Technical sin is not without its own peculiarities. Of course, one attraction to sin is the delicious anticipation of being caught. But in technical sin, not only do you rarely get caught, but if you are caught, someone else pays the price.
For example, suppose you create a grotesque programming language (there are a few about). You don't pay the price. Thousands of programmers do. They end up coding its contorted syntax.
A major technical sin is taking inappropriate technology and, by Herculean effort, turning it into a winner. There are lots of such technologies around, including the x86, color CRTs, and hard disks.
To an observer, the x86 is a variable-length ISA, register-limited CISC—one ill-suited for high-performance RISC implementation. Yet pour enough money and resources into it and it's a winner. Its advanced decoding and mini-RISC data-flow stages changed design. The x86 became a RISC killer. Just ask the PowerPC folks, who were creamed by early Pentiums.
Color CRTs are another apparently ill-suited technology with lots of complexity: three-color guns, phosphor displays, analog circuits, and so on. Today, though, a 15-in. color CRT costs $90 at Fry's.
Or how about hard disks? Take a platter, spin it at 10,000 rpm, float a read/write head a few microns above it—and call it reliable? Wouldn't a pure solid-state implementation be better? Ah, but it works, and it continually gets cheaper and denser.
But there's a larger canvas for technical sin, like developing a key technology and ignoring the lucrative embedded market. Take the PCI bus as an example. Developed for PCs, Intel and the PCI SIG ignored embedded technology, fielding an edge-card bus. Yet today, the PCI bus is fading, except for its embedded forms like CompactPCI, which are growing.
Is technical sin alive today? Are we pouring money down inappropriate technology rat holes? Are we missing the embedded boat again? You be the judge.
Reprints
