GEM was an early GUI for the IBM PC and compatibles and, later, the Atari ST, developed by Digital Research, the developers of CP/M and, later, DR-DOS. (Digital Equipment Corporation was a different company.) So what was it, and what happened to GEM?
It was very similar to the Apple Lisa, and Apple saw it as a Lisa/Macintosh ripoff and sued. While elements of GEM did indeed resemble the Lisa, Digital Research actually hired several developers from Xerox PARC.
DRI demonstrated the 8086 version of GEM at COMDEX in 1984, and shipped it on 28 February 1985, beating Windows 1.0 to market by nearly 9 months.
I read about GEM in the early 1980s, but didn’t actually see it until 1993 when I was in college. When using a friend’s 286, I spied a copy of GEM installed on the hard drive, so I booted it up. Having used a number of 1980s GUIs previously, I had no trouble figuring out GEM. The problem was the lack of software.
I’m sure performance was an issue on the pokey 4.77-MHz 8088 CPUs that were common in 1985. On the 286 I was using in 1993, which probably was 10 or 12 MHz, the speed was tolerable.
Competing with Windows
The lack of speed and lack of software pretty much doomed GEM on the PC. Windows didn’t do much better; it was 1990 before Windows, finally in version 3, gained widespread adoption and use. But DRI discontinued GEM in 1988, two years earlier.
I don’t think it was coincidence. By 1990, the 486 CPU was out. Few people could afford it, but it existed, and that pushed down the prices of 286 and 386 CPUs. Windows 3.0 was marginal at best on anything but the fastest 286s, but ran fine on the 386, and in 1990, the 386 was reasonably affordable.
In 1990 a perfect storm happened: PCs fast enough to run Windows existed, and Windows got to be good enough for people to want to use it.
One could argue DRI bowed out too soon. Then again, it’s questionable whether it would have won against Windows anyway. Microsoft was the larger company and had OEM agreements with all of the major PC makers. GEM only came with PCs from Amstrad and Atari, neither of whom were big PC sellers in the United States. They did better in Europe, and that’s why GEM did better in Europe than it did here.
Finding refuge in the Atari ST
In the meantime, GEM survived on the Atari ST. With an 8 MHz Motorola 68000 as the baseline, speed wasn’t a terrible concern. The 8 MHz 68000 was roughly equivalent to an 8 MHz 80386SX, had such a chip existed. In 1985, it was hot stuff. Hardware-wise, the ST matched up closely to its contemporary Macs and outperformed the PCs of its day, making GEM performance on the ST pretty much a non-issue. And since GEM was the default environment for the ST from the date of its release, available software was less of an issue. Third parties were going to develop for the ST, so they were going to use GEM.
Strangely, Apple didn’t sue Atari like they did DRI, and GEM on the Atari remained very Mac-like. I don’t know why Apple didn’t see Atari as a threat. Given what Jack Tramiel had done to the Apple II while heading up Commodore, Apple shouldn’t have wanted him competing with the Mac. Really the only thing that saved Apple from a repeat performance was Tramiel’s lack of understanding that the ST and its operating system needed refinement every few years. By the late 1980s, the ST line looked more dated than it needed to.
But the bigger problem was software piracy. Piracy was common on the ST, and that made developers less enthusiastic to continue ST development, and instead, they ported their good ST software to other machines. The ST eventually died due to lack of software as the platform aged. By the early 1990s, developing for PCs running Windows was more profitable, and a sufficiently powerful PC running Windows could match or exceed the ST both in performance and price, something that wasn’t true in 1985.
What happened to GEM? Niche uses
And it’s not entirely fair to declare 8086 GEM’s dying date as 1988. It lived on for several years as a graphical runtime library for DOS, most famously used by Ventura Publisher, one of the more popular desktop publishing packages for PCs.
In hindsight, it’s possible to see what went wrong. Had DRI been supplying the underlying operating system to PC makers (call it CP/M, call it DOS, whatever) and convinced one or more of the large US PC makers to bundle GEM with their PCs, and had DRI developed application software that used GEM, it’s easy to imagine an alternate history where GEM thrived the same way Windows did, and perhaps did it a bit sooner, especially if GEM had one or more killer apps and drove demand for ATs that could run it.
Linux vendor Caldera ended up owning the old Digital Research intellectual property. Caldera released GEM as open source under the GNU GPL in 1999, and it’s one of my favorite examples of how open source isn’t a silver bullet. As it stands, GEM is just another part of Gary Kildall’s mystique, sadly.