Phoenix in Cupertino

The reason is not hard to discover. You could have read it in June, for example, in the Merc. (That's the San Jose Mercury News, the daily paper of the Silicon Valley, known in cyberspace as http://www.sjmercury.com/.) The reason you will be so welcome, notwithstanding the occasional "Welcome to California. Now go home!" bumper sticker, is the desperate need for fresh meat. There's a worker shortage in the Valley. According to the Merc, the unemployment rate in Santa Clara County dropped to 3 percent in May from 3.1 percent in April. That's full employment, the way economists view it. More than full, it would seem. "There are," the Merc said, "simply more potential jobs than workers in Silicon Valley these days." The Merc also reported the May unemployment rate across the Bay in Alameda County as 4.3 percent (still impressive compared with a Statewide 6.3 percent) and up the peninsula in San Mateo County as 2.6 percent.

Santa Clara County alone added something like 30,000 new jobs in the first half of this year. Granted, they're not all programming jobs. There is also a critical waitperson shortage in the Valley. (You might want to bring a bag lunch.) But Sun Microsystems, to take a not-altogether-random example, currently has Bay Area openings for about 900 people.

So unless you smell awfully bad, you will be welcome in the Valley. Nice, huh? But what reports like the Merc's don't always get around to mentioning is that the average duration of a programming job in Silicon Valley is something like two years.

Okay, I don't have any survey results to back that up, but it is the accepted urban folklore statistic. It's apparently based on the following plausible logic:

The average programming project lasts about two years.

The end of a project is an excellent time for considering one's options.

It's often easier to move up by moving out than to get promoted by people who know your personal quirks as well as your professional accomplishments.

Anyway, short duration jobs are common. People get fired, get tired, get better offers.

And sometimes stranger things happen. Sometimes the job just disappears out from under you. Sometimes the whole company disappears. Once in a while, the company you signed on with morphs into some entirely different entity.

This is the story of one company that completely changed its mission not once but twice in five years, jettisoning programmers and their work each time.

Oddly enough, it's a success story.

The Incredible Alliance

In the beginning, there was Pink.

Pink was Apple's first attempt to develop a successor to the Macintosh operating system. While the Mac had all the visual appearance of an object-oriented OS, the object model didn't go much below the surface. Pink was to be completely object-oriented from the ground up. Also, it was to transcend the company's single-hardware platform heritage. By early 1991, Pink was up and running on Intel hardware, specifically, an IBM PS/2.

But it wasn't finished, and finishing was a tricky job, apparently-because Apple went looking for some help. Later that year, Apple and IBM signed a wide-ranging agreement that, among other things, created two new companies to be jointly owned by Apple and IBM. Kaleida would push the limits in multimedia, and Taligent would produce a new, next-generation microcomputer operating system. It was to be independent of hardware platform and completely object oriented from the ground up. It was to be Pink. Apple was going to codevelop Pink with IBM. Well, maybe not exactly Pink. As the story developed, it seemed that more and more IBM technology was going to be a part of the Taligent OS.

Wait a minute-Apple and IBM working together? The same IBM that Steve Jobs had painted as Apple's only meaningful rival? That Apple's memorable 1984 Super Bowl advertisement portrayed as Big Brother from Orwell's 1984? Now Apple was sharing its plans for key technology with IBM? Jobs would have been rolling over in his grave if he had been dead rather than off running a new company called NeXT and cutting his own deals with IBM.

It was fairly surreal for the Apple and IBM employees who went to Taligent and found themselves working for bosses still loyal to the opposition. Not a typical Silicon Valley career move, maybe, but perhaps a portent of other weird twists to come. Ignoring the politics as much as possible, the Taligent programmers buckled down and wrote a lotta lines of code.

Then the roof fell in.

The Taligent operating-system initiative collapsed. The partners lost faith. The market dissolved. The bureaucracy collapsed under its own weight. Stories differ. But whatever the immediate or ultimate cause, the plan for an Apple/IBM-developed next-generation hardware-independent object-oriented microcomputer operating system fell apart.

The Fate of the Foosball Table

We won't do a new operating system, the Taligent braintrust decided. People don't want a new operating system. What we'll do is, we'll build a thing, a layer, that sits on top of the operating system, any modern operating system, and provide all the object-oriented features anyone could want.

We'll provide critical services to applications. Shorten the development cycle.

And they did. It ran on top of AIX, HP-UX, OS/2, and Windows NT, and it was called CommonPoint.

Early in 1994, Hewlett-Packard decided to become a Taligent partner. Now you could be an Apple programmer working for an IBM boss who reported to HP. Or some combination thereof. Twisteder and twisteder.

The Taligent programmers ignored all that and buckled down and wrote a lotta lines of code: more than a hundred object-oriented frameworks, well over a thousand classes.

Taligent shipped it. They couldn't sell it. The Taligent braintrust said, people don't want an object-oriented layer that sits on top of the operating system. What we'll do is...

What they did was, as they phrased it at their own web site (http://www.taligent.com/), "focus on the technology and leave marketing to their partners."

By spring 1996, "partners" meant IBM. IBM bought out Apple and HP, and Taligent became a wholly owned subsidiary of IBM.

Which may turn out to be the smartest move the company ever made. Not everyone would agree, of course. Not the programmers who lost their jobs each time Taligent regrouped.

In 1995, Andreas Kyriakou and Roger Thornton had launched the Taligent Memorial Alumni Page (http://www.best.com/~zaveri/talumni/), a site for exemployees of Taligent following the second death of the company. The site's mission, tone, and attitude were all hinted at in its disclaimer:

Taligent and the Taligent logo, People, Places, and Things, A sense of community, Your Aunts, Uncles, and anything else you can think of are registered trademarks of IBM.

The Memorial site is still there. It includes a job board for Taligent exemployees, memorabilia, and a fair amount of black humor. It doesn't appear to have seen many visitors since 1995. The Bulletin Board features some two-year-old queries about the possible fate of the foosball and pool tables and "the Quicktake images from the ill-fated CP Easter egg," but little else of lasting importance.

You get the impression that there wasn't a lot of licking of wounds because...right, they all went out and got new jobs.

Born Again (and Again)

Of those who remained, the top managers have some pretty impressive credentials.

General Manager and Chief Executive Officer Debbie Coutant managed the porting of object-oriented system software to the PowerPC RISC platform at Apple and went on to manage operating-system development during Taligent's first life. She has a patent for techniques for symbolic debugging of highly optimized code, and has numerous publications regarding code generation and optimization for RISC processors.

Vice President and Chief Technology Officer Mike Potel was director of software engineering at Apple, where he initiated both Macintosh System 7.0 and the original "Pink" project.

Director, Core Technologies, Mark Davis is Mr. Internationalization. He cofounded the Unicode effort, is the president of the Unicode Consortium, is a coauthor and editor of the Unicode Standard, Version 1.0 and of the new Version 2.0. He coauthored the Macintosh KanjiTalk and the Macintosh Script Manager (which later became WorldScript), authored the Arabic and Hebrew Macintosh systems, and architected the CommonPoint international frameworks and Java 1.1 international libraries for Taligent.

Director of Planning and Program Management Rich Castro was formerly in charge of system-software development for new Macintosh systems, including the first open Macintosh system, the Mac II, and the first Quadra and PowerBook systems.

Director, Desktop Frameworks and VAB, Mairé Howard started on the Pink project at Apple, before Pink became Taligent.

There's a reason why they've stayed. Well, okay, the benefits are good, but so is the company's mission.

The stated mission is "to provide solutions to application developers, web-site programmers and groupware users...increase developer productivity by shipping products built with innovative object technology and cross-platform distributed computing...develop framework technology for the IBM VisualAge Product Family... provide Java solutions for software and web developers...make it easier for advanced developers and web novices alike to utilize the Web for software development" and blahblahblah.

Translated, this means: Unbundle all the technology that went into CommonPoint and get IBM either to release it in its own products or license it to others.

And it's working. Taligent is onto something.

Sun Microsystems thinks so. It has licensed technology from Taligent. So has Netscape-Java technology.

Sunsoft glommed onto IBM's JavaBeans Migration Assistant for ActiveX, a tool that turns ActiveX controls into JavaBeans. It's cool: It analyzes an ActiveX control's properties and creates a 100 Percent Pure Java container that takes on the features of the ActiveX control, then implements the component functions. Sunsoft is going to include it in its JDK. From the CommonPoint effort, Taligent gained a large collection of frameworks that bring huge chunks of functionality to developers. The Desktop framework gives an infrastructure for building components. The Compound Document framework gives developers a quick start on creating cross-platform compound-document applications. The web frameworks provide interfaces for extending a server's functionality, with support for Common Gateway Interface (CGI), Netscape Server (NSAPI), and Internet Connection Services (ICSAPI) servers. The Graphics frameworks...

Well, you get the idea. There's a whole bunch of object-oriented, framework-based technology in CommonPoint. And Taligent is pushing it out into the market. Or IBM is.

I guess it's easier to develop hot technology when the guys before you have already written most of it. Like inheriting from a rich uncle. And having another rich uncle to sell it for you doesn't hurt, either. Meanwhile, they're developing cute Java widgets and telling the same story everybody else in the software industry is telling: All of our software is an extension to Java. Sure, why not?

They seem to be having fun, are delivering some killer technology, and are just starting to get some (good) press. A nice spot to be in. And the benefits package isn't bad either.

DDJ