For hardware, this reasoning runs headfirst into the lintel on the low doorway of logic. Open-source hardware just doesn't make the same kind of sense.

Software design is like poetry; hardware design is like architecture. There are millions of amateur poets in the world because it's easy and costs nothing to try, but you don't see people throwing together buildings just to see if they stand up.

So it is with programming: It's easy for a clever engineer to hack out some code in her spare time. Programming is free, it doesn't hurt anyone and plenty of like-minded people are willing to help with test and debug. Open-source software agreements are like so many "open-microphone" poetry readings.

Hardware design, on the other hand, is riskier, more expensive and more structured. You can't dabble in hardware design because it costs too much; you can't e-mail prototypes to friends for tweaking and debugging; and you can't generally get EDA tools for free. Roll-your-own hardware can't be fully tested on an amateur's budget. Whether designing boards or ASICs, the costs and the complexity are just too high for engineers to do it for fun, or out of a sense of altruism.

Ah ha, I hear you saying. What about open-source hardware in the form of RTL (Verilog or VHDL) files? And what about programmable logic, such as FPGAs from Altera, Xilinx and others? Surely design files can be shared, and FPGAs make hardware prototypes less expensive.

They can, and do. But they don't solve the fundamental problem. Designing hardware is the easy part; verifying, testing and characterizing it separates hobbyist projects from bona fide products.

This doesn't mean hardware designs can't share some open-source benefits. Some newer user-configurable designs still benefit from incremental upgrades, plug-in enhancements and the collaborative spirit that makes Linux so popular.

Yet we have recent examples of 32-bit processors that thousands of people downloaded but nobody uses. The downloaders were largely college students and hobbyists, filled with curiosity but with little probability of ever producing silicon.

Who wouldn't jump at a free microprocessor and the chance to avoid licensing fees? Real engineers with real jobs and real projects, that's who. The cost of a license fee is irrelevant; the cost of missing a project deadline or of fumbling the entire development are astronomical. Companies can't afford to bet their products on "freeware" that has no visible means of support.

If your neighbor were selling trips to the moon on a homemade rocket he built in his garage, would you go? Would it matter if the trip were free? I didn't think so.

Jim Turley Is Senior Vice President Of Technology At Arc Cores Inc. (San Jose, Calif.), A Subsidiary Of Arc International Plc.