SOFTWARE ... ALWAYS ALMOST READY FOR PRIME TIME


By G. A. "Andy" Marken
President
Marken Communications, Inc.
marken@cerfnet.com

Apple took it on the chin when they had a "minor" heat problem with their Powerbook system. Compaq bit the bullet when they had an input device problem. After weeks of denial, Intel admitted to 486 problems and finally offered to replace millions of chips in the field.
We won't compromise when it comes to hardware quality. However, install a new software program, from almost anyone, and you know you will be looking at weeks and months of work disruption.
Today's software firms are almost proud of the fact that their products never finished and are always new. Translation ... you're running your business on partially-complete products.
That's akin to an auto manufacturer selling cars based on a rough design and customer complaints. Each one that is shipped is improved based on the input and experiences of people who purchased the car last week. As your new car proceeds down the assembly line, they make "minor" enhancements like brakes, engine, electrical, etc., based on customer beta testing and feedback.
While we won't tolerate problems with other products we buy for the home or office, we've come to expect problems with each new software release. We don't even call them problems anymore. They aren't bugs. They're "undocumented features."
Many people relish the fact that they are the first in their group to purchase these untested, unproved and problem-riddled products.
For more than three years as Microsoft struggled to develop "Win '95, the next-generation operating system." People around the globe
clamored to be the first to use the evolving beta software. Microsoft
handed out millions of copies of the software. IS and network managers as well as technically sophisticated individuals provided Microsoft testing and trouble-shooting services at no cost. They used the software, documented the problems and forwarded them to Seattle. As soon as Win 95 was "relatively stable" Microsoft released it to the world.
When Win '95 was released, people stood in lines in front of stores so they could rush in and be one of the first to purchase the new operating system--despite the fact that the press had been documenting the problems for more than six months. Rather than find and fix all of the problems themselves, Microsoft sold millions of copies of the software.
They were so successful that the software industry followed. They realized they were wasting time, money and effort trying to deliver a finished package. They saw that people were actually willing to purchase something that wasn't quite ready to use. In fact, people actually relished the idea of getting software that was buggy, flaky and constantly hanging up or crashing.
Rather than pay for expensive test programmers and engineers, they realized it would be much more cost-effective to let users determine what needs to be fixed.
Few of us would actually put down hard cash for a car for business that is "pretty close to right." However, for applications that we use to run our companies on a daily basis ... it's a great decision.
At shows like CES, PMA, NAB and NetWorld+Interop, people stand in line and vie for the honor to take home beta software so they can become no-cost guinea pigs for software companies. While the volunteer test engineers usually get the products for nothing, it is becoming so "acceptable" that developers have even begun charging for the partially-complete product.
Putting a new spin on the problem software publishers now proclaim that their products are never finished and always new. Users are expected to have problems with each new product release. Publishers send out products knowing that over the next weeks and months, they will have to send out patches and "fixes."
The exception has become the rule.
If the trend is going to continue, it is a strong case for returning to the mainframe era or, as we now call it, thin clients or NC systems. Then, users won't have to purchase and install new software. Instead, it can reside on centralized corporate computer systems. Then, firms will be able to do a better job of coordinating their "upgrades and enhancements" and simply pay per usage.
It's a unique, revolutionary concept that we haven't seen for nearly 25 years.
A problem may arise. IS and network management may be slightly less willing to license the beta software. They may insist that software publishers actually finish a product before they begin working on the next version.
If people stop buying work-in-progress software packages, perhaps we'll be able to produce the workflow improvements we were promised with computers. We'll have the time needed to master the software packages we have before the next rev is released.
And when we receive the software, we won't have to be test engineers for software publishers and can focus on getting work completed.