Ask any Web site manager or Internet systems manager what their biggest problems are and they'll certainly say crappy, crashy, incomplete software and difficulty hiring skilled developers. The skills shortage is a debatable issue I've covered elsewhere but poorly written software, rushed to market long before ready for prime time, is a well-known Web development nightmare. It looks like we may have to learn to live with it.
September 3, 1999
I doubt I need to explain the facts. If you have done much Internet development at all you have run into buggy, incomplete software that just doesn't work as advertised from HTML editors that crash and produce nightmare code to middleware that core dumps regularly under load. Developers are finding they spent an inordinate amount of time figuring out how to work around the flaws in the software they need to use to do the cool new things that a competitive Web site demands.
So what's the problem? Is it just impossible to write decent software anymore? Has the software development world reached the limit of complexity human software coders can handle? Not really. There are solid programs like Apache out there and that tells us complex software that is solid and actually works well can be built by human beings. But the Internet development tools market is extremely competitive. Internet years (measured like dog years at 1=7) are so short that market pressures to get new products to market fast mean testing, quality control and careful tuning are given short shrift. Documentation? Documentation is often worried about a few days before shipping and often lags well behind features actually in shipping products. There a several reasons software seems to be getting worse and worse.
Products developed on one platform, NT for example, may be difficult to port to other operating systems. Software developers tend to be fierce proponents of their favorite platform and often have little idea of the problems and strengths of the others. Often programs written specifically for one platform are given to someone not familiar with the program to get working on another platform. Developer's keen on NT may not be too worried about getting their cool new software to work perfectly with Solaris. They may not know how to get it to work well with other operating systems and there may well be no one in the company who does know how. How many companies can you think of that are equally competent in both NT and Solaris development? It's a rare quality but developers want software with all the cool new features they've been sold on and they want it to work on their favorite systems. New software has to work on several different platforms and that can be a bitch.
As Web tools builders get bought, sold, aggregate and merge, products built by completely different teams on completely different platforms are often forcibly merged and combined into a witch's brew of software bugs, work-arounds and incompatibilities. I know of one ad management system developed on NT merged with a visitor-profiling program built on UNIX. The product has now been marketed vigorously for over a year, has sold well propelling the company to a nifty IPO but the software barely works at all. It's also extremely expensive. I've been leading a development team trying to get the product working since January. We've done all the usual stuff to get the project going. We've thrown hardware at it, consultants, and had the software company's "experts" in for months at a time. We now have the product barely working although new, weird problems occur constantly. We'll never be able to run it without constant attention from an army of outside experts and our own team. The company has been very cagey about giving us contact information for other customers using their products successfully. Which is worrying but I'm not terribly surprised.
We've worked closely with these people for almost a year now. It's been a rough relationship at times as you might imagine. Frank, beer-soaked conversations with their sales and technical people confirm my suspicions: most of their customers are having serious problems with the software and they have had major customers drop them recently. Legal action looms from a few soured sales. Employee morale is low. "We need some help over here and no one knows what the hell they're doing!"
To be fair I believe the vast majority individuals working for software developer companies I have been in contact with are well-intentioned and good-hearted people who really believe in their software and want it to work right. Maybe I can blame it on the marketing people who set unrealistic launch dates. They're just responding to market pressures themselves so it would be unfair to blame the situation on them.
Maybe I should place the blame on site developers themselves who too readily and meekly put up with incomplete and shoddy software products. Why did I spend almost a year working with the product I mentioned above when it didn't work? Why didn't I just bin it after a month or so and buy a competing product or roll my own? Good question. The product in question has "features" that would give us a competitive advantage. If we give up on it and one of our competitors manages to get it working we're screwed. SO we put up with it. Say no more.
