The Focus Shifts to Palmpads

Even as notepad-sized pen systems begin to emerge, recent attention has nonetheless focused on the smaller, calculator-sized palmpad pen-based computers — Apple with its Newton, Tandy/Casio’s similar device. And, most recently, AT&T and GO’s “development relationship” for the purpose of creating a personal communicator device that will run PenPoint. Indeed, while PC has come to mean personal computer, it could well come to mean personal communicator within the next couple of years.

Before going too far, it should be emphasized that none of these devices exist. Apple optimistically says it will deliver Newton in early 1993, while AT&T/GO and Tandy/Casio also expect to deliver later in the same year. In the meantime, many problems and uncertainties need to be resolved before these palmpad systems are feasible, from improved handwriting recognition to seamless wireless communications.

But what is most intriguing about these announcements is that they shift the focus from general-purpose notepad pen systems such as the NCR 3125 to far smaller, special-purpose palmpad devices. There’s a good reason for the shift in focus: current notepad pen systems do not necessarily meet the requirements of mobile computing. The litany of shortcomings is well-known by now — they’re too expensive, too heavy, too hard to read, limited battery life, and so forth.

Enter the palmpad machine. By addressing the previously mentioned shortcomings, the marketeers say, palmpads render the notepad obsolete. All of a sudden we have highly intelligent devices weighing under one pound, with crisp readable displays, flawless handwriting recognition, wireless fax and modem communications, and software that’s so intuitive that it knows who “George” is when you scribble “lunch with George.” The system then puts the lunch date in your daytimer automatically or sends George a fax. And to top it all off, it runs for a week on AA batteries and costs less than $1000, maybe even as low $500. It sounds too good to be true — and, for the time being, it is. So let’s sort through the hype and see what these various alliances and new product announcements mean.

Apple’s Offerings

Of the three major announcements, Apple’s Newton has garnered the most attention and certainly the most speculation prior to the introduction of the machine at the June Consumer Electronics Show. As we discussed in our May, 1992, issue, the Newton is powered by the Advanced Risc Machine (ARM) processor, which, according to Apple, offers about twice the performance of a Mac II. (For a good technical overview of the Newton, see BYTE Magazine, July 1992.)

The prototype machine that we saw at the recent Mobile ’92 Conference fits in the palm of your hand and has a 5-by-3 inch LCD display/digitizer with pen input. The device includes a card slot that supports PCMCIA cards as well as more complex devices conforming to a specification Apple calls TRIMbus. A fax modem and audio speaker are built in and other peripherals will be offered on PCMCIA or TRIMbus cards. The Newton also has a built in infrared transmitter for communicating with other Newtons or infrared networks. System memory is supplied in the form of flash EPROM and will range from 2 to as much as 20 Mbytes.

Apple has designed from scratch a new multitasking operating system to operate the Newton. The OS includes handwriting and graphics recognition software and is built around a core piece of software called the Intelligent Assistant.

The Intelligent Assistant works essentially as the interpreter of the user’s actions on the screen. If you write someone’s name, the Intelligent Assistant will look for it in your address book and offer you a list of persons that have that name. Point at the appropriate name and the Assistant makes the phone call or sends the fax. Notes can be jotted down and the Assistant will convert them into a formal business letter (using predesigned templates, of course). Draw sketches and the graphics recognizer will convert crude shapes to the appropriate geometric figure (circles, triangles, etc.). As you write notes, they are stored in memory as they scroll up the screen. You can recall and tear off notes and save them at the end of your work session.

The Newton software is based on object-oriented data structures. For example, Newton has a single repository for all information that you’ll normally accumulate (business cards, phone numbers, dates, times, etc.). You’ll then be able to access the data in a variety of ways.

Handwriting recognition (developed by Apple’s Advanced Technology Group) will be based on the simultaneous use of several recognition technologies (including neural networks), each with its own strengths. Apple claims this will provide to better recognition and greater adaptability.

Additionally, Apple’s new language called DYLAN has been developed by Apple’s Cambridge, Mass. labs for Personal Digital Assistant (PDA) development. DYLAN, short for “dynamic language,” is a actually a hybrid derived from languages like Common LISP and Scheme. Primarily known as the language of choice of artificial intelligence aficionados, LISP uses “garbage collection” memory management techniques to address the memory limitations of small machines. Garbage collection basically works by throwing away unused data from time-to-time to save memory space.

A variation on the Newton theme is a line of multimedia personal digital assistants (PDAs) that will be jointly developed by Apple and Toshiba and manufactured by Toshiba, with each company releasing its own version of the device. It’s expected that the multimedia operating system will be licensed from Kaleida, the Apple/IBM multimedia joint venture. Standard PDAs will be designed for note-taking, communications, and the like, similar to a pad of paper; the multimedia variety will let you view content from a variety of sources — encyclopedias, videos, etc. One multimedia PDA software vendor that’s been mentioned is Time Warner.

Apple has been showing prototypes of the Newton machine running on a NuBus card on a Mac II. The demonstration is impressive. The handwriting recognition appears to be fast and highly accurate. In the demonstration, the audio speaker is used to provide the sound of pages turning or tearing off a notebook as you scroll up the screen. Part of the demo includes the use of a SkyTel pager to receive written messages on the Newton.

The one big question is power consumption. The ARM processor itself is very power efficient, consuming less than 500 milliwatts (mW), according to Apple. But other components of the system must also be designed to be extremely power efficient; in particular, the display, the fax modem, and the card slot peripherals.

Tandy/Casio/GeoWorks

There is considerably less information about the Tandy/Casio deal. The companies issued a press release announcing their intentions to “work together to design, develop, manufacture, market and distribute a new family of Personal Information Processors.” The announcement also states that GeoWorks and Palm Computing will provide the operating system and applications software respectively.

GeoWorks is working on a pen version of its GEOS graphical user interface, to be called PenGEOS, as we have reported in previous issues. The operating system is designed for the Intel architecture and runs on 8088 processors. However, demos that we’ve seen of PenGEOS show that it’s pretty slow on an 8088, particularly with the addition of handwriting recognition. Of course, Intel offers the advantage of a wide variety of processors to choose from so that a range of price and performance should be available in the Tandy/Casio product line.

Palm Computing is headed by Jeff Hawkins, formerly of Grid Systems, and the main architect of the handwriting recognition software used in the GridPad. Palm Computing is working closely with GeoWorks so that their partnership on the software side of the Tandy/Casio project is not surprising.

Tandy/Casio undoubtedly will aim at the mass merchandise consumer electronics market and will offer very competitively priced products. Whether they will have the processing power or sophistication of something like Newton remains to be seen.

AT&T/GO

Like the Tandy/Casio deal, there is only general information on the AT&T/GO arrangement. As we wrote in our last issue, the project will use AT&T’s Hobbit chip, based on the CRISP (C-Language Rational Instruction Set Processor) architecture that AT&T has been working on for some ten years. Knowledgeable chip watchers, such as Michael Slater, say that the Hobbit chip will be about twice as powerful as an 80386 (putting it somewhere in the same class as the ARM chip) and with very low power consumption (also around 500 mW). The relationship with GO makes good sense because PenPoint is written in C and the Hobbit chip is optimized for running C instructions. The Hobbit is reportedly also optimized for working together with digital signal processors, to be used for handwriting and speech recognition.

According to GO and AT&T executives, the two companies have been working together for about a year on the joint venture. PenPoint is being ported to the Hobbit architecture and is expected to be ready in mid-1993. AT&T and GO will license the Hobbit/PenPoint platform to other hardware vendors. It is widely anticipated that Eo Computer (a spin-off of hardware engineers from GO, as described in our May issue), will be the first company to introduce a system based on the Hobbit/PenPoint platform. Like Apple and Tandy/Casio, AT&T and GO emphasize the communications aspect of these portable devices. The version of PenPoint for Hobbit (as well as other versions of PenPoint) will include integrated messaging, allowing the system to automatically send and receive faxes, electronic mail, and so forth.

Other PalmPad Players

These companies will certainly not be the only ones venturing into the handheld personal communications business. Hewlett Packard is known to be working on a pen-based version of its DOS-based HP95LX. In addition, HP has the hardware know-how to come out with something much more powerful than the 95LX, perhaps along the lines of Newton.

IBM can also not be overlooked in this market. IBM has experience with its PC Radio product which is used for messaging in the field, and also is competing in the general purpose pen market with its ThinkPad.

Another important player in this market is Sharp Electronics. First of all, they’re working together with Apple on Newton and will market their own versions of the product. Secondly, Sharp has introduced a pen-based version of its Sharp Wizard, and has plans to expand its capabilities in new versions of the product. Motorola and Samsung of Korea have also announced a joint venture to develop a pen-based palmpad with wireless capability.

Where’s Microsoft Windows?

No discussion of a new market in pen computing would be complete without including Microsoft and Pen Windows. So far, Microsoft has little to say about these announcements, other than that the company is very interested in the palmpad market. It seems to us, however, that Microsoft Windows would be poorly suited for a palmpad machine. There’s simply too much overhead with Windows on top of DOS for a special purpose handheld device. What seems more likely is that Microsoft will develop a scaled down version of Windows NT, its 32-bit multitasking version of Windows, for use in handheld devices. But Windows NT is still in beta testing and probably won’t be widely used for at least another year.

Communications is the Key

The key and probably most challenging ingredient for the success of these machines is seamless, wireless communications (for more on wireless communications and palmpads, see related article in this issue). As Andrew Seybold put it in his newsletter, “without an umbilical cord of some kind (wired, infrared, or wireless), Newton is, at best, a much improved Sharp Wizard or Casio Boss.” Indeed, the whole concept relies on excellent communications. Even for use as a personal database or game machine, users will want to be able to purchase software via telephone and have it downloaded to their machines.

All of these machines will feature wireless communications. In the work world, the most likely form of this communication will either be cellular telephone or wide area networks such as ARDIS or RAM Mobile Data (home consumers will probably plug into their home phone system). The problem is that these are expensive forms of communication. It’s fine for a big corporation to spend a few hundred dollars a month on an executive’s cellular phone charges, but quite another matter for individual consumers or even small businesses. Paging systems such as SkyTel are also not cheap.

Nevertheless, these communications services will play a major role in the success of these devices. Many companies will be able to justify the cost of cellular or radio network communications for certain work operations.

Another key communications aspect will be central host services for mobile computers. Users will be able to sign up with a service through which they can send and receive e-mail. They will be able to back up data through this service, and purchase and download application software from the service. All of the above companies are working on this concept. Again, cost will be a major factor.

In conclusion, much work remains to be done. We are pretty skeptical that Apple can deliver a product as early as Winter of 1993. We think that all of the above players won’t deliver until at least the end of 1993. In fact, AT&T’s president of the Microelectronics division talked about a range of devices priced from $300 to $1500 appearing in 1995 (he also said that devices costing about $1500 would appear before then).

In any case, there is one very positive side to the palmpad announcements. It raises the standards for the entire notepad pen market — lighter weight, longer battery life, better communications, lower cost. Hopefully some of the features promised in these one-pound handheld wonders will appear more quickly in larger, but potentially very useful notebook sized machines such as the NCR 3125.

—

The following boxed announcement appears at the foot of column 1 on page 1:

Pen-Based Computing is now bi-monthly! We are pleased to announce that Pen-Based Computing will from now on be published on a bi-monthly basis (6 issues per year). In conjunction with the new publication frequency, we have raised the subscription price to $96 per year ($120 overseas). Current subscribers will receive the number of issues remaining on their subscription and then will receive renewal notices. We apologize for any inconvenience this may cause but we feel that the pace of new developments in pen technology requires us to increase our frequency of publication as well as our subscription price. The Editors