December 2, 2002
©2002 Rich Stillman, Waystation Partners
Do you like this article? You can read more. Better yet, subscribe to my mailing list and you'll receive reviews of new devices and technology that can make your electronic life better today!
Comdex and a few recent field trips have helped me put pen computing in perspective and come to terms with my belief that pen-based tablet computers implement a fundamentally flawed product, but handwriting recognition is key to a new generation of personal computers. The secret is: handwriting recognition belongs in the pen, not in the computer.
In spite of the minimal progress made in the past ten years, handwriting recognition will be part of a breakthrough in the way people work with computers. But tacking pen capabilities onto an existing computer form factor solves the problem backwards, creating a product that combines the pen's unreliability as a computer input device with the computer's limited portability and durability.
The opposite approach - adding computing capabilities to the pen - would leverage the strengths, instead of the weaknesses, of both. Current "digital pens" capture handwriting much like a tablet PC, avoid the minefield of handwriting recognition, and are far more portable, usable and cost-effective.
As Moore's Law continues to enable smaller and more capable devices, a smart pen - one that reliably recognizes handwriting, can be used as a stylus on a tablet-style screen, and carries the processing power, storage capacity and network connectivity of a decent laptop - would eliminate the keyboard as an obstacle to development of truly portable computers, and would be a very compelling form factor for a future generation of personal computers.
The introduction on November 7th of Microsoft's TabletPC brought back memories of the last Golden Era of pen computing, in the early 1990s. Back then, Microsoft put its engineering and marketing weight behind PenWindows, a set of extensions to Windows 3.1 that enabled the use of a pen for mouse operations, digital ink and handwriting recognition. Quite a few companies picked up the pen idea, either licensing Microsoft's technology or developing their own. The pen was so much more natural to use than keyboards and mice that the success of this technology seemed virtually assured. It was just a matter of getting out a few bugs.
Fast forward to 1994, and the pen manufacturers everyone talked about were once again Bic, Parker and Mont Blanc. The success of pen computing remained elusive for two reasons:
Fast forward again, to 2002. Pen-based TabletPCs are again being sold by Compaq, Toshiba, Acer and others. In place of the eight pound tablet of 1992, current tablets weigh two to four pounds - better but still heavier than paper. Battery life limits their useful portable life to around four hours, with a hard stop at the end. They are guaranteed to break if dropped on the floor. And handwriting recognition is improved, but near-perfect? Not even close.
Let's ignore handwriting recognition for the moment, and concentrate on the issue of portability. There are many ways to measure portability, but functionally, the size of portable devices can be broken down into only six important categories:
Each of these size categories is a hard threshold, and an enabler. Portability enables access - a more portable device is more likely to stay with its owner as he or she moves around, and is less likely to be left behind at home or on a desk. A device that fits in a handbag is more likely to be carried than one that requires a briefcase; the increased access opens the door to a different set of applications and allows a different set of human factor design tradeoffs.
The power of crossing a size threshold - or failing to do so - can be seen over and over in the recent history of consumer electronics:
The challenge of shrinking any device comes down to these six size thresholds. A hardware company that wants to design any breakthrough portable device has to cross one of these barriers. If you're going to create a new generation of a traditional device in the same form factor, your product had better do something radically new, or be a whole lot better than current products at doing something that's currently important to the mainstream user. And it's the interpretation of "a whole lot better" that causes the industry to go down the pen computing rabbit hole again and again.
I don't get much chance to look inside the marketing departments of companies like Microsoft and Compaq, but if I did, I'm guessing that I'd find this belief: The end-user community contains a large population who don't like to type, and who regularly work with information that can't easily be represented by standard computer data types.
The pen computer offers solutions for both problems to the optimistic product planner. Keyboard-phobic customers can handwrite standard computer-readable text, using a method they've known since first grade. And all those sketches, doodles, marginal notes, revisions and annotations can be captured in their original form as digital ink and saved for posterity.
Trouble is, the pen computer doesn't do the first of these things particularly well, and the second isn't particularly important.
At its current level of reliability, handwriting recognition requires that the writer constantly supervise the result of recognition so misinterpreted words won't reach the final document. This makes handwriting entry a circular process:
This cycle is not only far slower than writing on paper but also disruptive to the thought process. It is slower than typing, even for barely competent typists. The slowness leads to frustration. In most cases, users abandon the pen and go back to the keyboard in order to get their work done.
As for the capture of digital ink, most of the data that starts life in this form finds its way into conventional documents through standard tools like Visio, Photoshop or Word. Much of what hasn't been converted has been lost because no mainstream software exists to effectively index and search this data. Like it or not, hand-drawn sketches and marginal notes rarely make it into final documents, and it is the final, machine-readable document that has long-term value. Most of the time, inked materials are just an intermediate form of the data that ends up in these final documents. The ink-based originals are actually useful only for a few hours to a few weeks.
Widely used solutions already exist to take care of most ink-based data during its relatively short useful life. The ideal storage medium - abundant, cheap and easy to use - is paper. An effective transmission medium, found in every office, is fax. And for those few handwritten documents that must find their way into digital form for long-term storage, scanners work well. For long-term, non-digital storage, file cabinets do the trick. And while direct-to-digital note taking does simplify some operations, like document distribution via email, the advantages are not nearly strong enough to make the tablet a compelling alternative to paper.
To summarize the argument so far: TabletPCs solve a problem most people don't need solved. Cheaper, more effective, widely accepted solutions exist that solve the same problem. And the TabletPC crosses no size thresholds, so its hardware offers no portability advantage and enables no new ways of working. In short, tablets seem compelling in concept, but don't pass the test of the market.
Now let's consider how another relatively new product may begin to fill an actual consumer need. The mainstream digital pen has been sold for about four years, beginning with the moderate commercial success of the CrossPad. Advances in processor size and speed, battery life, memory technology and short-range wireless communication have improved the breed, and current models, costing about $100, do a pretty effective job of capturing digital ink for later transmission to a PC.
As a practical tool for capturing digital ink, the pen has one major advantage over the TabletPC: it fits in a pocket. This puts it two size classes below a notebook, and means that a worker who buys a digital pen is very likely to have it within reach when needed. Digital pens even offer a bonus that no PCs, tablet or otherwise, offer: a hard copy of the written document, created concurrently with the digital ink and without the use of a printer or any other external equipment. And unlike the images produced by the tablet, the paper copies won't become inaccessible if the pen runs out of battery power.
So digital pens, with the ability to capture uninterpreted handwriting and drawing, offer some of the advantages of a tablet PC, with far better portability and at much lower cost. Let the computer worry about handwriting recognition later - the first job is to capture the idea and get it into the computer, and for this the digital pen is a far better tool than the tablet PC.
Look beyond currently available products, and the digital pen offers an intriguing window into the future. The current digital pen, like any electronic device sold today, is a product of our current ability to cram a specific amount of circuitry into a given space. As we are constantly reminded, this ability is evolving according to Moore's Law, which seems to have a few generations of viability left in it. And today's technology already enables powerful, cheap and tiny computers. For example, disposable computers, like Given Imaging's camera computer that takes pictures of a patient's GI tract after being swallowed, are already on the market and are nearly small enough to fit in a pen.
Progress is also being made in handwriting recognition - it's not an insoluble problem, only a very hard one. The day will probably come when a recognition engine can be trusted to convert handwriting to computer-readable text without constant supervision. This is the point at which the mainstream will accept handwriting recognition as a viable input technology.
Put these two trends together, and the pen takes on a whole new purpose. A pen that does a good job of handwriting recognition becomes a standalone input device. Write words on paper, then download the machine-readable document to the computer. It's the input advantage of the TabletPC, shrunk to pocket size.
And the input device isn't the only thing that can shrink. With advances in high-density non-volatile memory technology, we ought to be able to fit a useful amount of mass storage inside. Wireless networking, Bluetooth and audio should fit also. Now the pen is no longer an input device for the computer - in almost every way, it is the computer.
Of course, I did say almost. Computers require output devices as well as input devices. But here's where another new Microsoft technology comes in handy - Mira, which connects a portable screen to a remote PC through a wireless network so the screen and its attached stylus can act as the PC's display and pointer.
Mira's design goals originate in the current truth that screens are easier to carry than computers, especially non-portable desktop computers. But turn this assumption on its head, make the computer small enough to carry around in your pocket, and any Mira-equipped screen within reach becomes your display. The pen takes on the added role of stylus/mouse.
By the way, if you'd rather not base this design on a proprietary Microsoft technology, that's fine. X11 approaches the same issue from a different direction, using an industry-standard thin client as the output device. Or, for real portability, buy a pair of glasses with a heads-up display, and have a fully functional computer that fits in a pocket. Imagine the applications that will be enabled by that kind of portability.
As for me, I'm looking for a nice quiet investment in a company that manufactures pocket protectors, and lobbying my local school board to bring penmanship back into the curriculum.