Smilodon and the Computer Operating System


I think pretty nearly everyone goes through a stage in childhood where they’re fascinated with dinosaurs. I certainly did. Along with dinosaurs, I was fascinated by the saber-tooth cats, a fascination that has stuck with me.

One of the most interesting facts about the saber-tooths was that the big, robust, saber-toothed monsters actually evolved no fewer than four times, from Thylacosmilus (which was really a marsupial, a sort of saber tooth possum) to Smilodon, which actually was a kind of cat. They also died out four times, and the general agreement is that they died out each time because the big saber-like canines are an advantage right up until they get so big they are a disadvantage. The saber teeth get bigger and bigger until they become a dead end and the saber-tooth whatever dies out.

Computer programs are like that. Most successful ones start out small (Martin’s Law of Software Projects is that the first version of any successful computer program is running within 90 days of the project kickoff party) and grow. Over time they grow so large, have so many people involved, and become so complex they can no longer be maintained; they die off, or they reach a stage where all available resources are devoted simply to keeping the things running. These programs are then replaced by other small programs. With the exception of real game-changers, like the first spreadsheet programs, those small programs don’t do anything radically different from the big programs. They just do things better.

Last week’s big tech news was that Google is coming out with a “new” operating system, called Google Chrome. The scare-quotes are because it’s not really new at all in some senses, since it will be based on the Linux operating system kernel. What’s new is that Google is building a new collection of basic lightweight tools on top of this kernel, in order to have a consumer operating system suited for netbooks and other small computers, that doesn’t require a degree in computer science to manage, and that can be given away essentially for free.

Why bother? I don’t think Google is particularly interested in being the “Microsoft killer”. They’re not in that business, the business of selling computer programs, and since this is intended to be a free open-source product, they don’t apparently plan to be in the business of selling computer programs anytime soon. Google’s business is selling access to information on the Web. They give away access to the whole Web via the search engine, and then sell advertising on their search pages to pay for it. It’s a business model that’s hard for people to understand — it seems like Google is giving away their services.

It works, though, because the economics of the Web are built on services that cost almost nothing to provide. It’s like the Crazy Eddie business model: they lose money on almost every use of the system, and then make it up in volume.

For that model to work, though, Google has to be able to provide those services to many many people so that the very rare people who actually click through an advertisement add up to the revenue Google dearly loves. Google needs new users for the web.

Google also knows that a very large proportion of computer users don’t do anything very difficult with their computers; in fact, a lot of people only use the computers for surfing the web. Most people use Windows for that, but Windows has become pretty unsatisfactory.

Expose a new Windows box to the unfiltered Internet, and it will be infected with some kind of malicious software within literally minutes. Even if it isn’t infected, Windows has tried to become an “enterprise” operating system; it runs on your desktop, but it also runs business applications, big databases, and large-scale web sites, all functions that have essentially nothing to do with reaching email and reading Drudge Report.

The answer Google has come to is Chrome: a small, robust, simple system that can be sold on a small computer, and that does nothing except make it easy to get on the web. They aren’t targeting Microsoft specifically; they just want more eyes on web browsers, for more of the day. A simple, cheap, unbreakable operating system would make that more likely.

The consequences, for Microsoft and very likely for Intel, will be devastating. A very large proportion of the people who use computers now will find that they can get by happily using Google Chrome the free operating system, and Google Chrome the free browser, on a hundred-dollar computer, and do whatever they need to do besides surfing on Google Docs or any of the competing web-based services. To do the same things with a conventional desktop computer means bigger hardware, a relatively expensive copy of Windows, and a copy of Office — and there is always a market for letting people do something for $100 that used to cost them $1000.

Intel potentially has the same problem: for years, they’ve worked on making better, faster, stronger, versions of their processors, always trying to stick to a price point of a couple hundred bucks. They’ve been very successful, but with Moore’s Law, that processor that you can buy for a couple of hundred bucks is now a supercomputer in a box the size of a candy bar, and people don’t really need to do weather simulations at home.

As Windows has grown, the processors have grown, but people are going to find that Google Chrome runs fine on the processor in an iPhone or a hand-held computer game.

Historically, Intel has been a more agile company than Microsoft, and after all, they can make small simple chips using the same factories that now make big complicated chips. They might be able to adapt.

Soon Microsoft will find itself in the situation that eventually killed Sun: fewer and fewer people need the size of computer on which you make your money. They’ve gotten the biggest, shiniest, sharpest teeth on the planet.