Dr. Dobb's Journal October 2000
The title of this month's column, as any lover of science fiction can tell you, is lifted, with a slight twist, from a Robert Heinlein time-travel classic, "All You Zombies -- ," a strange, wonderful, twisty short story. The history of the personal computer era is also a strange, wonderful, twisty short story, with characters who seem to die and get properly interred only to reappear some time later with dirt beneath their nails. This month is about some of those zombies.
Never do yesterday what should be done tomorrow.
-- Robert Heinlein
By the way, I've dropped some quotes throughout the column, like the one above. They're all from that Heinlein story. They are, at best, marginally relevant to the surrounding material, but they constitute a thread that runs through the column. Several other, more substantive threads twist through this story, so to keep from getting tangled up, let's get a firm grip on both ends of the bundle.
One end of the bundle of threads is the world of technology today, and you know what that's like. Here's one snapshot:
A few weeks ago I attended the O'Reilly Open Source Conference while doing my best to stay on top of MacWorld Expo, which was taking place at the same time. One conference was a gathering of revolutionaries, the other a revival meeting. I context-switched from one to the other: In an age of lightweight portable computers, fast modems, and live webcasts, it's not hard to be in two places at the same time. My portable looked bulky, though, in this age of the Palm device, and my need to find a phone connection to get online no doubt seemed hopelessly quaint to the cellularly enabled.
The other end of all the threads, arbitrarily selected for the purposes of this story, is 1980. Remember 1980?
If you had a PC in 1980 you probably bought it from Apple, Tandy/Radio Shack, Atari, or Commodore, unless it was one of the dozens of S-100 bus systems. And you thought the new Apple III looked like the machine to win over corporate America to the virtues of small computers -- you really did, even if six months later you'd deny you ever believed it had a chance.
For an operating system you were using something proprietary or you were using CP/M, but maybe you just booted right into Basic. You programmed in Basic, but you were exploring other options. A couple of early implementations of C were available on these microcomputers, one of them written by a well-known science-fiction writer not named Robert Heinlein.
You were probably exploring the online life too, if you were into small computers back in 1980. You logged onto special- interest BBSs and exchanged messages with like-minded folks and traded software. Unless you could afford one of those 1200-baud high-speed U.S. Robotics modems, you were communicating at 300 baud. 1980 was the year when Hewlett-Packard released what Byte magazine called "close to being the world's first pocket-sized personal computer:" the HP-41C programmable calculator. It was an exciting time for technology buffs.
For me it was the last days of innocence. In 1981 I started writing about this stuff full time and never saw it in the same way again. But in 1980 I was just out of grad school, working for Ray Borrill, founder of one of the first personal computer stores, maintaining the Alpha Micro systems we sold. Most of my classmates in the computer-science department had taken real programming jobs, and the most sought-after positions were at Bell Labs.
Bell Labs was clearly the place to be. We'd all used UNIX throughout grad school, and knew where it came from. It was said that managers at Bell Labs were so interested in pushing the limits of technology that they would pull the plug on a project if it looked like it had some near-term practical application. We didn't know much about Richard Stallman and free software, although we took a lot of its principles for granted. Back there in the Midwest, we'd never heard of Xerox PARC.
Ancestors are just people.
-- Robert Heinlein
Steve Jobs had heard of Xerox PARC, though. He and Bill Atkinson and others paid a legendary pair of visits to PARC in late 1979 and brought home to Apple the essence of the PARC graphical user interface. Jobs later admitted that he had missed the real essence of what was happening at PARC. Lisa and the Mac only implemented the surface of the object-oriented approach the PARCers had pioneered. The idea that software should be objects all the way down was radical, possibly too radical for the time.
If the Mac GUI was thin, the earliest versions of Windows were even thinner -- a translucent layer of GUI over MS-DOS. It wasn't hard to see the skeleton beneath. Windows got thicker over time. But back in 1980 Bill Gates had said, "UNIX's inherent flexibility...will make [it] the standard operating system for the computers of the '80s..." Well, not quite. He really said "Xenix," not "UNIX." Xenix was Microsoft's UNIX. Gates was sold on UNIX then. A little deal with IBM that happened shortly after he said this changed his mind, and within four years Microsoft had sold off Xenix to the Santa Cruz Operation. During the next two decades of UNIX balkanization, SCO stayed strong. Its SCO Forum was the place to be if you were a Unixhead. But neither SCO nor any other vendor brought the disparate threads of UNIX together.
In the Apple and Microsoft camps there was a sense that an object-based model was the right direction to go with operating systems. At Microsoft, one of the projects that might have led to something along these lines was Cairo, which eventually turned into a collection of technology, some of which showed up in some products. Apple made a lot of fuss about OpenDoc, which had the potential to deepen the operating-system model, but Apple pursued a lot of next-generation operating-system candidates through the late '80s and all of the '90s, and it didn't do anything with any of them. By the mid 1990s it was understood that Apple had to get a new operating system or die. Microsoft had not only successfully developed a new operating system, NT, but it had effectively killed off an even better one, OS/2, when it no longer looked like it would serve Microsoft's interests.
Bell Labs, meanwhile, was not so secretly working on the successor to UNIX. Named after an Ed Wood movie, it was called Plan 9, and was conceived in the late 1980s. The 1995 release saw use in a few labs around the world.
Plan 9 was an operating system for distributed computing. It employed just one protocol for all references to, and communication among, all resources: processes, programs, data. It gave users a transparent (this word has too many meanings in computing, doesn't it?) computing environment in which they didn't need to know anything about the details of what process was running on what device.
There were more devices to deal with by this time. Palm, a division of 3Com, was a distinct computing platform that insisted on being addressed. Palm users were not all as fanatical as the users of that HP-41C, but there were a lot more of them. 3Com, riding high on Palm's success, acquired another company in the mid 1990s: U.S. Robotics. Modems had gotten faster, and there were a lot of 300-baud modems in dumpsters that were never going to spring back to life. UNIX, the real zombie of the operating system world, found a rallying point with an Open Source kernel called "Linux." One by one, the commercial UNIX implementations began to drop out. Not all of them, certainly, but even the companies that were able to keep selling their proprietary UNIX variants began building bridges to Linux and Open Source.
Open Source was the term almost everyone used now, although the terms of the licenses varied. The Free Software movement that Richard Stallman had championed was trying to achieve credibility by distancing itself from that "free" label, and from Stallman. The Open Source Conference that I attended this year had been called a Perl conference two years earlier. It was still mostly about Perl and other scripting languages, but the Open Source label was hot and O'Reilly shrewdly embraced it.
It is earlier when you think.
-- Robert Heinlein
Fast forward to today. To the other end of all the threads. Over two years after Steve Jobs (or Gil Amelio, if you prefer that view of history) pulled Apple back from the dead, Apple is prospering and we're still waiting for that new operating system. This is the operating system, you will recall, that was Apple's only hope for survival, the operating system that there wasn't time to build, that had to be bought from Steve Jobs. Turns out that after they bought it they still had to build it, but the company's doing very well indeed with the same old buggy, antiquated operating system that nobody could stomach three years ago.
Credit it to that whole new line of computers in colorful, see-through cases. A lot of what's new about the new Apple machines is what's missing. No diskette drive, no SCSI or serial or ADB ports, no CD-ROM drive. In the latest addition, the cube, there's no fan. That's the Jobs influence, for sure. The other no-legacy-hardware decisions may have been in Amelio's master plan, but they fit with Jobs' break-with-the-past style.
Apple announced a lot of things at MacWorld Expo (although it didn't release OS X), but the most dramatic was the cube. For $1799 you get a 450-MHz G4 CPU, 20-GB hard drive and 64 MB of RAM, two USB and two Firewire ports, 10/ 100 Ethernet and a 56K V.90 modem, a DVD-ROM drive, and an ATI Rage 128 Pro video card. You can connect a VGA monitor or one of the new displays from Apple. You also get a remarkable looking piece of hardware: 7.7 inches square and a little taller, with that unusual and beautiful case design. And the DVDs eject vertically, like toast from a toaster.
Who will buy it is the subject of considerable controversy, although the answer will be apparent soon enough. It's not Palm Pilot cheap, but unlike the 20th Anniversary Mac, it's cheap enough and powerful enough that some people will buy it just because they like the way it looks. And it is pretty. All the surfaces are pretty at Apple now, although nobody on the outside seems to know what's going on inside. That particular transparency is no longer the Apple Way.
The other day Brandon, who does web development for my partner Nancy's farm and restaurant, and who works in my office, announced that he had a bug in his machine. I went to look, and sure enough, through the translucent case of his iMac we could see the silhouette of the tiny critter, a baby spider I think, crawling around inside. I said it wouldn't be lonely. It's not the only bug in there.
A paradox may be paradoctored.
-- Robert Heinlein
One thing that's not inside any Mac is OpenDoc. OpenDoc hasn't disappeared, but the latest OpenDoc-inspired technology turns out to be all about surface appearances, too.
DesktopX from Stardock is technology that lets desktop gadgets talk to each other. (That's the OpenDoc influence.) It sits on top of the default Windows shell, uses standard Windows COM plumbing, and lets users completely redesign their user interface without programming. Want your Windows machine to have the Aqua look and feel of MacOS X? Can do. OEMs may find this more exciting than end users, because, now that Microsoft can't dictate down to the last pixel what the Windows desktop looks like, the ability to customize that look is going to be an important differentiator for OEMs.
Stardock is also working on the object-based filesystem that would make this technology more than just eye candy. If DesktopX hasn't already put them in Microsoft's crosshairs, that certainly would. It would be the essence of Cairo, back from the dead and not owned by Microsoft.
A stitch in time saves nine billion.
-- Robert Heinlein
The keynote speaker at the Open Source Conference was Andy Hertzfeld, the chief architect of the original Mac GUI. Hertzfeld went on from Apple to General Magic and is now one of the principals at Eazel. Eazel is all about putting a friendly face on Linux.
That's a fine idea, and Hertzfeld seems to have exactly the right instincts about it. Work in the Open Source model, build on existing standards and technologies, work with other Open Source developers. He thinks that a lot of Linux and Open Source developers could benefit from developing a little more respect for the end user, and that Linux will never have a chance on the desktop until it has a world-class GUI. Imitating Windows is a losing game. You've got to look beyond.
Beyond in terms of ease of use of the user interface, that is. Hertzfeld's a GUI guy, and that's what Linux needs if it is to become a successful desktop OS. That's the message, and there's nothing wrong with it. But it's more eye candy, isn't it? Is anybody working on anything deep? Well, lots of people are, of course. Since the Open Source Conference evolved out of a Perl conference and still remains a very scripting-oriented show, it's not fair to hold it up as a model of what's happening in Open Source generally. There are deep projects afoot.
One of the deep projects is another zombie. It even has a zombie name. Lucent, the zombie revivification of Bell Labs, recently released Version 3 of Plan 9, and this time it's free. Plan 9 is small, powerful, and very flexible, and is seen by some as a serious, if long-shot, competitor to Linux and UNIX. Those who are not troubled by its lack of development tools, applications, middleware, and serverware may even be able to make use of it right now. High-performance technical computing, yes. Commercial web development, probably not today.
Okay, Plan 9 is mostly an academically interesting technology today, but that could change. Vita Nuova Holdings (http://www.vitanuova.com/) is working on the first physical distribution of Plan 9 and plans to offer support and enhancements.
I guess we'll know it's really arrived when Hertzfeld announces plans to develop a GUI for it. By the way, Hertzfeld went out of his way in his keynote to acknowledge Richard Stallman. Given that the Open Source movement finds it useful to let Stallman function as the radical fringe figure that makes the rest of them look mainstream, and given that Stallman is incapable of countering this ploy, it's nice to see the real father of Open Source acknowledged in this way. I'm not sure Stallman counts as a zombie since he never went away. And those who know Stallman can testify that he never will.
I know where I came from, but where did all you zombies come from?
-- Robert Heinlein
And those U.S. Robotics modems? Those are the real zombies. The U.S. Robotics name went away when 3Com acquired U.S. Robotics in 1997, a disastrous acquisition that some say was the undoing of 3Com. Now 3Com is spinning off its modem business and will give it back its old name. Before the year is out you'll be able to buy U.S. Robotics modems again. You probably won't, but makers of low-end PCs are likely to keep buying analog modems for years to come because the price is right. U.S. Robotics could have a new life.
This fall I'll be meeting an old friend at the Vintage Computer Festival. My old boss Ray Borrill, who has more than enough irony in his soul not to be offended by being mentioned in a column about zombies, will be there selling his Apple 1. Now that's a zombie.
DDJ