Original URL: https://www.theregister.com/2013/09/17/20_years_of_the_apple_newton/

Stylus counsel: The rise and fall of the Apple Newton MessagePad

20 years on, the story of the handheld that might have changed the world

By Tony Smith

Posted in Personal Tech, 17th September 2013 09:04 GMT

Archaeologic It will forever be remembered as the butt of a-thousand-and-one jokes about its poor handwriting recognition, but Apple’s MessagePad was bold in its conception. Its legacy is ARM’s conquest of the mobile microprocessor world.

The company said on 8 August 1993:

The Newton MessagePad is the first in a family of communications assistants from Apple. By combining Newton Intelligence technology with sophisticated communications capabilities, the Newton MessagePads help you stay in touch with friends and colleagues, organize your life, and keep track of your ideas.

Apple Newton MessagePad

Handheld assistant to hold your hand: Apple’s original MessagePad

You can take notes. Make a quick sketch. Format and print letters. Share and synchronize information with your personal computer. Send a fax. Receive pages and messages. Tap into on-line services or electronic mail. Even exchange business cards with a colleague via built-in infrared technology.

And wherever you go, the powerful, under-one-pound personal digital assistant goes too, tucked in your pocket or briefcase.

The MessagePad was priced at $699, and the first 5,000 devices sold out within hours of being made available to buy. Some 50,000 went in the first 10 weeks. It’s easy to forget now, but the arrival of the MessagePad, rushed though it may well have been, was a topic of real enthusiasm among not only Mac fans but the broader tech community too.

At the time, notebook computers were chunky, weighty devices priced well beyond the wallets of most users. Palmtops were small and cheap but lacked sophistication and power. Might the much less expensive MessagePad at long last open up the world of mobile computing?

Apple began work on what would become the MessagePad back in the late 1980s. In 1987, Steve Sakoman, an Apple engineer, decided it would be a good idea to make a device capable of interpreting its user’s handwriting. Indeed, the user would interact with the gadget entirely with a pen, not a keyboard. The device might share information wirelessly.

Aye, a pad

Sakoman had a well-established interest in mobile computing. He’d been poached by Steve Jobs from HP three years previously on the back of his work developing the HP Portable, though at Apple he’d spent his time overseeing the development of desktop Macs: the Plus, the SE and the II.

With his attention back on a mobile device, Sakoman discussed his "writing reader" notion with former Lotus CEO Mitch Kapor, who proved very receptive to the idea. His ideas validated, Sakoman fleshed out a development project proposal and went knocking on the door of Apple’s engineering chief Jean-Louis Gassée, who was also Sakoman’s boss. Gassée was also impressed with Sakoman’s thinking.

The three men – Kapor’s ex-Lotus colleague Jerry Kaplan made a fourth – met to discuss how they might take the project further. The involvement of Kapor and Kaplan, neither of whom were Apple employees, shows there was a clear belief that the way forward was to develop Sakoman’s device outside of Apple. Gassée seems to have been particularly keen on this approach. But Sakoman was unwilling to make the break.

Apple's Newtons

Apple’s Newtons: (top) the eMate 300, (bottom) MessagePads original, 100, 110, 120, 130 and 2000
Source: moparx

Gassée's solution was to grant Sakoman the opportunity to pursue the project within Apple, but well away from the efforts of the company’s main hardware teams, and without much management oversight. Perhaps his cheeky notion was that if it couldn’t be done, Apple would have covered the cost of the evaluation, but if the project proved to be a runner there might still be a chance to establish a new firm to take it to market.

Sakoman's goal was to create his pen-based communication and information-organising device and evolve it to the point where it could be commercialised. He reckoned this process might take up to three years, and might ship for perhaps $2,500. He codenamed the gadget "Newton" after the scientist’s appearance in the first Apple logo. As the early work progressed, he pulled in Apple engineering talent to help, among them Glenn Adler and Eric Gruenberg who had both worked with Sakoman at HP and had followed him to Apple. Steve Capps, a member of the team that developed the first Macintosh, came on board too.

Together they and others worked on the A5 notepad-sized hardware Sakoman had originally envisaged, and which he was now calling Newton. They also started writing an operating system to manage the hardware, software to interpret the user’s scribbles, and a toolbox to support applications running on top of the OS. Apple had acquired a software company called Arus and got some rudimentary handwriting analysis code. Sakoman’s team took that as their starting point, but were soon gifted with algorithms handed to Apple by a band of Russian programmers working behind the Iron Curtain on handwriting recognition.

Russian intrigue

Apple senior VP Al Eisenstat was unexpectedly handed a floppy disk by a nervous Russian during a 1987 trip to Moscow. Back home, he passed the code on to Gassée who passed it on to Sakoman. The disk contained cursive handwriting recognition software.

That’s one version of the tale. Another, more prosaic telling has Russian software boffin Stepan Pachikov, founder of Moscow-based software house ParaGraph International, much later hawking his PC-based handwriting recognition code around Silicon Valley. Apple’s Larry Tesler, who took over the Newton project in 1990, and Steve Capps, another Apple engineer, both agreed to watch a demo, and ended up licensing the ParaGraph software for Newton.

But that came later. During the late 1980s, as Steve Sakoman’s three-year deadline began to draw close, Gassée began to believe they had all bitten off more than they could chew. In any case, the engineering chief’s own star was no longer in the ascendant. His "no compromise" attitude to product engineering at the expense of such – to him – petty matters as release schedules and usability didn’t go down well with Sculley. In January 1990, Apple CEO John Sculley named Michael Spindler Chief Operating Officer, knocking Gassée down a rung on the corporate ladder. On 2 March 1990, Gassée gave his notice and said he’d leave the company on 30 September.

Apple Newton ad

Apple advertises the MessagePad
Click for larger image

This management reshuffle didn’t go down at all well with Sakoman, and that same month he quit Apple too. Later that year, he hooked up with Gassée, and the two went on to co-found Be to develop a new desktop computer and a multi-processing, multi-tasking OS to go with it. Sakoman went to work for Palm Computing spin-off PalmSource when it acquired Be and, in 2003, briefly rejoined Apple. His LinkedIn page now lists him as a computer software consultant.

Back at Apple, Sculley began taking a closer interest in the Newton project. He saw a product that was very technologically advanced which would enhance not only Apple’s reputation but his own. It was no pet project of Steve Jobs, as the Mac had been. It was a product for which Sculley himself could claim ownership and establish his credentials as a forward-thinking head of a technology company. It reminded him of a concept he’d described in his 1987 book Odyssey: the "Knowledge Navigator", a kind of hardware information browser. The erstwhile Pepsi chief would be viewed as something more than a clever seller of sugary drinks.

Of course, Sculley was also aware of his limitations as an engineer. A visionary he might think himself, but he knew he wasn’t a product designer. So he put Newton in the hands of Larry Tesler, an engineer who had left Xerox’s prestigious PARC facility in 1980 to join Apple and by this stage was running the company’s internal think-tank, the Advanced Technology Group.

The $8,000 tablet

Tesler found the Newton team, now grown to include 30 engineers, to be imbued with a strong sense that they could indeed deliver the product. Unfortunately, as adherents of Gassée’s philosophy that good engineering was all that mattered, and without having to report to managers eager to ship product, they were preparing to deliver a device that had swollen to become a 22 x 28cm product codenamed "Figaro" that would cost buyers up to $8,000.

The money went on batteries that could run for weeks, a touch-sensitive active-matrix LCD screen, infra-red wireless networking, a hard disk for storage, and multiple microprocessors. It was impressive. But it also weighed 3.6kg. Even Tesler, a man with perhaps more of a lofty, academic view of computing technology than the hard "yes but will it sell?" attitude of the salesperson, knew the Newton prototype was never going to succeed in the market in its current form. He told the team: make it smaller and cheaper. And get it done by April 1992, added Sculley.

Had Tesler left it at that, the MessagePad would have ended up more like a chunky iPad than a beefy iPhone. It would have retailed for around $1,500. But Michael Tchao, the marketing man Sculley had sent to Tesler toward the end of 1990 to determine how Newton would eventually be brought to market, pushed to get the form-factor down to the jotting pad size Steve Sakoman had originally – and coincidentally – envisaged. Tchao wanted the Newton to be a handheld product, not one you took out of a bag and placed on a desk. Having worked on 1989’s ill-fated Mac Portable, Tchao knew big was not always better.

Apple Junior prototype

Newton - MessagePad 120? - in prototype
Source: Grant Hutchinson

Tesler, taking his cue from the Newton team, told Tchao it couldn’t be done, at least not with current technology. But not all the Newton team were in accord. Steve Capps, for one, was as keen as Tchao to make a truly handheld device. And, having co-developed the Mac in the shadow of the Lisa, Capps suggested creating a second, lesser, sub-$1,000 Newton dubbed "PocketNewt" while everyone else on the Newton team was finishing off the tablet, then called "Newton Plus".

In fact, a third effort was soon underway, this one undertaken by System 7 software contributor Paul Mercer and conducted entirely separately from the Newton project. In partnership with a team of four other engineers, Mercer spent 1991 designing and building what he codenamed Swatch. It was essentially a pocket Mac running Mac OS but with Newton’s handwriting recognition built in. It worked, it would have cost less to buy than even Capps' cut-down Newton, just $400, but John Sculley killed it early in 1992.

During the latter quarter of 1991, the CEO had persuaded Sharp to co-manufacture Newton. Mercer had separately managed to interest Sony in coming on board to produce Swatch. That was perhaps appropriate: Mercer had originally been inspired by a Sony palmtop prototype. But Sculley didn’t think Apple could partner up with two rival Japanese giants. Larry Tesler was opposed to it because it wasn’t a Newton. So even though Newton still wouldn’t be ready for at least six months, Sculley chose to cancel Swatch, which was essentially ready to go there and then.

From pocket Mac to Pocket Newt

Sculley might have gone with Mercer’s machine had he not just given a keynote at the 1992 Consumer Electronics Show (CES) promising that Apple would ship a new category of device which he coined a “personal digital assistant” - the first use of a term that would become widely used through the 1990s and beyond.

It would be a gadget capable of allowing the user to communicate through email and handle all their personal information: diary appointments, to-do lists, their address books and such, all the while smartly spotting connections between them. Lunch with dad on the fifth? Sculley’s PDA would automatically load dad’s contact details so they were there if you needed to email or call and cancel. It would learn that you went swimming every Wednesday at 6pm and remind you accordingly.

The PDA business would be worth a then-staggering $3.5 trillion by 2003, Sculley predicted. Even today, that’s a colossal figure. Smartphone sales are in the billions, not trillions, even if you add app sales and service fees. Voice calling aside, what Sculley was predicting was indeed the smartphone market, had he but known it.

But it would take companies like IBM, Symbian, Nokia and Ericsson many years to create the first true smartphones. Even without the phone component, Newton just wasn’t going to come close to the vision Sculley outlined in his speech.

Apple MessagePad 130 ad

Promoting portability: the MessagePad 130
Click for larger image

Still, the potential Sculley had outlined was enough to pique the curiosity of US telecoms giant AT&T. Its bosses could see, if Sculley and his engineers perhaps hadn’t, that a Newton-like device with access to the phone network – and by this point in time analogue mobile phones were widely available, and digital ones not very far off – could be big business in the years to come.

AT&T put out feelers, found Sculley receptive and, by early 1993, the two companies were meeting to discuss turning Apple into an AT&T subsidiary. They came close to an agreement, but AT&T’s earlier, unsuccessful acquisition of NCR and its already in play purchase of a mobile comms company, McCaw Cellular, gave AT&T bosses’ cold feet, and in April 1993, the Apple buyout talks were off.

And soon so was Sculley, forced to step down from the CEO job in June that year in part because of the delays getting Newton to market, though he stayed on as chairman until October when he resigned and left the company.

Back in 1990, as Larry Tesler was taking over the Newton project, that seemed a very unlikely outcome. And no one would have bet that a scheme set in motion by Tesler would eventually yield the world’s most widespread microprocessor platform, but that’s what happened.

Advancing ARM

Casting around for a chip that could sit inside a battery-powered handheld device and render a (for the time) high-resolution display without hammering the battery, Tesler got wind of Britain’s Acorn and its Acorn Risc Machine chip.

The ARM had been developed to power Acorn’s computers when the 6502 processor – which it had used in the Atom, BBC and Electron micros – ran out of steam. Acorn used the ARM in its desktop Archimedes computers, but its very low power draw while running and near zero power consumption when idling struck Tesler as just what the Newton needed. It was certainly better than the in-development AT&T part, codenamed "Hobbit", the Newton team had been considering thus far.

Conversely, Tesler’s Newton project must have seemed to Acorn co-founder Hermann Hauser as a dream come true: a realisation of an electronic book concept he’d been talking about since the late 1970s. Here was Apple building what was not so very far from the device he’d envisaged. Best of all, it could be powered by as Acorn chip.

The result was the creation, in November 1990, of an Acorn spin-off company, Advanced Risc Machines (ARM), part owned by Apple, part by Acorn’s majority shareholder Italian manufacturer Olivetti, and part by Hauser himself. Chipmaker VLSI had a stake too. Acorn handed over its ARM intellectual property.

Apple Junior prototype

Junior cased, complete with dropped-before-launch screen cover
Source: Grant Hutchinson

Separated from Acorn, ARM would survive its parent’s collapse. Established with a plan to design and license processors for others to manufacture, ARM had the strategy that would encourage it to sign up partners who would put its chip architecture into, eventually, millions upon millions of mobile devices.

It’s fair to say that the Newton project made ARM the powerhouse it is today. Apple initially put in $2.5m in return for 43 per cent of ARM. In March 1998, it sold 18.9 per cent of a reduced stake and made $23.4m. Further sales would net it $792m in total, far more than the $100m John Sculley once said Apple had spent on the Newton project.

Apple had found a source of silicon: a 20MHz ARM 610 chip. The first product to use it would be Steve Capps’ Pocket Newt. In February 1991, Capps’ sponsor, Michael Tchao, had buttonholed Apple CEO Sculley on a flight to Tokyo. He made the case for canning Newton Plus - too big, too expensive, technically brilliant but just not sexy enough - and focusing the attention of the Newton team on the handheld. Sculley was persuaded: Pocket Newt would be the device Apple you take to market. But development of the slate would continue, to form the basis for a follow-up product. It became known as "Senior" to Pocket Newt’s "Junior".

So, in May 1992 at the Summer CES in Chicago, when the press got to see the Newton prototype for the first time, it was Junior they saw, not Senior. It wasn’t ready, of course. That summer, the Newton team became part of a new Apple division, Personal Interactive Electronics (PIE) under ex-Philips consumer electronics chief, Gaston Bastiaens. He was able to show a working Newton at the Winter 1993 CES, but wisely kept quiet about when the thing might actually be released. In fact, he had a new deadline: 29 July. To make it, that March he told the world he had frozen the Newton software. From now on the team would be tracking down and squashing bugs.

Launching MessagePad

Within Junior, the ARM 610 chip was given 640KB of RAM to work with and connected to the 4MB of ROM that would hold the Newton OS, to be branded "Newton Intelligence" in the marketing material, along with the NewtonScript API and the bundled applications, all of which would appear on a 240 x 336 monochrome LCD panel.

The whole thing was powered by four AAA batteries. There was a PCMCIA card slot for expansion, an eight-pin mini-DIN serial connector for peripherals, and an infrared port, installed on the insistence of Sharp. The new device weighed just 410g. The case was designed by a young Apple employee called Jony Ive, who joined Apple in 1992 at the behest of company design chief Robert Brunner.

NewtonScript was developed when the original choice for the Newton’s native programming language - codenamed "Ralph" and later called Dylan, short for "Dynamic Language" - had to be dropped when it became clear it would not be ready in time and in any case would not have fitted into Junior's reduced memory footprint.

Dylan had been designed with Senior in mind, not its diminutive sibling. Like Dylan, NewtonScript was an object-oriented language. Newton 1.0’s bundled apps – Book Reader, Calculator, Clock, Dates, Formulas, Names, Notes and Works – were all written in NewtonScript.

Sharp ExpertPad ad

The other original Newton: Sharp’s ExpertPad

It’s not clear when the device became the MessagePad and not simply Newton, a name retained for the OS. During the beta-test period, the MessagePad was actually called NotePad, but the marketing team made the switch to emphasise the device’s communications abilities over its role as a note-taker.

By the time of the device’s launch on 3 August 1993 at the Boston Macworld Expo, Newton was the platform, MessagePad the product. Perhaps it was thought that, in time, Senior might be yet be brought to market as a second Newton product even though its development had come to an end in the summer of 1992. Apple certainly said the MessagePad was “first in a family of communications assistants from Apple”. Indeed, Apple had already commissioned a series of mock-ups of other devices that might be based on the Newton platform, from larger devices to videophones, and versions for sports fans and for kids.

Sharp announced its own version of the MessagePad, the Expert Pad PI-7000, on the same day. In Sharp, Newton had its first licensee – one of what would soon be many, Apple hoped. Sharp’s first ExpertPad was announced right alongside the MessagePad. In truth others did embrace the platform. Motorola built a Newton-based communicator device with an integrated cellular modem and released it in 1995. So did the less well-known Digital Ocean. In 1996, Siemens put Newton into a desktop phone-cum-information manager. The following year, oilfield and business tech company Schlumberger based a mobile medical data gathering unit on the Apple platform.

Apple’s own Newton efforts first saw the release, in October 1993, of Newton OS 1.10, which was packed with bug fixes. It improved the handwriting recognition code too, which in truth was never quite as bad as the critics and comedians had made out. Ironically, the real flaw was not the recognition code but a memory-handling bug which caused the recogniser to lose data unless the MessagePad was frequently rebooted.

Pilot flies high

Newton was programmed to adapt to your handwriting style and that took time and training. The memory bug meant it took even longer, and a limited word-recognition - rather the letter recognition - database of just 10,000 words didn’t help either. Unfortunately, too many users got fed up with its early, often incorrect efforts and gave up using it – Newton never got the training it needed to improve.

Even Apple got fed up, and Newton OS 2.0, released in 1997, incorporated a new printed-character-recogniser that was developed in-house. Undoubtedly the barbs that followed the release of the original MessagePad prompted Apple’s Advanced Technology Group to create the new recogniser - the work was done by Brandyn Web, Michael Kaplan and Larry Yaeger - but the arrival of the pocket-sized Palm Pilot may have pushed it too.

Palm Computing’s Jeff Hawkins had been working on software to simplify entering text into a computer by pen since the early 1990s. Palm was founded in 1992, and the following year Hawkins started to develop the company’s first hardware, Zoomer, based on the GeOS operating system. It was announced in June 1993 but not released until November, when it was perceived as a Newton knock-off and, despite the involvement of Casio and Tandy, was a flop.

Palm Pilot

Jeff Hawkins’ Newton killer: Palm Pilot

Not so Hawkins’ next effort, the Pilot 1000. By the time of its release, in 1996, Palm had been bought by modem company US Robotics, and Hawkins had developed Graffiti, a printed character recognition system that used a pre-designed set of glyphs to represent letters, numbers and symbols. It also had dedicated character-entry areas. Both made Graffiti’s character recognition much more straightforward than Newton’s and made it far more accurate.

US Robotics also pitched the Pilot, which was small and pocket friendly, as an adjunct to a desktop PC. You could enter data on your computer and then sync it over to the Pilot, which became simply a portable data viewer. The Pilot wasn’t as sophisticated as the MessagePad, but it worked and it was (relatively) cheap: $299. US Robotics sold plenty of them.

Steve Capps had always wanted Junior to be as small and as pocketable as the Palm Pilot later turned out to be, but in 1991 couldn’t cram Newton into such a compact case. By 1995, it could be, and Apple engineers build a small Newton prototype called the MessagePad 110/LC. It combined all the MessagePad’s chips into a single part devised by ARM and LSI Logic. Had production been authorised, it might well have beaten the Palm to market.

Evolving the MessagePad

As it was Apple stuck with the larger form-factor established by the original MessagePad. That device was replaced by the $499 MessagePad 100 in March 1994. The 100 was identical to its predecessor but for the more up-to-date Rom it contained. Alongside it debuted the $599 MessagePad 110, an even larger device – to accommodate longer-running AA batteries – with more 1MB of memory rather than 640KB and a 240 x 320 screen, slightly larger than the original one.

The following October, Apple introduced the MessagePad 120, which upped the memory to 2MB – a cheaper version had 1MB – but, more importantly, featured Newton OS 2.0 with its better character recognition. March 1996 saw the arrival of the MessagePad 130, a 120 with a backlit screen, 8MB of ROM and 2.5MB of RAM.

A year later, Apple introduced the MessagePad 2000, which featured a bigger, 320 x 480 greyscale display, 5MB of memory – 1MB of RAM and 4MB of Flash – a second PCMCIA slot and a new peripheral connector, the InterConnect port. It was bigger even than the 110, and was the most expensive MessagePad yet: it retailed for $799. It was “what the Newton should have been in the first place”, CEO Gil Amelio would say after his ousting from Apple, but it wasn’t the star of the launch.

Apple MessagePad 2000 box

UMPC: The MessagePad 2000
Source: Jeff Kubina

The eMate 300 outshone the 2000. This was a Newton laptop, a 2.2kg clamshell unit with the same screen as 2000 but arranged in landscape orientation above a physical keyboard. The 2000 ran on a 162MHz StrongARM 110 chip, but the education-oriented eMate had to make do with a 25MHz ARM 710a.

It was “the first of a new class of affordable mobile computer”, as Apple put it, and it’s arguably the first true netbook – particularly if equipped with the optional PCMCIA modem Apple offered. Never a serious productivity machine, it was nonetheless a great writer’s portable, able to upload work to a desktop Mac or PC.

The 300 shipped early in 1997, but its arrival was overshadowed by Steve Jobs’ triumphant return to Apple as “advisor” to then CEO Gil Amelio. The following July, Amelio was out and soon Jobs was back in control, albeit as “interim CEO”. He would eventually drop the "interim" but from July 1997 he ran the company right up until January 2011, the start of his final leave of absence on medical grounds. He died the following October.

The death of Newton

Newton died – as an Apple product, at least – on 27 February 1998, the result of Jobs’ aggressive pruning following his return to Apple’s helm. Apple released the MessagePad 2100 in November 1997, but everyone expected Jobs to axe Newton sooner or later. Unlike the Mac, they said, Newton had not been developed under Jobs’ watch.

That may well have reinforced Jobs’ thinking, but killing Newton was first a pragmatic decision. Apple had failed to license the platform far and wide as Sculley wanted to, though he’d been ousted before such an effort could really begin.

His successor, Michael Spindler, didn’t pursue it aggressively - he was keener to license the Mac OS instead - and considered shutting the Newton operation down on the recommendation of McKinsey, a consultancy.

After Spindler, Gil Amelio, who initially considered ditching Newton - which was costing the company $15 million a quarter, he later said - too but rejected the notion, attempted to make more of the platform, sensing its value as "not Mac", and that led to both the eMate and a plan to spin off Newton as a separate company, Newton, Inc. It almost happened, but Jobs pulled it back, perhaps feeling that a company in need of major revivification didn’t need the distraction of a spin-off process.

He soon realised that without major endeavour Newton wasn’t going to become the widely supported platform Sculley had dreamed of, and in its current state Apple couldn’t afford to do that – not if it wanted to revamp the Mac OS as Mac OS X. The Mac was the reason why Jobs’ company NeXT and its NeXTStep OS had been acquired by Apple in December 1996.

And even with more money: could Newton had cut it in the face of competition from Microsoft’s Windows CE, the precursor to Windows Mobile; the Palm OS; and the emerging Symbian platform? Probably not.

Apple eMate 300

The eMate 300: designed for kids, loved by writers on the move
Source: raneko

Had Jobs kept Newton, perhaps the OS would have matured and, when the CEO eventually turned to the mobile market, found a home in the iPhone, as many Newton fans hoped when rumours of Apple tablets and such began to leak out in the mid-2000s. I think basing the iPhone on Newton OS (as it was) would have prevented Apple enjoying the success it did. Newton was too firmly rooted in the old world of pen computing, as were Windows Mobile and even Symbian, which is one reason why iOS was able to leap ahead.

iOS is based on Mac OS X, and thus came with a toolchain familiar to all Mac developers, not just the small number who’d also embraced the Newton OS.

But Newton has its place as one of the great "might-have-beens" of Apple and, indeed, mobile technology history. It still has its adherents, and for the early 1990s – despite the many, many gags – was impressive technology.

But it was too early. Even had Apple integrated cellular communications – as Motorola did with with Newton hardware, and as IBM showed in November 1993 with the BellSouth Simon, though that wasn’t based on Newton – it would have raised the price beyond the point where most businesses, let alone individuals, could justify the expenditure. ®