Apple Newton 2100 (Pt. II)

The Newton 2100 had so many great features that it’s impossible to list them all in a simple blog post. Given that this is one of a series of posts about pen-based computers and PDAs, however, I’ll focus on a few points that individually seem relatively minor, but which between them show just how much care and attention the Apple engineers paid to the Newton platform.

The Newton was designed from the ground up to be used with a stylus, and to be used on the move. Unlike so many devices that have followed it, the Newton had no pretences at being a desktop machine. There was no attempt to shoehorn a desktop UI into a handheld device; instead the whole system was designed around the stylus, rather than simply trying to use it as a poor mouse replacement. Text and images were deleted by simply scribbling through them. Handwriting could be recognised on-the-fly, or deferred for recognition later (ideal for taking notes without the recognition slowing things up), and there was even a mode for recognising hand-drawn sketches which would transform my rough scribbles into perfect lines, squares, triangles and even circles.

Having created a modernist masterpiece (or more probably sketched a map) it was extremely simple to edit and change the design. Pressing the stylus to the screen for a couple of seconds elicited a small squeak from the speaker, and switched the Newton into selection mode. Drawing on the screen in this mode produced a thick line which could be used to circle multiple items for selection. Tracing over individual lines – or parts of lines – would also select them, so having drawn a square, for example, it was easy to select just one side of it for removal.

Once selected items could be moved by dragging, resized by dragging one corner, or deleted with a quick scribble. Dragging selected text or drawings to the side of the screen caused a smaller translucent representation to be docked there which could be dragged back into a different application – copy and paste using only the stylus. An extension for the 2100 allowed you to access a hidden OS preference to set the number of these docked items that could be held at once – up to a maximum of 20. This made it simple to copy and paste multiple different items without losing track of any of them. How I wish that this feature was implemented in a desktop system.

The attention to detail in this pen-based system even went as far as the hardware design. Not only did the 2100 have a full-sized and very comfortable stylus that clicked satisfyingly into a sprung chamber in the body of the device, but they also included a small pull-out stand to rest the stylus in when it wasn’t in use. Again, this is a small thing that’s sadly lacking from most modern pen-based systems.

Finally, still on the subject of hardware, it’s impossible to really consider any mobile device without giving some thought to its battery life. This is one area where the 2100 really did excel. Whilst its green-backlit greyscale screen might look primitive compared with more modern full-colour devices, it had one huge advantage: low power consumption. With just an hour’s charge the 2100 could nominally run for 24 hours with the backlight off. In practice the battery life was less than this, but the way in which a PDA is usually used – just a few minutes at a time – meant that a single charge could easily last a week, or even a month. Travelling overseas and worried that the charge might run out? No problem, just use the Newton’s international charger with simple slide-on adaptors for different power sockets. Or just take the battery tray along with you and rely on easy-to-obtain AA batteries instead.

What if the worst did happen, and your poor Newton ran out of power completely? Thanks to the early use of flash memory in the 2100 it wouldn’t lose any data. When you finally found another power source, your Newton would boot up with all your data intact. Which is something that definitely can’t be said of the Palm Pilot…

Apple Newton 2100 (Pt. I)

Fast forward to 1997. I’d finished university and had a real job. It was also one of those rare periods when I had no significant financial burdens. With no major outgoings, and fresh from the penny-pinching days of studenthood, I seemed to have a lot of spare cash floating around. What was a man to do, other than spend stupid amounts of money on new tech?

So I found myself traipsing off to the (not very) local Authorised Apple Reseller to have a play with a Newton 2100. This was the latest and greatest of the Newton line: a king amongst PDAs – with a royal price tag of £699. I may have had money to burn, but not even I was stupid enough to pay full price to an Apple reseller. My trip there was simply to try the 2100 out, to make sure that the upgrade of the operating system hadn’t ruined the things that I’d come to love about the MP100. I wasn’t disappointed – although the salesman at the shop definitely was.

I had a couple of friends who were also impressed by the MP100 – but not £699 worth of impressed. They decided to buy a Newton 120 each – the same basic operating system as the 2100, but with significantly less powerful hardware. The three of us pooled our money to make a single purchase with a mail-order company in order to negotiate a better deal. So it was that in December 1998 my Newton 2100 arrived. Little did I know at the time that only two months later the whole Newton line would be discontinued by Apple.

The 2100 addressed some, but not all, of my issues with the MP100:

  • Although it didn’t have a case, it did have a removeable screen cover which is still the best I’ve seen on any PDA
  • The faster processor (162MHz rather than the 20MHz of the MP100) meant that the handwriting recognition was far faster – and it also seemed to be more accurate
  • The 2100 had a new version of Newton OS – version 2.1 compared with 1.3 in the MP100. One of the things that the engineers concentrated on when developing the 2.x series was reducing the amount of handwriting that was required. I might expand on this in a later post, but suffice to say that the reduction in the amount of writing that was required was significant
  • The synchronisation tools were still quite poor. With the abandonment of the platform a couple of months later this is a situation that was never to improve
  • It was still physically large – moreso than the MP100. The functionality and power it offered made it a viable replacement for a laptop in many ways, so its size could actually be considered as quite small. However for those occasions when you just needed an address book, it was a rather hefty lump of plastic to have to carry

On the whole it was a great machine: powerful, extensible and delivered everything it promised and more. In the next post I’ll look at some of the highlights of the MP2100.

Apple Newton 100 (Pt. III)

For the price I paid, the MP100 was a great device. But looking at it from a more independent (i.e. less money conscious) viewpoint, it was clear that there were several shortcomings with it:

  1. The £40, fell-off-the-back-of-an-apple-cart, model lacked any sort of case or screen protection
  2. The handwriting recognition was slow. There was an option in the handwriting preferences to set the trade-off between accuracy and speed – but going for fast recognition only ended up taking more time overall due to the need to make corrections.
  3. It relied too heavily on the handwriting recognition
  4. The tools for synchronising it with a desktop machine left a lot to be desired
  5. It was physically large

The first issue was easily dealt with by a combination of an old plastic videotape case, some green fabric (for the outside), some soft furry fabric (for the inside), some glue and a couple of hours embroidering the Newton logo during the Eurovision Song Contest (no, really). It took a while to construct, but the result was a dirt-cheap hard case which housed the Newton quite snugly.

The second issue was more problematic. The slow handwriting recognition made the whole system seem a little sluggish. The Newton developers had clearly also seen this as an issue, and included a useful mode called “deferred recognition”, which would leave your scrawl untranslated, but allow you to trigger the recognition routines later on. This was handy for taking notes, but not much use for entering data into more structured applications, such as the address book. Of course the Newton had an on-screen “soft” keyboard (several different types, in fact), but using that was also quite slow (due to the hunt-and-peck nature of on-screen keyboards, rather than due to the speed of the machine itself). Indeed the handwriting recognition was so problematic for some people that an early version of Calligrapher was available for the Newton as a third-party application – long before the invention of the Palm Pilot, which used it as the primary method of input.

The speed of the handwriting recognition might not have been such an issue if it wasn’t for the fact that it was used everywhere throughout the operating system. It was almost as though Apple was so proud of the handwriting recognition that they felt it should be used as much as possible – even where it wasn’t necessarily the best option.

The tools for synchronising with a desktop machine were not good. Whilst the Newton’s internal applications were excellent, getting data to and from a PC or Mac was fraught with problems. It didn’t help that the Newton’s applications were just so much more flexible than just about any desktop PIM software of the time – so there was inevitably a mismatch between what the computer could do and what the PDA was capable of.

Finally, the size of the Newton was an issue – it’s just that nobody realised it at the time. It was the size of a paperback book, but far more flexible than any one book could be. But unfortunately for Apple a lot of people don’t want to carry a paperback book when all they actually need is a small address book. It took Jeff Hawkins to realise this when he created a small wooden prototype of the Palm Pilot – perhaps the device that did the most to secure the Newton’s eventual demise.

When I left university in 1997 I was still enamoured with my Newton, despite these shortcomings. So as soon as I could afford to, I decided to treat myself to the latest and greatest (and, ultimately, the last) Newton: the MP2100. Did this upgrade address any of the issues above? All will revealed in the next exciting instalment

Apple Newton 100 (Pt. II)

The term “Personal Digital Assistant”, or PDA, was originally coined specifically for Apple’s Newton. These days it’s a more common term, applied to everything from Palm devices to Windows Mobile systems. But perhaps it’s bandied about a little too easily these days, because whilst these young pretenders to the name are both “personal” (they’re small and pocketable, and typically used by only one person) and “digital” (they’re… well… digital), they’re not really “assistants”.

The Newton was most definitely an “assistant”. It even had a built in application called “Assist” which did exactly what you might expect of a personal assistant – or at least of a digital one. Scribble “Lunch with Simon on Tuesday” into the assistant, and it would schedule a lunch appointment with the most likely Simon from your address book. Scrawl “fax this to Bob” and it would find Bob’s fax number in your address book and fax him the document you were currently looking at. It may have drawn the line at beating up Martin, but otherwise the Assistant worked pretty well.

The reason why the Assistant was able to perform its magic is due to one of the underlying design decisions made by the Newton team. Rather than storing data in simple text or binary files, as most PCs do, they were stored instead in database files called “Soups”. Each soup was accessible by any application on the Newt, and it was possible for a program to access a conglomeration (or should that be consommé) of all of the soups stirred together as one big database. This allowed any application to access the data from any other application – and the databasey nature of things gave it all some structure.

The obvious result of all this gazpacho was the Assistant. It could use the super-soup to easily search through all of the data in the Newton, and because the database model provided a little structure to things, it could safely add a new appointment here, or an alarm there. It was the Google Desktop of its time – only moreso, because it wasn’t limited to just searching for data, it could also create it.

Soups are just one example of the innovations that went into the Newton. The whole operating system had been designed from the ground up to be pen-driven, and to thrive as best it could within the limitations of the Newton hardware. Clever design tricks allowed the built-in applications and widgets to be easily extended by third parties, with soups letting them seamlessly work with the data from the native applications (or from other third-party applications). The “object oriented” operating system made it possible to extend the Newton’s core code (which was stored in read-only memory) without having to use up much of the precious RAM (the early Newtons only had 640KB of RAM – and at the time ROM was cheap, but RAM was not).

Technically the Newton was a triumph. The handwriting recognition worked well enough. The operating system could be twisted, extended and expanded in ways that its designers hadn’t even imagined. It really did live up to the idea of a “Personal Digital Assistant”, and I certainly considered it £40 well spent. But not everything was perfect with the MP100, and that will be the subject of Part III

Apple Newton 100 (Pt. I)

The Apple Newton was released way back in 1993 and although I thought it looked interesting, I steered clear of it for two reasons:

  1. It was expensive. Very expensive. Several hundred pounds of expensive.
  2. I’m left-handed, so figured that a device which relied so heavily on handwriting recognition probably wasn’t for me.

Fast-forward to 1996. I was a student in Manchester, and a whisper was spreading round the electronics department like wildfire: a local Apple store had a huge box of Apple Newtons that they were selling cheap. Very cheap. £40 of cheap.

I rushed to the shop, and joined the long queue of students hoping to bag a bargain. It seems that the rumour had spread to our department from the computer science department – and just about all of them had got there first. The queue was so long, in fact, that I expected to finally reach the door just after they sold the last one. But I was in luck: they actually had two huge boxes of Newtons – and I really do mean huge. By the time I arrived there was easily another hundred or more in the box, so I had no difficulty in getting one.

The truth is that a lot of the Newtons in the box were “dead” or locked. Some of the computer science people managed to track down the magical incantations required to restart or reset them, and picked up a vast quantity of “dead” machines at £5 apiece – which they then sold on at a profit, of course. For a few days afterwards there was much swapping and trading of devices as everyone shuffled parts around trying to get a better machine, complete with a battery cage (so that it could take standard AAA batteries – those who failed to get one had to subsequently get creative with a soldering iron instead).

So although I started with a first generation Newton – or an Original MessagePad (OMP), as it’s known – I quickly upgraded to an MP100 – the second generation which looked identical, but had a slightly later version of the operating system. I managed to get one with a battery cage, too – though I did have to solder up my own serial cable to connect it to a desktop computer.

Having dealt with issue (1), it was time to move onto issue (2) – how would the MP100 deal with my lefty scrawl? The answer, it turned out, was “surprisingly well”. It took a little training to get the recognition accuracy up to an acceptable level, but we’re talking hours rather than days. Once the little black box was used to my writing – and with a little adaptation on my part, I’m sure – it was pretty damned good.

Perhaps more important than the recognition was the fact that it was easy to fix mistakes. With each correction it would learn a little more, so that the recognition became better over time, but even without this benefit the OS designers had obviously realised that being able to correct mistakes easily is just as important as getting things right in the first place. When a word was incorrect you would simply double-tap on it to see what other possibilities it could be – or to edit it letter-by-letter if the Newt had really fouled up. When it did get something completely wrong, a quick scribble through the word (or letter) would blow it away in an animated puff of smoke. Very intuitive, and also very satisfying.

So having got the first obstacle out of the way – could I actually use the thing – it was time to take a look at the software which was built into the Newt… and that will be the subject of Part II.

Another pen-based computer

I’ve just bought a Nokia 770 internet tablet (because it was going cheap). It’s yet another pen-based computer system of one sort or another, to add to the ongoing list of stylus-driven systems I’ve owned over the years:

You could argue that the Nitendo DS and DS Lite shouldn’t be listed alongside more general computer and PDA appliances, but with the recent addition of an R4 DS I can use various PIM applications and even run Linux on it.

I only have one TC1100 which dual boots between Windows Tablet PC Edition and Ubuntu (currently 7.04, “Feisty Fawn”).

Over the next few posts I’ll go into more details about each of these devices, why I bought them, and what I think of them, starting with the Apple Newton 100.