History of Apple Inc./Hamstrung by an Established User Base

A factor in the failure of any technology company that must be considered is technical incompetence. Apple displayed a measure of this, perhaps most of all in its 90's operating system development work.

See "hitting the wall": 

The initial incarnation of the Macintosh OS had been created for the minimal hardware platform that was available in 1983; tradeoffs had been made to meet time-to-market and cost constraints. By 1989, affordable hardware had become much more powerful, and Apple management was meeting to discuss redesign of the OS to adapt to the changing needs of the market. Some features, such as a user interface with color, were added fairly quickly, but the less superficial features necessary to make the Mac competitive beyond the early 90's proved more challenging. Indeed, by 1997, this next generation of Mac OS, codenamed "Copland", was incomplete with no foreseeable ship date. On the advice of engineering manager Ellen Hancock, Apple CEO Gil Amelio killed Copland. What had happened?

Copland seemed to incorporate all the right ideas. For instance, a marquee Copland feature was "memory protection". Memory protection prevents poorly written software from causing system crashes -- this hadn't been a significant issue in the earlier Macs, which only ran one program at a time, but beefed-up early 90's Macs ran many more programs concurrently and accordingly, crashed much more often. Memory protection technology had been present in UNIX for decades; cheaper and more powerful hardware in the early 1990s made it both technologically feasible and a competitive must-have for personal computers, and appeared in 1992 versions of Microsoft Windows. Copland also included a "microkernel" and operating system "services": both respected means of organizing a modern OS. Copland also included a virtual machine called the "blue box" which would run an older version of Mac OS for backwards compatibility and a simple upgrade path for existing customers. Apple's heart seems to have been in the right place.

But technically Copland proved difficult to execute. To illustrate one technical hurdle: Copland aimed to use no more than 4MB of memory, but the backward compatibility mode demanded a complete, memory-resident version of Mac OS 7, consuming 2.5MB of memory;  two-thirds of the Copland memory budget was consumed before the first line of code was written. Interestingly, the 4MB limit may have seemed reasonable in 1992 when Copland was initially scoped out, but, as Copland fell behind it became quaint: 64MB of memory was standard by 1997.

Insiders report that technical difficulties like this were only part of the problem. Apple by the mid-90s had assumed the posture of a research organization. The regular releases, firm timelines and deliverables lists characteristic of successful software companies ceased to exist. Mac OS upgrades had become little more than bundles of shareware apps -- a widely touted but hardly notable feature of Mac OS 8 was an interface tweak known as "springing folders". "Why Apple Failed" talks of "Snowball Projects": projects which, like snowballs rolling down hills in cartoons, had grown larger and larger until they had only a snowball's chance in hell of completion. These projects became unkillable, because in their growth they had also grown to contain strategic technologies the company believed it couldn't do without. Copland was not the only failed project of the era -- X, Y, and Z also bombed.

The Copland goals eventually materialized, years later, in Mac OS X.

See "The Rise of Windows" Rather than working to maintain its lead of the graphic desktop, in the early 90's Apple seemed content to doodle around with "advanced technology" and gained comfort in pointing out that it had already delivered what Microsoft was trying to build with Windows. That turned out to be a disastrous path for Apple, but the full consequences were hidden under a temporary illusion of apparent productivity. Apple had delivered A/UX, a Unix distribution with impressive support for the Mac system software; played around with porting System 7 to the PC; experimented with a new RISC workstation platform; begun work on a media architecture called QuickTime; developed the Newton handheld computer and a line of cameras, scanners and other peripherals; described a new messaging platform called PowerTalk; and incubated various other enabling technologies, but was having trouble creating any significantly marketable products beyond the Mac. Notable platform lesson: Real artists ship!