"Inking" out of the box: manufacturers of pricey tablet PCs bet digital ink will go mainstream.

IT'S DEJA VU ALL OVER AGAIN. Pen-based computers are resurfacing, thanks largely to Microsoft's new Windows XP Tablet PC Edition, introduced in the fall of 2002. There's no denying that the new Tablet PCs are, well, fun. But handwriting recognition as computer input has had several major forays into the marketplace, most of which were spectacular and costly failures. The Apple Newton comes to mind. What's changed so much that Tablet PCs priced at twice as much as a serviceable laptop will make consumers dig for their checkbooks this time around? Perhaps the best single-word answer: Microsoft. None other than Bill Gates is talking up Windows XP for the Tablet PC and Microsoft's inking technology.

Half a dozen hardware manufacturers are on board, ranging from pillars of the laptop industry like Toshiba, to upstarts like Motion Computing. Unveiling new Tablet PC products that emphasize handwriting via digital ink as the preferred input mechanism, they are betting inking is ready for prime time. How well does this work and what are the larger implications of moving away from our long suffering pal, the ASCII character set keyboards use? Glad you asked ...

Making a mark

There are two basic jobs a pen-based computer has to accomplish. First, it has to detect the stylus as it moves over the screen, providing a form of feedback that gives the user the feel of writing on a piece of paper. Second, it has to store the "inked" input and retrieve it in a useful and relatively convenient fashion. Let's look first at how tablet PCs simulate the writing experience.

The key difference between the hardware of an ordinary laptop computer and that of a tablet PC is the digitizer that overlays the LCD screen. Conceptually, a digitizer is a grid that creates an electromagnetic field. The digitizer and the pen interact electronically to track the motion of the pen. When the pen is in contact with the digitizer, its position is recorded as a series of X/Y coordinates on the digitizer grid. The "handwriting" you record in this way is really a series of small vectors. As the stylus moves across the screen, its position is sampled about 130 times per second. (Studies show that anything less than 100 samples per second gives the user the sense that they're "dragging" the ink, rather than writing freely and normally.)

Here are a few subtle but important points. A single ASCII key-press generates a couple of bytes of data at most. In contrast, detecting handwriting requires the computer to do a stupendous amount of work. A single handwritten character might generate a thousand digitizer grid sample points. With today's fast CPUs, that might seem irrelevant, except for one thing: power consumption. Tablet PCs are designed to be highly mobile devices, so battery life is an issue. Powering the digitizer and handling the vast amounts of data that inking generates take a serious toll on battery life. Today's Tablet PCs are heavier, bulkier, and more power-hungry than their similarly capable laptop cousins because they use power less efficiently than keyboard-based devices.

In addition, writing on a Tablet PC isn't like writing on paper. The stylus has a slippery feel on the screen, which causes most people to write in a larger script than normal. This makes writing on a tablet slower, at least initially. Also, the stylus and the tablet are a matched pair, and you have to have both of them in order to do anything with either. This isn't the case with handheld devices that let you input data using a stylus. Handhelds use pressure-sensitive screens, so you can write on them with anything pointed--even a fingernail. The Tablet PC stylus has an induction coil to help the machine track the pen point more accurately. If you lose a Tablet PC stylus, it's going to cost you.

The act of writing

Not all of the handwriting processing takes place on the Tablet PC. A significant amount of neurobiological research shows the act of handwriting triggers a physiological organizing process in the brain. This explains why note-taking improves memory retention, even if the notes are incomplete. Handwriting also aids the brain in establishing relationships between stored information. Important and closely related items of information are coupled more tightly in memory; less important or extraneous information is moved, or even forgotten.

This bit of human biology has long held the "paperless office" at bay, but it's a point in favor of the Tablet PC. Some of the users of 1:abler PCs will prefer to retain their handwritten input as "ink." Windows XP for Tablet PCs and its ink-enabled Office applications elegantly support this (figure 1). Things become more complex if you need to store handwriting as ASCII--and you'll have to do this if you want to treat the input as data (i.e., search text, cut/paste, etc.).

[FIGURE 1 OMITTED]

An expression of biology, handwriting is almost as highly individuated as speech. Herein lies one of the great challenges to handwriting recognition technology. All handwriting recognition systems rely on databases of rules, that are developed by analyzing large numbers of handwriting samples. A recognition engine uses the rules to match a particular group of handwritten marks to a character. Then, given a group of characters, it tries to match known words. Using rules about grammar and context, it recursively corrects and improves its guesses about the characters and the words. It's a statistical process that simply can't be 100 percent accurate out of the box because of the individual nature of handwriting and the generalization necessary to create analysis rules.

However, analysis accuracy can closely approach 100 percent with training. By training, you teach the engine to decipher the peculiarities of your handwriting, providing samples with which the recognizer augments its rules. Training has proven unpopular with consumers because it's a lengthy and tedious process. However, without it, even the best generalized recognition engine has an error rate of about 3 percent. That doesn't sound like much, but in a 1,200-word article such as this one, that means 36 typos. Yikes!

What if you're happy with storing your handwritten notes as ink? Your problems are solved, right? Not exactly. Although you can send inked messages to anyone as e-mail, if the recipient isn't running an ink-enabled application, your e-mail turns up as a bitmapped image. This can be problematic for a couple of reasons. As I mentioned, you can't full-text search or cut/paste text from a bitmap. Second, there's an issue of bitmap size. That's cumbersome. Also, if you want to collaborate with others who are generating inked documents, annotated spreadsheets and the like, you'll need to update both your operating system and applications to run the new versions of Microsoft's products. That's costly.

In the applications available from Microsoft, the inking format is more than a simple bitmap. At a hundred and thirty samples per second, you generate a huge amount of data, but a lot of it is redundant. Think about how far you can push an ordinary pen in one second, and divide that into 130 increments, Everybody understands that at that kind of sample rate, we're going to do a lot of data compression, probably at several stages: throwing away duplicate input coordinates, smoothing curves, and using some kind of analysis to summarize stored data. Otherwise, recipients of inked e-mail would just throw it away without reading it. Imagine naively sending huge bitmapped pictures of handwritten e-mail-what if the screen resolutions don't match? There are other technical problems like that. Inking can't be broadly successful without some kind of fairly sophisticated storage format and input handling technology.

The price of convenience

There is some suspicion about the greater significance of Windows XP for the Tablet PC, and especially about inking. The way XP captures ink, recognizes handwriting, and stores the compressed input are all proprietary Microsoft technologies. If ink takes hold, it will certainly confer market advantages on applications designed to exploit it (say, for example, Microsoft's Office suite).

As has been the case with other Microsoft technologies, the trade-off for convenience is flexibility. Relying on a strategy not unlike the one that landed it in court for the last several years Microsoft is keeping key aspects of the inking technology to itself, shutting other application vendors out of the market Thus, Microsoft gains a toehold in an important emerging market and creates conditions that may sequester user data, binding it to Microsoft applications far into the future. This limits what you can do with your data. To get a feel for the potential impact of this, imagine the consequences if one company owned the rights to the ASCII character set.

It's the economy, sweetie

Even in the spotlight of Microsoft's high-wattage persuasion, the Tablet PC and its cleverly packaged digital inking features look like an expensive novelty. Technology businesses continue to cut staff, and industries across the board are slashing IT budgets, making technology forecasters sad men and women, indeed. In these times, be you an ordinary consumer or an IT manager, you need a sound and rational reason to pay the premium attached to a Tablet PC. Tablet PCs and digital ink will be strong contenders in the niches already occupied by pen-based computers, such as shop floor processing and the vertical solutions pioneered by overnight shippers, but Microsoft and its manufacturing partners are going to have to make some shifts in strategy and technology to see Tablet PCs move into the realm of low-cost laptops and subnotebooks.

MOBILE BUSINESS BENEFITS

Many business users subscribe to the hunt-and-peck school of typing and, In some situations, a keyboard can be awkward; so, at first glance, tablet PCs seem like a boon to the business world. But, there are downsides to digital ink, and the Microsoft XP Tablet PC platform, that may significantly narrow its appeal.

* If the Tablet PC is a good fit for your work environment, go to htttp://www.advisor.com/doc/11455 for more information some of the models that are available.

Nancy Nicolaisen is a software engineering consultant, trainer, and author of a new book, Making Win32 Applications Mobile: Porting to Windows CE, published by John Wiley & Sons. Her firm, N2 Software Engineering, recently released FieldTrip, a forms-based data collection program for handheld computers. http://www.fieldtrip2go.com.