USS Clueless              Voyages of a restless mind

Main: normal long no graphics Contact Log archives Best log entries Other articles Site Search

Stardate 20010807.1939 (On Screen): Whenever an engineer designs something, one of the decisions he faces over and over is "Build or buy?" Sometimes that decision is made on the basis of technology, but usually it's driven by money. The reason why is that a "build" decision involves adding engineering time and expense, but a "buy" decision usually adds cost per unit, for royalties or for purchased assemblies. Generally speaking, the more of something you produce, the more likely you are to build. With smaller numbers, "buy" is usually more attractive.

Of course, that depends on the licensing scheme for the buy. They usually involve a royalty, but if you can buy for a flat rate then it may make sense even for a high volume product. But usually they don't. When I designed an internal test tool which was expected to have a lifetime build of 12 units, I bought as much of it as I possibly could in order to finish it in less than two months with a minimal staff. But it is actually worthwhile for General Motors to spend five million dollars in order to reduce the manufacturng cost of a car by fifty cents.

The Internet is the ultimate high-volume business. Reproduction of the product is essentially free and it's not always possible to figure out ahead of time how big the production run will be (i.e. the number of copies of your web page will be downloaded). A conservative design will err toward "build" over "buy" for that reason, because per-unit royalties can eat you alive. What's bad about that is that it can't be budgeted. Something which is predictably expensive is better than something which is probably cheap but might be grossly expensive. Uncertainty is evil. Which is why the licensing agreement being proposed for the CURL language is preposterous and condemns the language to failure.

It's possible for a proprietary format to become a web standard. Much as I despise FLASH, Macromedia has done it right. First, the player is free. Second, the tools for creating Flash animations cost a fixed amount. Third, and most important, there's no royalty. Once you buy the tool, you can make and post as many Flash animations as you want without giving Macromedia anymore money. As a result, it's become the tool-of-choice for smart web pages. Adobe's PDF is the same way: the viewer is free and the producer cost is onetime fixed, with no use royalty. If the CURL language were licensed that way, it could revolutionize the web. But that's not what they're asking for. Rather, they're going to let private individuals use it for free, but charge corporate users a royalty per download. What they're hoping is that private users will make it a standard which will force corporate users to adopt it. That's naive. Even if private users flock to it, corporate users will continue to build their sites using existing tools, so that they can control costs. As in most cases like this, the build/buy decision will be driven by money, not by technical merit. (discuss)