The Borland software cycle is well known to users of their software. At predictable intervals, Borland kicks out new versions of their software. Borland development teams struggle in cloaked secrecy to create a product. Working only with a tiny group of beta testers, the Delphi team usually gives birth in November or December. Whether the product is ready or not, whether it has enough new features or not, regardless of consequences, it will be shipped out to the world each year when the cold winter rains pound the shores of Monterey Bay.
Borland’s Delphi 8 product was buggy, unstable, and incomplete. Many people who follow Borland closely believe Delphi 8 was flawed because the team was forced to ship before they were ready. Delphi 8 had a huge new feature set, and it appears to many that the company did not give the team enough time before forcing them to ship. From a certain point of view, Delphi 8 seems to highlight the flaws in a market system that successful free software teams never face. But let’s not jump to conclusions too quickly.
Consider, for example, the JBuilder team. Twice each year, at sixth month intervals, they ship a version of JBuilder. Sometimes you need to be an expert to detect the difference between one version of JBuilder and the next, but nonetheless the versions keep coming, Over the long term, the JBuilder team, working at this frenetic pace, created a series of remarkably innovative products. From the first iteration of the all Java based JBuilder 4 platform, and on through JBuilder 2005, the team has continued to put out a series of solid products that break new ground. JBuilder had the first widely used all java editor. Innovations followed in refactoring, in code insight, in the editor’s ability to automatically insert try..catch blocks where needed, in its automatic handling of import statements. Again and again JBuilder led the way in programming editor enhancements. And wondrously, the editor worked nearly flawlessly on both Windows and Linux! No other IDE had ever done anything remotely as impressive.
JBuilder’s record proves that there is nothing inherently wrong with Borland’s approach to software development. You can successfully put a team on a break neck schedule, lock them up in near total privacy, and release products beneath a veil of secrecy so thick that at times it seemed like sacrilege to even consider marketing a product which had, during its development, been so carefully hidden from the prying eyes of the world.
Looking at these examples, one can conclude that the market forces driving Borland could successfully create high quality software, but could not guarantee the production of high quality software. But for free market capitalism to meet the claims of its adherents to be the best, and only practical, system, it must not only be successful, but be shown to actually outperform the competition.