Hope and Horror

Summary: The electronics market is set to drastically expand. or fail. Take your pick.

President Obama is a commie socialist because of his stimulus packages.

Or he's a saint who didn't spend enough.

Put two economists in a room and you'll get three strong opinions. Keynes was right! Or not. Frankly, I've given up. These guys spend decades earning PhDs and their Nobels, yet their opinions are so divergent one has to sigh and acknowledge Carlyle's notion of economics as the dismal science. Their prognostications make weathermen look prescient.

Happily, though, there are organizations that specialize in predicting the future for narrow industries. It stands to reason that in these limited domains we can expect pretty good accuracy.

For example, IDC projects that the embedded space will double in the next four years (http://www.networkworld.com/news/2011/090911-idc-embedded-systems-market-to-250716.html). We embedded people are on a gravy train! Start spending those anticipated raises now, folks, as that kind of stimulus will save the economy. Many economists think so, anyway.

IDC says that this means 14.5 billion microprocessor cores will be shipped in 2015. Odd, that. My data suggests somewhere around 12 billion were sold last year. I'm not an economist, but am pretty sure 2 times 12 billion is not 14.5 billion.

(Hopefully there's a Nobel prize lurking in that startling observation).

If 14.5 billion represents a doubling, then either 7 billion were shipped in 2010, or IDC anticipates that multicore will fail as we designers race to cut the number of CPUs from our products.

The CBO (http://cbo.gov/ftpdocs/123xx/doc12316/EconomicTables.pdf) is projecting a 14% increase in the GDP by the end of 2015. But 6% of that is expected to be from inflation. Not that the CBO gets any credit for being Nostradamus, but if one works the numbers, real GDP growth will be $1 trillion, exactly the amount IDC says embedded systems will grow. Various sources (e.g., http://geography.about.com/od/obtainpopulationdata/a/uspopulation.htm) suggest the population will grow by over 10 million in the USA by 2015. We'll be buying more electronic gadgets, but eating less.

Gartner, another refuge for otherwise-unemployable economists, thinks (http://eetimes.com/electronics-news/4227364/Chip-sales-slowing-Gartner) that the semiconductor industry is poised to - wait for it - shrink by 0.1% this year. One priceless quote reads "This is a significant change from Gartner's previous projection, given in the second quarter, for 5.1 percent growth this year." That's an enormous revision in the course of only a few months.

Next year they project 4.6% growth in semi shipments, which is half their previous estimate.

Note the apparent precision embedded in the predictions, expressed to a tenth of a percent, even though they are wildly volatile even over so short a period of one quarter.

I would love to watch these folks at work. Do they use a dartboard labeled with growth figures? Tarot cards? Or maybe they're more like the editors of National Inquirer: a giggling gaggle of pranksters trying to come up with something truly absurd to spring on their credulous customers.

But what really amazes me - and, heck, I'm no economist so don't have the extensive academic credentials to understand difficult microeconomic problems of this nature - is that there's a market for this so-called "data." The standard deviations are so huge, over such short periods of time, that one is inexorably drawn to the conclusion that their reports are meaningless.

Published September 16, 2011