Where PalmSource Went Wrong

Opinion: With a few million users hooked on an obsolete software platform, not even the hardware side could get moving.

The inertia of a few million users hooked on an obsolete software platform has meant that even when the software side of Palm was split off from the hardware side, the hardware side simply couldnt get moving. David Nagel pushed and pushed and pushed—and broke his back.

Heres how it went wrong: Palm was too successful.

The Palm Pilot was easily the most successful thing ever invented when it hit, and it was adopted by more people, month by month, than the original Walkman. The tiny handheld without a keyboard astonished the world, and the software world rushed to write programs for it.

So great was that success that the company was bought by U.S. Robotics, then a fiercely successful modem builder that was challenging Hayes for dominance and was moving into corporate networking. Its success in corporate networking attracted 3Com, which swallowed USR only to lose interest and spit it out.

But Palm, ah, that tasted too sweet. The 3Com giant sucked on that sweet candy and drained it of juice.

What it should have done, of course, was to invest and innovate. With the resources of 3Com, the genius of the founders, the licenses to IBM and Sony—well, that was the definitive handheld standard. The world might have had Psion, which already existed; and that might have given us the Symbian standard, but its hard to see how Windows CE could have gotten off the ground.

But 3Com was hooked on the sweetness of Palm and simply drained it of cash—and spent the money on corporate adventures that came to nothing. The despairing founders left to form Handspring, licensing the Palm operating system. Theyre the people who gave us the Treo—a very successful smart phone.

So, how could all that go wrong?

Pick up a Treo and install one of the many thousands of independently written software packages on the market. Say, an IRC (Internet Relay Chat) program. You can log on to the Internet, join an IRC server and start chatting.

Now, suppose that this phone receives a call. What happens? Easily said: "Bang! The IRC session is dead." The "Garnet" version of the PalmOS is single-tasking, and it cant run two applications simultaneously.

PalmOS (the old version) is riddled with such absurdities. The networking stack is actually an application layer, the development system has only just been taken in-house, and everybody in the Palm family knows it is simply out of date.

David Nagel made no secret of this fact—indeed, how could he? Every developer knew it. But he said his job was to produce an up-to-date version of the OS—and he did, when Cobalt was launched last year. We all expected PalmOne to launch a new machine using this OS at the same time.

Here we are nine months later, watching Palm, the main licensee of the PalmOS, launching the latest Treo, and it still runs Garnet.

To blame Nagel for this is simple idiocy. Some idiots have already done so. Im distraught at the thought that he has given up, but if it was his choice, then I cant blame him; and if it wasnt, then the people who are letting him go are the idiots.

I dare say Ill have to rephrase that when I meet his successors and listen to them explain how they are going to turn the company around by "correcting all his errors." But the truth is that the villains of the piece are those 3Com executive vice presidents who let Palm and its OS rot for the best part of a decade, while the rest of the world moved on.

/zimages/3/28571.gifClick here to read about the new Treo 600.

The problem was that the people inside Palm didnt accept 3Coms inertia. Instead, they fought it. They adapted the platform so that applications could be built into the application layer, and the application executive was multitasking internally. As long as you were part of the package that Palm itself sold, operations could multitask. It was just impossible for ISVs to plug into that without Palms help and cooperation.

And that meant that it was impossible to migrate to Cobalt without leaving all of the Garnet applications behind.

Some have argued that Java was the right alternative to Garnet. This is a little like suggesting that a fleet of a thousand 50cc moped bikes is a sensible cargo-shifting alternative to a 60-ton truck. The writers of software for Palm need a modern, multitasking alternative to Windows CE or OSE or any other RTOS capable of robust wireless control, not a shim on top of an obsolete and inappropriate base layer.

As to whether Cobalt was that modern alternative, we may perhaps find out. PalmSource, split off from PalmOne so as to clarify competition with other PalmOS licensees, has little choice but to move forward with it for the time being.

The optimist in me says: "The China Mobilesoft takeover gives PalmSource the breathing space it needs to grow in the midrange phone business while the licensees and the developers in the top-end smart-phone market move forward." And it adds: "Nagel and his team know what to do and have started most of what needed to be done; his successors will now inherit the fruits of his labors."

The realist in me hears the chorus of informed voices saying: "If theyd done this five years ago, theyd be unstoppable." I fear that success is now the albatross around PalmSource, and that it cant move forward fast enough to survive—before the investors decide to quit while theyre still ahead and pull the plug.

If the new management has a time machine, it can solve the problem. Without one, I hardly know what to say beyond, "I hope your investors are as dumb as you think they are."

Contributing columnist Guy Kewney has been irritating the complacent in high tech since 1974. Previously with PC Mag UK and ZDNet UK, Guy helped found InfoWorld, Personal Computer World, MicroScope, PC Dealer, AFAICS Research and NewsWireless. And he only commits one blog—forgiveable, surely?

/zimages/3/28571.gifCheck out eWEEK.coms for the latest news, reviews and analysis on mobile and wireless computing.