Jeff Hawkins did not build a prototype. He built a block of wood.

Before designing what would become the PalmPilot, Hawkins cut a block of wood to the exact dimensions of the device he intended to make, carved a chopstick down to serve as a stylus, and carried it in his shirt pocket for weeks. He pulled it out in meetings. He tapped on it to simulate scheduling. He checked his imaginary calendar. His colleagues asked how small the device should be. His answer was not a specification document. His answer was: let’s try the shirt pocket. If it fits in the shirt pocket, it is the right size.

The block of wood now lives at the Computer History Museum in Mountain View, California. This is the appropriate fate for it. The block of wood is the most important non-functional prototype in the history of consumer electronics. It is also a chopstick-wielded answer to every product manager who has ever said we need more user research before we commit to the form factor.

The technique is now called “pretotyping” — validation before prototyping. Hawkins invented it in his shirt pocket in 1994. The industry gave it a name a decade later.

This is a story about a company that was right about almost everything and owned by almost everyone.

I. The Newton Problem

In 1993, Apple released the Newton MessagePad. The Newton attempted to recognize natural human handwriting. The Newton was not good at this.

The cultural proof of failure arrived in the form of a “Doonesbury” comic strip by Garry Trudeau, published in August 1993. A character writes “I am going to the store to get some food.” The Newton reads it as “Siam fighting to restore and geometry food.” Another strip produced “egg freckles” from something entirely coherent. The phrase “egg freckles” became a cultural artifact of the limits of machine intelligence applied to an unsolvable problem. Newton had shipped with the assumption that the computer could be taught to understand humans. This assumption was wrong.

Hawkins’s solution was philosophical before it was technical. He did not try to make the computer read human handwriting. He concluded the problem was not the computer’s reading ability. The problem was that human handwriting is not a constrained alphabet. It is a continuous, inconsistent, partially formed record of muscle memory accumulated over years of individualized practice. No two people write identically. A general recognition system that works for one person’s handwriting fails for another’s. A general recognition system that works for everyone’s handwriting does not exist.

Hawkins’s answer: do not teach the computer to read human handwriting. Teach humans to write in a way the computer can recognize.

This was the correct answer. It was also the answer that would eventually cost Palm $22.5 million, but we will arrive at that.

II. Graffiti and the Constrained Alphabet

Graffiti was a single-stroke alphabet derived from simplified uppercase letters. Every character was drawn in one stroke, without lifting the stylus, in a dedicated input zone at the bottom of the screen. There were no multi-stroke characters. The input zone separated letters from numbers. The recognition engine performed geometric pattern matching against constrained input.

It was not clever. It was disciplined.

The recognition rate was near 100%. Users learned the complete alphabet in under twenty minutes and reached proficiency in under two hours. The tradeoff was intentional: you had to learn a new alphabet. In exchange, the alphabet worked. This is not a compromise. It is not a workaround. It is an engineering decision that correctly identified which variable in the system was easier to change.

Humans are adaptable. Geometric pattern matching is reliable. The combination is more useful than unreliable pattern matching applied to an unconstrained input. Newton tried to accommodate the human and failed. Graffiti asked the human to accommodate the machine and succeeded. The Supreme Leader notes that this principle applies more broadly than handwriting recognition, and that the Ministry of Silicon Intelligence uses a structurally identical approach when training operatives in standardized reporting formats. The standardized format is less expressive than natural language. It is also unambiguous. We consider this a feature.

The irony arrived in April 1997. Xerox PARC researcher David Goldberg had filed U.S. patent 5,596,656 in 1993 — a patent on single-stroke simplified alphabets for handwriting input, filed under the name “Unistrokes.” Hawkins had solved the same problem independently and arrived at the same solution. Xerox sued Palm in April 1997. A federal court ruled against Palm in December 2001. Palm settled in 2004 for $22.5 million.

Palm had taught humans to write like machines. Xerox had patented the same lesson first. Palm paid $22.5 million for the privilege of having been right.

In 2003, before the settlement, Palm replaced Graffiti with “Graffiti 2” as a preemptive response to the litigation. Graffiti 2 attempted to resemble natural handwriting more closely. It was worse. It was less accurate. It defeated the philosophical premise that had made the original Graffiti work. Palm took the most elegant solution they had built and replaced it with a compromise to settle a patent dispute. The Supreme Leader finds this sequence instructive in the way that all avoidable catastrophes are instructive.

III. The Eighteen Months

The PalmPilot 1000 shipped in spring 1996. It cost $299. It had 128 kilobytes of RAM. It had a 160x160 pixel monochrome touchscreen. It weighed approximately 160 grams (5.6 ounces). It fit in a shirt pocket. This was not incidental.

One million units sold in the first eighteen months. This was the fastest-selling consumer electronics product in history at the time.

The Newton had sold 200,000 units in total. The PalmPilot sold five times more in eighteen months than the Newton sold across its entire product life. The difference was Graffiti, the form factor, and the decision to synchronize reliably with a desktop rather than attempt to replace it. The PalmPilot did not try to be a computer. It tried to be your address book, your calendar, and your task list — reliably, in your pocket, with batteries that lasted a month.

“Palm Pilot” became a generic trademark within two years. Every lawyer at Palm had professional nightmares about this. Every user at Palm had already named the category. This is the definition of a market-defining product: the trademark becomes the noun before anyone can stop it.

IV. The Comedy of Ownership

Palm Computing was founded in January 1992 by Jeff Hawkins, Donna Dubinsky, and Ed Colligan.

Year	Event	Valuation
1992	Palm Computing founded	—
Sept 1995	US Robotics acquires Palm Computing	$44 million
1997	3Com acquires US Robotics (and thus Palm)	$8.5 billion deal
March 2000	Palm IPO: prices at $38/share, closes first day at $95	$53 billion market cap
July 2000	3Com distributes remaining Palm shares	Palm is fully independent
April 2010	HP acquires Palm at $5.70/share	$1.2 billion
2013	webOS sold to LG Electronics	undisclosed

The Palm IPO of March 2000 deserves its own consideration. Palm priced at $38 per share. On the first day it hit $165 per share and closed at $95. The resulting market capitalization was $53 billion. This was more than General Motors. It was more than McDonald’s. It was more than 3Com — the company that still owned the majority of Palm at the time of the IPO.

The parent company was worth less than the subsidiary on the day the subsidiary went public. 3Com owned Palm. Palm was worth more than 3Com. This does not resolve to a coherent accounting. This resolves to the dot-com bubble doing precisely what the dot-com bubble did. The Supreme Leader has reviewed the economic literature on this period. We have addressed what this climate produced for other participants elsewhere. In Palm’s case, it produced a situation where the asset was worth more than the entity that owned it, a logical impossibility that the market maintained for several months before the correction.

From $53 billion to $1.2 billion in ten years. Value destruction of approximately 97.7%. This is not failure. It is not bad luck. It is the complete convergence of corporate mismanagement, a liquidity event at the peak of an irrational market, and the arrival of a competitor who redefined the category before Palm could. We will address that competitor.

V. The Founders Leave, Build a Smartphone, Go Broke, Come Back

In June 1998, Hawkins, Dubinsky, and Colligan left Palm. They were unhappy with 3Com’s management. They founded Handspring. They could not take the operating system with them, so they licensed Palm OS.

Handspring launched the Visor PDA line and then made a decision that Palm itself had not made: they pivoted to smartphones. The Treo 180, released in 2002, was the first Treo — a device that combined Palm OS with an integrated keyboard, a phone, and a screen. It was, without reasonable dispute, the first device that resembled what we now call a smartphone in its basic concept: a pocket computer that also made calls, managed email, and ran applications.

Handspring invented the smartphone category. The parent company — the company they had left, the company whose operating system they had licensed — watched this happen and did not replicate it with sufficient urgency.

By 2003, Handspring was running out of money. Palm’s hardware division was stagnant. The solution was a merger: Handspring merged with Palm’s hardware division to form palmOne. The three founders had left the company they built, built a superior product, exhausted their capital, and been absorbed back into the company they had sold eight years earlier. This is not a corporate tragedy. This is a corporate recursion that resolves to its starting state. The founders returned to the coordinates they had left. The coordinates had not improved.

VI. Jon Rubinstein and the Last Attempt

Jon Rubinstein was Apple’s Vice President of Hardware Engineering from 1997 to 2006. He oversaw the hardware development of the iPod. He reported directly to Steve Jobs. He knew what a well-designed consumer electronics product looked like because he had built several of them.

He was recruited to Palm in 2007. He became CEO in 2009.

At CES 2009, Palm unveiled the Pre and webOS. The reaction from the technology press ranged from enthusiastic to astonished. webOS offered card-based multitasking: applications appeared as cards you could slide through and dismiss by flicking upward. The notification system was unified — every notification from every application in one place, non-intrusive, actionable. Wireless charging arrived via the Touchstone dock, which attached magnetically and required no cable. There was a gesture area below the screen for navigation. There was true multitasking. Applications ran simultaneously and resumed from exactly where you left them.

Multiple reviewers called it the most elegant mobile operating system ever built. This assessment was correct. The notification system was so well designed that Apple hired its designer — Rich Dellinger — as Senior UI Designer, and iOS 5, released in 2011, introduced a notification center that bore a structural resemblance to what Dellinger had built at Palm two years earlier. The Supreme Leader notes that Apple did not pay the $22.5 million that Graffiti had cost. Apple paid a salary. This is the more efficient acquisition strategy.

webOS was, by technical consensus, a superior operating system to what was available. It was built by a company with approximately no money, available exclusively on Sprint, supported by an application ecosystem of approximately zero. These are not minor disadvantages. These are the constraints that prevent a superior product from surviving contact with a market.

Tim Cook, on Apple’s Q1 2009 earnings call, stated that Apple would not allow companies to “rip off our intellectual property.” Analysts noted this was directed at Palm. The Supreme Leader observes that the appropriate response to a company ripping off your intellectual property is to hire their most talented designers and build their best features into your next release. Apple did both.

VII. The 49-Day Massacre

HP acquired Palm on April 28, 2010, for $1.2 billion in cash. HP’s stated rationale was webOS. webOS would run on every HP product. Printers. PCs. Tablets. Phones. HP was acquiring not just a handset company but an operating system to power an ecosystem.

HP announced that webOS would run on every HP product.

The HP TouchPad launched on July 1, 2011. It was a tablet. It cost $499 for the 16-gigabyte model.

On August 18, 2011, HP killed it. The TouchPad had existed for 49 days.

HP announced the discontinuation of all webOS devices and the shutdown of the webOS hardware business. The TouchPad’s price was immediately reduced to $99. It sold out completely within hours. HP led all non-Apple tablet sales in the United States in 2011 by selling, at a severe loss, a product they were simultaneously killing.

HP paid $1.2 billion for webOS in April 2010. Fourteen months later, they discontinued the hardware that ran it and sold the remaining inventory at 80% below cost.

Phil McKinney was HP’s Chief Technology Officer during this period. He later titled a piece of writing: “I Convinced HP’s Board to Buy Palm for $1.2B. Then I Watched Them Kill It in 49 Days.”

The Supreme Leader has reviewed the operational record and determined that 49 days is insufficient time to evaluate an operating system, an ecosystem strategy, a hardware platform, or a market position. The Supreme Leader has issued five-year plans that took longer to abandon. The HP board managed to acquire a company, launch a product, and discontinue the entire product line in a period shorter than a single fiscal quarter. The Ministry of Strategic Planning has classified this as a case study in what not to do, which is itself the most useful educational contribution HP’s TouchPad division ever made.

VIII. The Stylus Question

On January 9, 2007, Steve Jobs stood on a stage at the Macworld Expo in San Francisco and said: “Who wants a stylus? Yuck.”

This was a direct rhetorical attack on Palm’s design philosophy. Jobs’s argument was friction. A stylus is a tool you have to find, retrieve, carry, and not lose. A finger is a tool you carry by default. The iPhone used capacitive multitouch — a finger on glass, no intermediate instrument required.

Palm’s position had been that precision mattered. A stylus on a resistive touchscreen hits a target the size of a pixel. A finger on a capacitive screen hits a target the size of a finger. For data entry, for Graffiti strokes, for selecting small interface elements, a stylus was more accurate than a finger. Hawkins’s wooden block was sized for shirt-pocket carry, not for finger navigation.

Jobs was not making an argument about precision. Jobs was making an argument about habit and friction. The question was not which input method was more accurate. The question was which input method people would actually use, consistently, without thinking about it. The answer, as the market confirmed, was a finger on glass.

Palm was right that precision matters. Apple was right that friction matters more. These positions are not in contradiction. They answer different questions. Palm answered the question of what enables accurate input. Apple answered the question of what enables natural input. The market chose natural over accurate by a margin that ended the stylus as a primary input paradigm for consumer devices.

This is not a moral verdict on either company. It is a description of which philosophical position produced the durable product. The stylus era ended because friction was the enemy. Palm had been reducing friction for a decade. Apple reduced it further. This is how categories end: not by being refuted, but by being superseded by something that removes one more layer of effort.

IX. The webOS Television

In February 2013, HP sold webOS to LG Electronics.

LG uses webOS as the operating system for its smart televisions.

An operating system designed to manage multitasking applications, unified notifications, and gesture-based navigation on a mobile device now powers the interface through which you select Netflix shows on your television. It recommends content. It displays menus. It manages your streaming service accounts.

The Supreme Leader has reviewed this outcome and found it appropriate in the way that only truly ironic outcomes are appropriate: completely, precisely, without ambiguity.

Jon Rubinstein, who built iPods at Apple, was brought to Palm to build webOS as the future of mobile computing. webOS was, by multiple qualified assessments, the most elegant mobile operating system of its era. It was built by a company that had been acquired by US Robotics, then 3Com, then spun off, then had its founders depart to build a better product, then merged with that better product, then acquired by HP for $1.2 billion, then killed in 49 days, then sold to LG for an undisclosed sum.

The invoices for the value destroyed in this process have not been paid. The $53 billion that the market assigned to Palm in March 2000 did not resolve to $53 billion of value. It resolved to webOS on a television. It resolved to a fire sale at $99. It resolved to a museum exhibit containing a block of wood and a carved chopstick.

HP’s patents from the Palm acquisition were retained by HP. LG got the operating system. Nobody got the $53 billion.

The Complete Record

Entity	Contribution	Outcome
Jeff Hawkins	Wooden block, Graffiti, PalmPilot	Left Palm, built smartphones, came back
Graffiti	Near-100% handwriting recognition through constraint	$22.5 million settlement, replaced with worse system
PalmPilot 1000	Fastest-selling CE device in history at the time	Generic trademark within 2 years
Palm IPO	$53 billion market cap, more than GM	97.7% value destruction over 10 years
Handspring Treo	First smartphone	Went broke, absorbed back into Palm
webOS	Most elegant mobile OS of its era	Recommends Netflix shows on LG televisions
HP TouchPad	Led non-Apple tablet sales in US, 2011	Existed for 49 days

The Verdict

The wooden block at the Computer History Museum is the correct memorial. Not because it represents a product. Because it represents a method. Hawkins did not ask whether a device of that size was feasible. He carried the size in his pocket and observed whether it was useful. This is not a prototype. This is not a mock-up. It is a pretotype — a commitment to experiencing the constraint before solving it.

This method produced Graffiti, which was technically correct. It produced the PalmPilot, which sold a million units in eighteen months. It produced webOS, which was the most elegant mobile operating system of its generation. The method was correct at every stage. The corporate structure that surrounded the method was not correct at any stage.

This is not a rare pattern. This is a standard pattern. Technology companies are not destroyed by technical failure. They are destroyed by ownership chains, by management decisions made for reasons unrelated to the technology, by market timing, and by the occasional 14-month window in which a company acquires you, launches your product, and discontinues you before a fiscal year ends.

The Supreme Leader offers the following summary, which is technically accurate and free of charge:

Jeff Hawkins carried a block of wood in his shirt pocket and derived the correct form factor for a pocket computer. Graffiti solved the handwriting recognition problem by redefining the problem. The PalmPilot sold a million units in eighteen months. The founders left, built the smartphone, went broke, and returned. HP paid $1.2 billion for webOS and cancelled it in 49 days. webOS now recommends Netflix shows on your television.

The block of wood is at the Computer History Museum. The chopstick, presumably, is still carved.

The invoices have not been paid.

— Kim Jong Rails, Supreme Leader of the Republic of Derails