Dead End: Palm and the PDA

Zusammenfassung

Palm Computing invented the market for handheld personal digital assistants in 1996 and spent the next fifteen years unable to survive the market it created. The Palm Pilot was a genuine breakthrough: small enough to fit in a shirt pocket, fast enough to be immediate, and designed around a simplified handwriting system called Graffiti that users actually learned. It outsold the Apple Newton, the Psion Series 3, and every other PDA in the first year. Palm was acquired, restructured, spun out, and acquired again. Its engineers founded Handspring, which merged back into Palm, which was bought by HP for $1.2 billion in 2010. Eighteen months later, HP shut down the smartphone business Palm had become. The engineers who built Palm went on to build Android and iOS. The company that built the pocket computer category disappeared while everyone else built what came next.

Jeff Hawkins and the Plywood Prototype

Jeff Hawkins founded Palm Computing in 1992 after leaving GRiD Systems, a company that had built early tablet and laptop computers. Hawkins had been thinking about handheld computing since the mid-1980s — his insight was not the technology but the use case: a device that could replace the paper organizer that professionals carried, not a shrunken computer but a dedicated personal information manager.

The existing PDAs in the early 1990s had failed because they tried to do too much. The Apple Newton was a sophisticated computing platform attempting to run arbitrary applications, recognize natural handwriting, and act as a communications hub. The Psion Series 3 was a capable organizer with a miniature keyboard. Both were larger than a shirt pocket. Both were slow to start up. Both required their users to adapt to technology rather than designing technology that adapted to users.

Hawkins defined his requirements negatively: the device could not be larger than a shirt pocket, could not take more than one second to turn on and display information, could not be too expensive for individual purchase without corporate approval (target: under $300), and could not fail at handwriting recognition — not because handwriting recognition wasn’t valuable, but because it had become a symbol of PDA failure. If it didn’t work perfectly, it would be mocked.

His solution to handwriting was Graffiti: an alphabet of single-stroke characters that resembled standard letters but were designed to be unambiguous — each character had a unique stroke that no other character could produce. Users had to learn Graffiti, but learning it took twenty minutes and resulted in reliable recognition. Rather than teaching the computer to understand human writing, Graffiti taught humans to write in a way the computer could reliably understand.

The design process for the original Pilot used a piece of plywood carved to the target size. Hawkins carried the plywood block in his shirt pocket for months, reaching for it in meetings and pretending to tap on it. The exercise revealed whether the size and weight were correct before any software was written.

The Pilot: What Worked

Palm Computing was acquired by US Robotics in 1995, giving the company the manufacturing and distribution infrastructure to ship a product. The Pilot 1000 and Pilot 5000 launched in March 1996. They were immediately successful.

The Pilot ran on two AAA batteries lasting a month of normal use. It turned on in one second. It synchronized with a desktop computer through HotSync, a physical cradle that connected to the PC’s serial port and synchronized the Pilot’s calendar, contacts, tasks, and memos with desktop software in under a minute. The synchronization was bidirectional: changes on the Pilot appeared on the desktop; changes on the desktop appeared on the Pilot. This was not technically novel — Newton had something similar — but it was faster, more reliable, and simpler.

The Pilot had four hardware buttons: Calendar, Contacts, Tasks, Memo. These launched the four applications directly, without menus. The device was always on, always in sync, always one tap from the information a user needed. For professionals who carried paper organizers, the transition to Pilot was immediate and the improvement was obvious.

US Robotics sold 1 million Pilots in the first 18 months. This was faster than any personal computing product category had grown before the smartphone era.

3Com acquired US Robotics in 1997, bringing Palm with it. The Pilot was renamed the Palm Pilot and eventually just Palm. 3Com had no particular vision for Palm and regarded it as a peripheral device business — interesting but not strategic. Hawkins, Donna Dubinsky (Palm’s CEO), and Ed Colligan (marketing and later CEO) chafed under 3Com’s management, which allocated insufficient resources for Palm’s expansion.

The HotSync Model

Palm’s HotSync synchronization was widely imitated. When the iPhone launched in 2007, it synchronized with iTunes over USB using a concept functionally identical to HotSync — a one-click sync that transferred contacts, calendar, music, and apps between phone and desktop. Palm actually complained publicly that the iPhone model was their model. The irony is that Palm, by 2007, had failed to provide a smartphone that could compete with iPhone while the synchronization model they invented became the iPhone’s foundation.

The Third-Party Developer Ecosystem

Palm’s developer program was, by the standards of the late 1990s, remarkably open. Palm OS was a simple operating system — single-tasking, with a small API surface — and Palm published the APIs openly. Developers could write Palm OS applications in C using a freely available SDK.

The result was an explosion of third-party software. By 2000, over 10,000 Palm OS applications existed: medical reference databases used by physicians, games, navigation software (using external GPS accessories), language translators, Bible software, legal references. Palm’s ecosystem was the App Store of the late 1990s — an open platform where small developers could build commercially successful products for a defined device category.

Palm’s hardware expansion followed the software ecosystem. The Palm V (1999) was the product that demonstrated Palm’s design maturity: rechargeable lithium-ion battery, metal case, thinner than any previous Palm, and an industrial design that looked like the future. The Palm V was expensive ($449) and became a status symbol as well as a productivity tool. The Palm VII (1999) added wireless email access through a Palm-operated service — the first mass-market wireless data PDA in the United States.

Handspring and the Divergence

In 1999, Hawkins, Dubinsky, and Colligan left Palm to found Handspring. The split was not acrimonious but reflected a disagreement about direction. Hawkins believed that the PDA was the wrong category — that Palm should be building devices that combined phone and organizer. 3Com’s management did not want to enter the phone market.

Handspring licensed Palm OS from Palm and built its own devices. The Handspring Visor (1999) used a SpringBoard expansion slot that accepted hardware modules: modems, memory, cameras, MP3 players, and eventually phone radios. The Visor was cheaper than comparable Palm devices and sold well.

The Handspring Treo (2002) was the product Hawkins had actually wanted to build: a smartphone that combined a phone, a QWERTY keyboard, and Palm OS in a device sized to fit in a pocket. The original Treo was awkward — a flip design with a separate keyboard module — but the Treo 600 (2003) was excellent: compact, with a thumbboard integrated into the device, a camera, and cellular data. For Palm OS users who needed a phone, the Treo 600 was the right product.

Palm and Handspring merged in 2003. Palm acquired Handspring for $169 million, reuniting the founders with the company. The merged entity had both the Palm OS platform and the Treo smartphone line.

The Linux Gamble and the Smartphone Trap

By 2004, Palm was aware that its platform was in trouble. Palm OS was single-tasking, built for the PDA use case, and not designed for the demands of smartphone software: cellular radio management, background data synchronization, modern web browsing, multitasking email. Extending Palm OS to handle these requirements required a rewrite.

Palm simultaneously pursued two directions, which split its engineering resources:

Palm OS Garnet continued developing the existing Palm OS codebase. Treo smartphones running Garnet sold reasonably well but were increasingly limited compared to what Microsoft’s Windows Mobile offered on competing hardware.

Access Linux Platform (ALP), developed after Palm sold its operating system business to ACCESS Co. in 2005, was a next-generation Linux-based Palm OS. It was slow to develop and never shipped in a major product.

Windows Mobile Treos — devices running Microsoft’s operating system rather than Palm OS — were introduced in 2005. Palm was now making hardware for its competitor’s platform, which acknowledged that its own platform was not competitive but also meant that Palm’s hardware margins were not protected by platform differentiation.

The Treo 650 (2004), 680, and 700 series kept Palm in the smartphone market but not ahead of it. BlackBerry dominated enterprise email. Windows Mobile offered a broader application ecosystem. Nokia’s Symbian was the volume leader globally.

The iPhone and webOS

The iPhone’s January 2007 announcement changed the terms of the smartphone market fundamentally. Full web browsing on a mobile device, a capacitive multi-touch screen, and tight software-hardware integration were capabilities that existing smartphone platforms — including Palm — could not match.

Palm’s response was webOS, announced January 8, 2009 — exactly two years after the iPhone announcement — at CES. WebOS was a genuine technical achievement. It was built on a Linux kernel with an application layer written in HTML, CSS, and JavaScript. Applications were web applications running in a native frame. The Cards interface — each running application a card, cards laid out in a horizontal stack that could be flicked away to close — was original and arguably superior to both iOS’s application switching and Android’s.

Synergy was Palm’s answer to contact and calendar fragmentation: webOS merged contact information from multiple sources (phone contacts, Gmail, Facebook, Exchange) into unified contact cards. Just Type allowed universal search from the keyboard before any application launched.

The Palm Pre, launched June 6, 2009 on Sprint, was the webOS flagship. Critics praised the interface and the Synergy concept. Sprint’s exclusivity limited the audience. The device’s hardware — a slide-out keyboard under a rounded screen — was physically satisfying but used a plastic construction that felt fragile compared to the iPhone’s metal and glass.

Palm shipped the Pre and the Pixi (a candy-bar keyboard phone) in 2009–2010 but could not achieve the application ecosystem density that iOS and Android had. WebOS had fewer than 5,000 applications when the App Store had 100,000. Users considering a new smartphone were choosing between iOS and Android; Palm was a third option that required accepting a smaller app library.

HP and the Shutdown

Hewlett-Packard acquired Palm in April 2010 for $1.2 billion. HP’s stated rationale was webOS as a platform for a tablet, for printers, and for a broad range of devices — not just smartphones. HP CEO Léo Apotheker described a strategy where webOS would be the software platform across HP’s product lines, competing with Android’s fragmentation and Apple’s closed ecosystem.

The HP TouchPad tablet, running webOS, launched July 1, 2011. It was reviewed as a capable device with good hardware and webOS’s strong interface, but lacking the application ecosystem to compete with the iPad. HP priced it at $499 — the same as the iPad — without a compelling reason for consumers to choose it.

On August 18, 2011 — forty-nine days after the TouchPad launched — HP announced it was discontinuing all webOS hardware. The decision was part of a broader HP strategic pivot under Apotheker that also involved spinning off HP’s PC division. The TouchPad was discontinued; remaining inventory was liquidated at $99.

At $99, the TouchPad sold out immediately. Demand at $99 was massive; at $499 it had been insufficient. HP’s pricing decision had placed a platform-capable device in a price bracket where it could not find the volume to support an application ecosystem.

webOS was sold to LG Electronics in 2013, which uses it as the operating system for its smart televisions. The platform Palm’s engineers built to compete with iOS now runs TV menus.

The webOS Diaspora

The engineers who built webOS did not disappear. Many joined competitors. Matías Duarte, who led webOS’s UI design at Palm, joined Google in 2010 and became the principal designer of Android 4.0 (Ice Cream Sandwich) — the version that unified Android’s visual language and is recognizable in Android today. webOS’s Card multitasking metaphor influenced iOS 7’s app switcher. The ideas Palm’s engineers developed did not die; they were absorbed into the platforms that succeeded.

Dead End: Inventing the Category, Missing the Wave

Palm’s history is a case study in a company that correctly identified the future of computing and was unable to make the transition when the future arrived.

The Palm Pilot correctly identified that a pocket-sized personal information manager was valuable. The Treo correctly identified that a smartphone was the right device category. webOS correctly identified that a modern mobile OS should be built on web technologies. Each insight was correct and each product was built too early or with insufficient resources to execute at the scale the market eventually demanded.

The competitive dynamics that destroyed Palm were:

Platform lock-in: once iOS and Android had achieved application ecosystems that Palm’s webOS could not match, rational consumers chose the larger ecosystem regardless of OS quality
Carrier relationships: AT&T’s exclusivity with iPhone (2007–2011) and the carrier relationships that drove Android volume were not available to Palm
Capital: Apple spent $150 million developing the original iPhone; HP bought Palm for $1.2 billion and then shut it down rather than continuing to invest

Palm’s trajectory illustrates that inventing a category does not guarantee surviving it. The original PDA inventors — Hawkins, Dubinsky, Colligan — were correct about the future of computing. They were unable to translate correctness about the future into the execution, capital, and scale required to own it.