Imagine how different the tech industry might have been had Gary Kildall accepted IBM's offer, back in 1980, to license his computer operating system for a top-secret project. CP/M would have been the OS that shipped with the original IBM PC, and the world might never have heard the name of Kildall's competitor, who eventually accepted the contract: a Mr. Bill Gates.

For all the amazing advances that the computing industry has brought us over the years, some of its most pivotal moments are memorable for all the wrong reasons. Not every idea can be a winner, and not even Microsoft can avoid every misstep. But as they say, those who forget history are doomed to repeat it -- and then again, others just keep screwing up. In the interest of schadenfreude, then, here is a look back at the last 20 years' worth of blunders, fumbles, also-rans, and downright disasters that you may have forgotten about -- or wish you could.

25. IBM PS/2. The original IBM PC hit the market like lightning in 1981. Unlike earlier IBM computers, it was built with off-the-shelf parts instead of proprietary components, making it the most affordable business machine yet. But by the late 1980s, IBM found itself edged out of the market by Compaq and the other PC clone makers. Its solution? Try again with proprietary components, of course!

The Personal System/2 series, introduced in 1987, was meant to be "software compatible" with the PC, but its Micro Channel Architecture made it incompatible with existing hardware. The clones had no such problem. Like the disastrous PCjr before it and the PS/1 series to follow, the PS/2 convinced customers that lightning would never strike twice in IBM's PC division.

24. Virtual reality. In 1982, the movie "Tron" imagined a man traveling the eerie internal landscapes of a computer. Fifteen years later, the technology arrived to make it happen -- sort of.

Building a spatial interface for the Internet was all the rage in the late 1990s, owing in part to VRML (Virtual Reality Markup Language). The problem was, it didn't make much sense. The Web put the world of information at your fingertips; leave it to software engineers to find a way to send it back down the street, across a bridge, and up two flights of stairs.

The concept lives on today in Second Life, which seems to think the problem is not enough advertising. But the truth is that mainstream users have never warmed to VR. Wake us up when we can ride real lightcycles to work and meet our clients on the Game Grid.

23. Compression wars. What do you do when another software company copies your code and releases an improved version of your own product? Sue them into the ground, right? That's what System Enhancement Associates (SEA) thought in the 1980s, when Phil Katz released a clone of SEA's archive compression program, Arc.

Katz's hand-optimized assembly language provided better performance than the original Arc, but because Katz had borrowed code from SEA's product, SEA successfully sued for copyright infringement. Customers, however, felt betrayed. They saw SEA as a bully trying to stifle Katz's superior software. When Katz came up with his own high-performance archive format in 1989 -- called Zip -- they ditched Arc in droves, and SEA's business never recovered.

22. Apple OpenDoc. Long before the Cocoa and Carbon APIs earned raves from Mac OS X application developers, Apple put its weight behind another innovative programming technology. Called OpenDoc, it was a way for developers to build applications out of lightweight, modular components. After all, what is a word processor but a text editor, a spell checker, a file manager, and a few other modules all thrown together? With OpenDoc, developers could mix and match, building their applications out of all the best bits.

Unfortunately, the concept never caught on. As it turned out, most applications weren't really as modular under the hood as they appeared on the surface -- and it didn't help that those so-called lightweight components turned out to be memory hogs that ran like molasses. After five short years, the book on OpenDoc was closed.