A History of Palm, Part 1: Before the PalmPilot

Palm Computing was largely the creation and vision of one man, Jeff Hawkins. Palm first brought tablet computing to consumers in the form of PDAs (but was beaten by Apple and its scions). The later – and more momentous – goal was to bring consumers to PDAs through simple and very fast user interfaces. This second goal brought us the original Pilot and an entirely new form-factor that millions embraced.

Palm Pilot and TreoIt was only until the introduction of multimedia-rich smartphones that Palm stumbled, though it was one of the leading manufacturers.

PalmPrint

GRID logoHawkins was the ideas guy at GRiD Systems, which had been founded in January 1979. GRiD had pioneered the concept of ruggedized laptops, the GRiD Compass (released April 1982), and object oriented programming environments.1

Hawkins, who avidly researched the brain before he arrived at GRiD (though he had no advanced degrees at this time), had developed theories about how the brain recognized text. He developed these ideas into his first pen-computing product, PalmPrint.

PalmPrint worked differently from other handwriting recognition software developed by companies like ParaGraph and Microsoft. Instead of comparing entire words to a dictionary of commonly used words, PalmPrint compared each character to a master set of four or five different forms of the character. The result was much less resource intensive, and, according to later Palm surveys, more accurate than the other systems.2

For a fascinating overview of the history of character recognition, see Palm:I’m Ready to Wallow Now by Thom Holwerda on osNews.

Tandy Corporation, the parent of Radio Shack and a pioneer in personal computing, bought GRiD Systems in March 1988.

GRiDpad tablet computerPalmPrint provided the impetus for GRiD to start developing ruggedized tablet PCs, GRiDpads (released in 1989). The GRiDpad measured 9″ x 12″ x 1.4″ and weighed 4.5 lb.

There was a lot of demand for such devices from enterprise and government customers who needed access to databases on the go. A laptop would be too cumbersome to use outdoors away from a desk (or a lap), so these businesses were quite receptive. GRiD quickly became the market leader in tablet devices, partly because of PalmPrint and partly because of GRiD’s easy to use development environment (originally on a GRiD-developed operating system, but later ported to Windows for Pen).

Hawkins started daydreaming about bringing PalmPrint and pen-computing to consumers. Unfortunately for him, GRiD was not the company to do it. GRiD had no consumer presence. It didn’t even sell its products to individuals. Instead, it had a fleet of salespeople who targeted high-profit customers who would buy dozens or hundreds of GRiD products at a time.3

The Zoomer

The specification that Hawkins showed to Tandy CEO John Roach (Hawkins’ manager at GRiD prior to Tandy’s acquisition) and other possible investors and partners was for a relatively simple device called the Zoomer (from consumer). The Zoomer would be no bigger than a paperback book (quite a bit smaller than the GRiD tablets, the smallest of which were the size of an A4 sheet of paper) and would have a large screen and no keyboard. Text entry would be done entirely through an improved PalmPrint, which Hawkins owned the rights to.

When the Zoomer shipped in 1992, it had a 4.2″ x 6.8″ footprint and was 1″ thick. It weighed in at 16 oz. It was much smaller and lighter than the GRiDpad.

Though the Zoomer could serve as an independent PC, it would be used most as an organizer, so the PIM (personal information manager) applications were imperative. This is what users would work with most, and they would have to offer a clear advantage to the paper DayPlanner-type organizers that executives already carried around.

Hawkins reasoned that synchronization with the PC and superior organization features (categories, search, etc.) would make the Zoomer viable.

Launching Palm Computing, Inc.

Palm logoRoach tried to convince Hawkins to stick around at Tandy to develop a Tandy-product. Hawkins saw things differently. He wouldn’t have the independence he needed if he was in the midst of a massive corporate parent, especially one for whom his product was not its top priority.4 Hawkins created a new company to develop his idea further; Palm Computing, Inc., opened its doors on January 2, 1992.

At that time, his primary goal for the company was to release software for a handheld computer. However, in a concession to his former employer, Palm would not develop the hardware. Roach had made a $300,000 investment in Palm and would market the Zoomer under the Tandy brand. Casio (one of Tandy’s largest OEM’s) would manufacture the devices and sell some of them under the Casio brand.

Tandy Zoomer PDAPalm would not even develop the operating system. That task was delegated to GeoWorks, which owned GEOS, a graphical operating system that was already growing long in the tooth. Palm’s role in the Zoomer project was confined to developing most of the application software, including a PIM and handwriting recognition.5

Despite all this baggage (with four partners, Hawkins feared that every design decision would be questioned), he was optimistic. He had easily secured venture capital funding for his startup through his contacts with established investors. Merrill, Pickard, Anderson & Eyre, a legendary venture capital firm, had given the small company – Hawkins had hired two software engineers, Art Lamb and Ain McKendrick – two offices to use temporarily.

Work began immediately on the Zoomer’s applications, but progress was hampered by GeoWorks. The company hadn’t even developed the operating system for the low-power Intel processor used in the Zoomer, let alone the development packages Palm needed to create the Zoomer applications. As a result, much of the software that Lamb and McKendrick worked on would never end up on a device. Instead, the two software engineers experimented with GEOS and with interface design, one of Hawkins’ passions.6

There was one thing at Palm that Hawkins did not enjoy – being a manager. He detested having to set up predefined compensation packages and dividing the potentially valuable stock options himself. Moreover, he was not much of an accountant (though he was very frugal with his investors’ money) and did not understand business or accounting law.

His most important early hire – perhaps his most important while at Palm or Handspring – was Donna Dubinsky, who had left Apple’s software subsidiary, Claris. While she was there, she had worked for Bill Campbell (later CEO of Intuit) and had developed a massive network of business contacts that would work to Palm’s advantage when they needed favorable analyst reports.7, 8 Around the same time, Hawkins also hired Ed Colligan to serve as VP of marketing.9

Bogged Down

Unsurprisingly, development of the Zoomer got bogged down, partly because of GeoWork’s slowness to get the Zoomer version of GEOS completed, but also because of feature creep instigated due to design by committee.10

According to David Pogue, in his book Piloting Palm, Casio was a particularly difficult partner to work with. Their relative inexperience with software and hardware development (the company’s major portable products were digital wristwatches, calculators, and inexpensive pocket organizers) made them irrationally intolerant of any bugs (at least in the eyes of Palm’s engineers), no matter how minor or how unlikely to affect the user. Casio also expected too much from the hardware. The company demanded instantaneous startup and 100 hour battery life, two features that are mutually exclusive in such a small device.

Hawkins’ concept of a personal pen-computer was slowly diluted by feature creep. The GEOS operating system had been first released in 1985, and the version that was being adapted for the Zoomer had last been updated in 1990. Though GeoWorks was revamping the user interface for a PDA, GEOS was still far from simple.

Later, reviewers complained of the desktop-like interface with widgets too small to be easily manipulated with a pen, the multiple steps required to copy and paste text, and that GeoWorks had not implemented a universal clipboard (ten years after the Macintosh had introduced the feature – eons in technology time).11 GeoWorks was also slow to get Palm the development tools the company needed to start serious work on the PIM applications that would be bundled with the Zoomer.

Work on PalmPrint was also bogged down. ParaGraph’s software, the company run by Soviet immigrants that had licensed its handwriting recognition software to Apple, was adaptive. As a user used the Newton, it would learn his handwriting and recognition would improve over time. On top of that, ParaGraph included support for cursive and block letters, as opposed to PalmPrint, which only supported block letters.12 The lack of “Newton Intelligence” (the catch phrase Apple used to describe its handwriting recognition and language parsing – and later the operating system itself) was not a fatal flaw, but it meant that there were only a few different ways of making letters. Since Hawkins had developed the early versions himself, PalmPrint preferred his handwriting. In Piloting Palm, we learn that software engineers had actually started writing like Hawkins in order to get better recognition.

As the Zoomer was readied for release, Hawkins became nervous about the product’s success. He had envisioned an organizer, but the Zoomer had turned into a personal computer complete with Pocket Quicken, AOL, a Zodiac database, and a 100,000 word dictionary. In his eyes, these utilities were an example of feature creep. It was unlikely that any user would buy the Zoomer because of a Zodiac database, but it was another feature that Casio and Tandy could put on their specifications.13

Though Hawkins had largely lost control over product development, the price stayed low. Tandy and Casio would debut the Zoomer for $599, over $100 less than the Newton MessagePad, so it was still competitive with the high end Sharp Wizards and Casio BOSS’s that the Zoomer would have to compete with.

Press Reception

Hawkins’ apprehensions about the Zoomer launch were well founded. The press had become somewhat enamored with the relatively unambitious Zoomer after years of being teased with a vaporous and quite ambitious Newton. When the Newton was released and was found to have deeply flawed handwriting recognition software, the press was all the more excited over the simple PalmPrint.14 This was also partly a result of Apple’s extravagant spending on its carefully choreographed press events for what was widely perceived to be vaporware (a la Pink, Senior, and the many other products that Apple had touted well before they were completed) that were more like rock concerts than like product demonstrations.

When the Zoomer went to conferences, the partners used Tandy’s well worn booth and usually rented conference rooms to give reporters hands-on time with the Zoomer. Casio once splurged on one such event and bought all the presenters baseball caps with an embroidered Zoomer on the front (for a description of Newton’s lavish press events, see Defying Gravity: The Making of Newton, the official history of the development of the Newton.)

Despite the paltry PR budget, the Zoomer got a lot of favorable buzz from the press. Many reporters and analysts were becoming increasingly frustrated by the lack of results with the Newton.15 The Zoomer was much less ambitious and seemed more feasible. Newton Intelligence (perhaps the name was another sign of hubris) was a fully object oriented operating system, replete with a relational database that stored all of the data for the device. Newton Intelligence went further than just an innovative programming interface and data storage system; it was able to parse sentences into commands.16

The less ambitious (and less innovative) Zoomer worked like a PC, complete with a file browser with icons. At the same time, it was not just a PC with a touchscreen bolted on. PalmPrint allowed for ‘inking’ (unrecognized handwriting) in all fields, allowing the user to jot down information quickly.17

After its release in August 1993, the Zoomer received somewhat favorable reviews, but most of them were negative.18, 19 The device was severely compromised. It had spectacular battery life, but at the expense of a powerful processor and readable display. One reviewer said (subscription required) that the Zoomer “has about as much snap to it as soggy cereal.” Other reviewers lauded PalmPrint for being more accurate than the Newton (or the other smaller players, like Amstrad), and many liked the Palm apps. PC Week‘s headline – “Zoomer doesn’t zoom, but for some users this PDA makes sense” – was typical.

The Zoomer was not the only PDA released in 1993 to receive a tepid reception. The much ballyhooed Newton had unreliable handwriting recognition (less accurate than PalmPrint, according to a few reviewers), and all of the PDAs were slower than an ordinary DayPlanner for organization tasks. The primary strength these devices had over their print compatriots was their ability to synchronize data with users’ PCs. However, this feature was often treated as an afterthought, if at all. All manufacturers sold synchronization software and cables separately, if at all.

It was intended that the PDA would supplant the PC. In fact one manufacturer, EO, actually included VGA and serial ports so it could be used as a dedicated PC, even though its PenPoint operating system was hardly a replacement for Windows or the Macintosh as a desktop platform.

Survival

Apple sold 80,000 first generation Newtons while Tandy and Casio only sold 20,000 Zoomers.20 It appeared that the Zoomer line was finished, so Palm needed a brand new source of income. It maintained its relationship with GeoWorks and helped adapt its applications and PalmPrint to PDAs from Sharp. Ultimately, GeoWorks went on its own, forming partnerships with HP (for which Palm would port PalmPrint, but not its PIM) and Novell.21

Palm ConnectPalm ultimately subsisted on a product called PalmConnect, though a new product would soon be under development.22 PalmConnect was a direct result of the survey cards user filled out to register their Zoomers. The most demanded feature was synchronization with their desktop PIMs. Palm complied and created PalmConnect, which synchronized the Zoomer’s databases and files with the PC.

Palm eked out an existence selling connectivity software to existing Zoomer customers and (after a rewrite) to users of the popular HP palmtops that ran MS-DOS. This income would not make Palm a billion dollar company, one of Donna’s ambitions, but it kept the company solvent.23

Hawkins began to look for ways to improve the user experience of the Zoomer. He apparently agreed with the reviewers and decided to address the two great weaknesses of the device. The first was its handwriting recognition. Since PalmPrint recognized individual characters, it had to have several different samples of each letter. This resulted in accurate recognition if the user formed his or her letters like the samples – but this wasn’t always the case. Ron Marianetti, one of the senior engineers, noted that PalmPrint best recognized handwriting if it mimicked Hawkins’ handwriting. In a stroke of inspiration, PalmPrint was adapted to recognize only a single, very simple sample of every letter. Instead of having the software struggle to adapt to humans, it would be easier to have the humans adapt to the needs of the software.24

The new package, named Graffiti by Marianetti, was a huge PR coup for Palm. It was released first for the HP OmniGo, but after Apple opened the Newton handwriting recognition API to outside developers, it debuted on the Newton to rave reviews. “If your PDA can’t understand a word you’re writing,” Windows Sources extols (subscription required), “then get Graffiti.” Another reviewer, in Macworld (subscription required), found that his “typing” speed on the Newton with Graffiti was “up to 20 words per minute at 90 to 95 percent accuracy.”

Orders flowed in, and Palm actually became profitable, finally making a return on the $1.3 million that its investors had initially invested.

Touchdown

Zoomer’s second major problem was its speed. It took 10-15 seconds to boot up and to switch between applications, seriously hampering its usefulness as a serious business tool. A DayPlanner was much faster. It was available as soon as it was opened to the correct page, a process that took only a fraction of the time the Zoomer took booting up. For a handheld to become a constant companion, it would need to move between applications almost instantaneously.

The user interface would also have to improve dramatically. With the Zoomer or Newton, users could only see from 8:00 a.m. to 3:00 p.m. on the (relatively) huge displays. A paper planner was much smaller and allowed the user to see his or her entire day. Little quirks like this and others turned off business users.

Rob Haitani devised a theory to explain how a handheld could succeed on the market. It was called the Zen of Palm (PDF). Any handheld would have to have clear advantages to paper DayPlanners and very few disadvantages. The UI would have to be impeccable. If a user had to navigate multiple dialog boxes, it wasn’t worth the money to get a PDA. A simple benchmark of the efficiency or inefficiency was to count the number of taps to create an appointment or add an entry to the address book. This required that all of the most used features be easily accessible, not buried behind menus or in dialog boxes. This concept of ease of use had eluded many of the early PDA’s.

Unfortunately, it appeared that the Zen of Palm would be impossible with another manufacturer or platform. Design by committee had not worked with the Zoomer, and the top-down design had not worked with the Newton. Even after complaints about the complexity of Newton Intelligence, Apple added more features with the 2.0 release of the software that did little to improve the user experience. Instead, the Newton gained a slight speed bump and a new communications tool. It was still painfully slow to search for a contact or add a new appointment.

Ron tentatively named this project (or concept) Touchdown, and it would save the company.

Hawkins, Dubinsky, and Colligan (with many misgivings) decided to go ahead with the project. With support from one of Palm’s major investors, Bruce Dunlevie, the board of directors was convinced. However, this was not enough. According to Colligan’s estimates, Palm would need $5 million to develop, manufacture, and promote the new product. Palm only had $3 million.

Finding a Partner

A corporate partner or buyer would have to be found. A few major candidates were considered (Motorola, Compaq, and Nokia), but none were willing to give Palm control over product development, nor would they give Palm a big enough share of sales to recoup costs. Compaq even proposed that Palm sell its new PDA via mail, relegating it to the pages of PC Magazine and Macworld instead of the shelves of big box stores across the country.

Palm’s buyer (and a secure feature for the Touchdown) was secured in a surprising way. During product development, Hawkins, Dubinsky, and Colligan were traveling the country promoting Touchdown as the platform of choice for hardware and software developers. One of the companies that the trio met with was Salt Lake City-based Megahertz, which had recently been acquired by USRobotics, which would be a good company to develop a portable modem for the Touchdown.

There were no working prototypes yet, so Hawkins was only able to present the Zen of Palm and pass around a balsa model of the Touchdown. Even with this somewhat bare presentation, Jon Zarkin (one of Megahertz’s executives) was wowed.

Palm Pilot with USRobotics logo

Palm Pilot with USRobotics logo

In 1995, Palm was acquired by USRobotics, which provided the resources for a marketing campaign targeting consumers and national distribution to retailers. Palm’s first PDA, the original PalmPilot, was released on March 24, 1996.

The Palm story continues in Part 2.

Index to A History of Palm

If you’re a Palm user, please consider joining our new Palm OS Users group on Facebook to share your knowledge and provide user-to-user support.

This article is adapted and updated from Tom Hormby’s “Early History of Palm (Part I)”, which was published on Silicon User on Oct. 1, 2007. It is published here with his permission.

Footnotes

  1. Neal Boudette. “AST, GRiD sign wide-ranging pen pact,” PC Week 10, no. 3 (January 25, 1993): 20.
  2. Pat Dillon. “The Next Small Thing: What does it take to change the world? Obsession. Tenacity. And lots of mistakes. That’s the untold story behind the PalmPilot – a 15-year saga that produced the kind of breakthrough that every startup dreams of.,” Fast Company no. 15 (May 1998): 97.
  3. Ibid.
  4. Daniel S. Morrow, “Jeff Hawkins Oral History,” Computerworld Honors Program International Archives, http://www.cwhonors.org/archives/histories/Hawkins.pdf (accessed accessed July 19, 2016).
  5. Ibid.
  6. Dillon, Fast Company, 97
  7. Morrow, Computer Work Honors Program International Archives.
  8. “Donna L. Dubinsky, MBA 1981,” Alumni Achievement Awards, https://www.alumni.hbs.edu/stories/Pages/story-bulletin.aspx?num=1997 (accessed July 19, 2016).
  9. Dillon, FastCompany, 97
  10. Ibid.
  11. Patrick Marshall. “Tandy Z-PDA Zoomer,” InfoWorld 16, no. 16 (January 25, 1993): 91-96.
  12. Edward Rawson, “Natural Handwriting Recognition,” Infotivity, http://www.infotivity.com/hwr.htm (accessed July 19, 2016).
  13. Mike Cane, “The Palm Foleo Disaster: Part Two,” Mike Cane’s Blog, http://mikecane.wordpress.com/2007/09/12/the-palm-foleo-disaster-part-two/ (accessed July 19, 2016).
  14. Michael Zimmerman. “Separate but equal PDAs from Apple, Tandy/Casio,” PC Week 10, no. 23 (June 14, 1993): 25-27.
  15. Zimmerman, PC Week, 22-27.
  16. “Newton MessagePad,” https://msu.edu/~luckie/gallery/mp100.htm (accessed July 19, 2016).
  17. Marshall, InfoWorld, 91-96.
  18. Ibid.
  19. Rich Santalesa. “PDAs; here, there and everywhere: Apple Newton MessagePad, Sharp Electronics Corp.’s PI-7000 Expert Pad, EO Inc.’s ATandT EO 440 and 880, Casio Z-7000, Tandy Z-550, AST Research GridPad 2390 and Fujitsu Personal Systems Inc.’s 325 Point PDAs) (includes related articles on remote communications and PDA keyboards,” Computer World 14, no. 3 (March 1994): 200-8.
  20. Cane.
  21. Junko Yoshida. “HP taps Geos OS for portable organizer: HP’s OmniGo 100 handheld organizer runs Geoworks’ Geos operating system,” Electronic Engineering Times no. 870 (October 16, 1995): 16.
  22. Dillon, Fast Company, 92-96.
  23. Mark Scardina. “PalmConnect: This connectivity software lets you run Windows versions of PhoneBook, Appointment Book, and NoteTaker on your desktop PC, and keep them synchronized with your HP Palmtop files,” HP Palmtop Letter, http://www.palmtoppaper.com/ptphtml/21/pt210054.htm (accessed July 19, 2016).
  24. Shawn Barnett, “Jeff Hawkins,” PenComputing, http://www.pencomputing.com/palm_orig/Pen33/hawkins2.html (accessed July 19, 2016 via archive.org).

Keywords: #palmpilot #zoomer #palmprint #jeffhawkins #palmos #gridpad

Short link: http://goo.gl/T4dfqv