© Jerry Stern 2000, All Rights Reserved.
As seen in ASPects, July 2000

Lately, my hardware clients have been having hard drive problems. They still have some older Pentium-based computers running simple record-keeping applications, and the hard drives are getting up to crash age, and they’re dying, early on a schedule. But that’s not the problem.

The problem is that simply replacing the drives doesn’t work. Standards for drive controllers have been changing about twice a year for some time now, far more quickly than motherboard designs can update the controllers. The hard drives available now are running Ultra-DMA 100 or 66, and the motherboards are expecting the old ATAPI mode 4 drives, or slower. The drives are allegedly backwards compatible so that they’ll work with older drive controllers, but they generally won’t work on any board more than two years old.

The fix? We have a choice of adding a modern controller on a PCI card, or rebuilding the PC with a motherboard upgrade. The real problems? Combining components with different useful lives is bad, and so is welding together products that were originally designed as modular units. Motherboards with drive controllers and a built-in kitchen sink aren’t anywhere near the worst of the trend, however. There are televisions, with no moving parts and about a fifteen-year life expectancy, grafted onto video cassette recorders with mechanical parts galore and a five-year expected life. When the VCR goes bad, you can’t throw it away; it’s part of the television, and it’s not worth repairing, because a low-end VCR costs less than the expense of a simple repair.

Software is worse. Around the end of the 16-bit days, the big shelfware companies started substituting bundling for innovation, and shipping size instead of speed. Try to buy a shelfware word processor these days–all three of the big office suites from IBM, Corel, and Microsoft include spreadsheet and slideshow/graphics programs, and each has a bundle with a speech recognition package. They’re all grafted together, too, thanks to the contribution to the mess by the magazine reviewers that rate programs by feature count rather than feature quality and relevance. Grafting or ‘integration’ makes it worse; the bugs can’t be isolated, and upgrading one product forces upgrades on all the others.

It’s not just word processing that is a mess. Internet browsers include E-mail software and FTP upload features. Anti-virus programs have become suites that include disk and file management. Scanners and digital cameras ship with half a dozen graphics applications, heavily overlapped in features, but with no functional reason for the combination.

Nearly all the big publishers have left behind speed and utility in favor of suite-ness and bulk. This would be the thump factor gone mad, except that you can’t have thump factor when the box contains a CD in a paper sleeve, a registration card, and a bizarrely-folded chunk of cardboard that keeps the box from being crushed when it reaches the bargain bin.

So, is this a good thing or a bad thing? It’s good for shareware publishers when the customers of shelfware realize that the graphics software that came with their word processor has taken over the extension associations used by the photo software that came with their scanner. Or, they find that their word processor is being used as a programming language by macro virus and E-mail worm authors.

Yes, it’s good for shareware authors, because our marketing and distribution methods force us to have reasonably-sized applications with easily-described purposes. Any application designer who markets by shareware can’t get away with bloatware; there is no way to describe a suite combo platter in a 45-character description without loss of focus, loss of downloads, and the loss of a business. Tightly-targeted applications that sell benefits over features, and speed over bloat, can compete with the big suites easily, just by being lean, easy to use, and above all, practical. You remember practical software that solved real-life problems–that’s what the shelfware mega-corporations used to publish, back when they were entrepreneurs and start-up companies.

Jerry Stern is the editor of ASPects, the author of Graphcat and FileTiger, runs Science Translations Software, and is online at www.pc410.com.