Thursday, January 17, 2008

Anyone can make a faster chip

Editor's note: In addition to practical business suggestions, employees also submit inspirational quotes. Today's post is based on one I found quite inspirational indeed.

"Anyone can build a faster CPU. The trick is to build a fast system"

--Seymour Cray

I find this quote remarkable, and indeed most inspirational. To think that I, a lowly Powerpoint slide maker, can refocus my attention and make a faster chip is truly energizing. But why should I keep this pleasure all to myself? Instead of hogging the glory of making a faster chip, I think I will share this joy with my family. So tonight, my preschooler and I will commence work on a faster chip. It really shouldn't be hard--he's got lots of lego blocks and toy trains. And I'm sure I've got some wires and other stuff in the garage. We might have to buy a few switches and a soldering iron from the hardware store, but I'm pretty sure we can have something put together before too long.

OK, who am I kidding. Despite my advanced degree and years of experience in the industry, I could no more make a chip, let alone a faster one, than teleport myself to Dubai. In fact, I find Mr. Cray's choice of the word "anyone" to be quite interesting. While I'm sure there are exceptions here and there, I think "anyone" would be limited to electrical engineers, who make up 0.25% of the workforce in the United States.

Just for simplicity sake, let's say that Electrical Engineers constitute 0.25% of the workforce worldwide, knowing that in some countries it will be higher (Taiwan and S. Korea, for instance), and in some countries it will be lower (say, Haiti or Kenya). Given that only about half of the population is actually in the workforce, 0.25% of the roughly 3,000,000,000 workers in the world is 7,500,000 Electrical Engineers. Now according to the most infallible of all sources, Wikipedia, there are eight major subdivisions within the field of electrical engineering, so let's say, again for simplicity, that roughly 1/8 of our pool are working in microelectronics. Rounding up, that leaves us with an even million worldwide.

Of course, building a faster CPU means that you work for a firm that builds that kind of chip. There are probably 600-700 companies making semiconductors, but only about six or seven that actually make CPUs. Now these six or seven are among the largest, so let's assume that they have about 10% of the engineers. We're down to 100,000 engineers making chips. But 20% of these folks are managers, so we know they're not actually coming up with the ideas, so we're down to 80,000. Another 20% have changed jobs in the last year, so even if they're on the key projects, they haven't come up to speed to the point of actually making a contribution. So at best, when you say "anyone," what you really mean is the 60,000 who might be in a position to contribute to such a project out of the 6,000,000,000 or so people worldwide, or one out of every 100,000 of us.

To put that into perspective, if you live in the United States, you are five times more likely to be murdered than to make a faster chip. You are 460 times more likely to get cancer than to make a faster chip. You are 9 times more likely to bowl a 300 game, 20 times more likely to get a hole in one, 27 times more likely to injure yourself mowing the lawn, 800 times more likely to die this year, 455 times more likely to write a New York times bestseller, 6,700 times more likely to have diabetes, 177 times more likely to catch a ball at a major league baseball game, and even 1.1 times more likely to date a supermodel. That's quite an interesting definition of "anyone."

Indeed, some of the few things that you're less likely to do than build a faster chip are win an Olympic medal (6.6 times less likely), win the Powerball lottery (1,461 times less likely), or get canonized (200 times less likely), all of which I would take over building a faster chip for the simple reason that building it is one thing, but selling it for a profit is quite another.

Oh, and as for the part about "the trick is to build a faster system," who cares? Because even if you did build a faster system, Apple would build one that's smaller and more fashionable, and Google would build one that's free.