least 15-20 years, if not the 30-40 years of a 1950s Maytag.
Older consumers will finally catch up with the technology, but they will naturally demand that it be solidly built and durable, providing them with reliable value for a decade or so without the need for replacement, and for expensive and annoying re-installment of their favorite software and re-learning a new collection of pointless operating system changes. In emerging markets, too, the demand for a basic smartphone that is also rugged and durable will increase, as it becomes primarily a tool for enhancing the livelihoods of its users.
If Maytag was an independent company today, it could usefully go into the cellphone business, drawing on its long-standing if battered brand reputation for quality and durability. That would be in the fine tradition of the original Frederick Maytag, who diversified in 1907 from farm machinery into its first washing machine, a hand operated model with a wooden tub (the electric motor came four years later.) A Maytag smartphone, with durable steel case and an estimated 30 to 40-year life, would initially be a niche product selling at perhaps 30% above the price of an iPhone, but would quickly spread as it became clear to consumers that the steel case and greater durability was worth more than any new features.
Like its young engineers, Silicon Valley must eventually grow up and accept the constrictions of middle age, while the intellectual cutting edge moves on to other areas such as biotech. Silicon Valley’s engineers may believe they can achieve a renaissance in the spread of robotics, but they lack the capability to be cutting-edge in that area. Robots, after all, require gyro-mechanical abilities that are only controlled by software. Being largish pieces of equipment and requiring substantial investments, they will also require a much higher level of reliability and durability than Silicon Valley traditionally builds in. They are, in other words, much more like washing machines than cellphones.
(The Bear’s Lair is a weekly column that is intended to appear each Monday, an appropriately gloomy day of the week. Its rationale is that the proportion of “sell” recommendations put out by Wall Street houses remains far below that of “buy” recommendations. Accordingly, investors have an excess of positive information and very little negative information. The column thus takes the ursine view of life and the market, in the hope that it may be usefully different from what investors see elsewhere.)