What's the cure for bad software?

By Richard Morochove

First published October 2, 1997

Why is there so much bad software out there? There are too many programs that are difficult to install, don't operate how they're supposed to and arrive late, far later than promised.

My initial enthusiasm for each new box of software that arrives is tempered by my realization of the problems it may bring. I often install a new program on a test computer, so if it blows up I don't risk losing really important data.

After analyzing the problems I've had with recent program releases, I think software quality would improve dramatically if developers paid more attention to just three areas: documentation, testing and planning.

I still remember the days when I'd get a big thick three-ring software manual with every program. In the front was a tutorial guide that walked you through the important aspects of using the program. At the back was an indexed user reference that explained operations in more detail. Today, you're lucky if you get a skimpy paperback that explains more than the installation instructions.

Software developers can get by with poor documentation today because it doesn't hurt their bottom line. When most programs offered unlimited free software support on toll-free phone lines, poor documentation could cost a developer a bundle. Now, it just means users fume, waiting on hold as toll charges mount, listening to some company pitchman rave about yet another great new program.

Under pressure to release products soon, most software isn't properly tested before its release to the public. Granted, today's programs are huge and it's harder to test software with millions of line of code, so big it can be distributed only on CD.

Yet there's more that can be done. Last month I attended a technical conference of a software developer who makes automated testing tools available so third party developers of add-on applications could ensure they worked. To my surprise I found that many of these add-on developers didn't feel it was worthwhile to invest in this testing technology.

Too often the testing falls to the unfortunate user who decides to take a flyer on a new program. Is it any wonder that many savvy computer users avoid the x.0 release of a program, preferring to await the x.1 version that fixes the initial release?

In house testing can be a long and expensive process, so many companies have moved to extended beta testing. In theory, this beta software is given a rigorous workout by a select group of users outside the developer. In practice, many of these beta tests appear aimed at testing the market acceptance of the software rather than highlighting the program bugs.

A perfect example is the beta testing program for MS Internet Explorer 4.0, the new Web browser officially released earlier this week. It was available to anyone who wanted to download a beta test version, weeks earlier. Microsoft maintains the fiction that beta testing improved the final product, when it was really just another tool in its marketing arsenal to compete against Netscape's Communicator.

How many software developers have a long term plan for product development and stick to it? If they don't have a workable plan, is it any wonder that users are disappointed when the software finally arrives?

Some companies love product development plans. That's why they have so many of them. Corel Corp. seems to whip up a new development plan every couple of months. In Corel's case, its plans seem more like trial balloons, sent up to see which way the wind blows. Yes, it likes Java. No, it doesn't. Yes, it does, but only for the enterprise market. Is it any wonder Corel Java evangelist Chris Biber left the company last week?

Apple Computer is an outfit that makes plans but has difficulty completing them. Apple revised its plan for the Copland operating system every six months or so. New features were added and the release date was pushed back further into the future. Of course, Copland never did get released.

That's why I'm skeptical about Apple's finely laid out plan for the release of its next generation Rhapsody operating system. There have been wholesale changes throughout the company since the OS plan was announced last January, yet the company claims it's still on schedule for a release next summer. It won't happen.

When it comes to future development plans, Microsoft seems to have a better track record than almost anyone in actually delivering what it promises. Mind you, it's often late. Windows 98 is just the latest example of a delayed release. However, I have no doubt Win 98 will arrive, with most of the features Microsoft promises.

Furthermore, Microsoft isn't quick to give up on a product. Windows CE, designed for handheld computing devices, sold poorly considering the number of high-profile hardware vendors who developed boxes using the software. Yet Microsoft listened to user complaints and, earlier this week, launched Windows CE 2.0. CW