Tablet PC interface: The write stuff

More details are emerging about the Tablet PC interface, which could turn out to be the most important IT innovation since the browser if it moves us from the keyboard to a more natural way of interacting with intelligent machines.

Microsoft, the master of hype, is being unnaturally cautious about its claims and intentions for the first generation of the platform, playing down the role of handwriting recognition.

Perhaps it has in mind Apple's pioneering pen-driven Newton handheld, which was killed as much by hype-driven expectations as by its poor reading skills.

Pen-driven Tablet PCs have, of course, been available for years; what has changed is that Microsoft has designed an interface around them, and has the power and perseverance to push it into the mainstream.

Like many major Microsoft projects, Tablets may not be an immediate success. But the company and its developer partners have taken great pains to ensure that the interface is immediately useful.

Early specialist applications will involve the kind of form-filling that was the mainstay of older Tablets. Well-designed electronic forms with tick boxes can get around many of the limitations of pen input.

But e-forms on the Tablet PC will also support both direct pen annotations (on accident-report diagrams, for instance) and handwriting recognition.

Curiously, it seems that, although recognition in on-screen forms can take context into account (the knowledge that a particular field will contain, say, only numerals), the facility will not extend initially to Microsoft Office.

So the recognition engine will get no help from knowing the data type in a particular Excel cell. Neither will it be able to use the facility in Word that expands short word forms to the correct spelling.

The Tablet PC specification insists that pens should be 'active', i.e. not relying on the screen to sense pressure and location.

This avoids the problem I reported with pressure-sensitive screens on early Tablet PC prototypes, on which recognition was confused by palm smudges.

A more contentious compromise involved not making the handwriting recognition adaptive: you can add words and names to its dictionary but it does not learn your particular little ways.

Bodin Dresevic, lead developer of the Tablet PC team, explained that this was to avoid new users having to go through a learning process and continually having to cope with correction boxes.

"There is 14 years' work behind the handwriting recognition in the Tablet PC and we have examples of thousands of scripts from people across the world," he said. "We don't need to make it adaptive."

Indeed the engine is the best I have seen, although it's far from infallible. But as people use it year after year they will want to use short cuts and perhaps even evolve a shorthand.

The best software is always easy for first-timers but has hidden depths for the more experienced.

Surely, I asked Dresevic, Microsoft could have allowed us to turn on an adaptive facility to allow us to be more creative in our use of the Tablet?

"I'm inclined to agree with you," he replied. "It is something we will probably look at for a future version."

The engine does allow you to combine speech and handwriting input, so that you can use one to correct the other, which is another area that may benefit from user experimentation.

Dresevic believes that people will start to use digital ink (as opposed to a text translation of it) as a primary medium for note taking and email.

This could mean that it will become more usual to pen a message rather than trying to tap out an SMS; a short digital-ink note takes around 30KB, about the size of an MMS picture file.

You can do many of the operations with digital ink that you can with word processed text, including paragraph and word inserts and searching for keywords.

There will be an odd feeling of déjà vu for anyone who has gone through early revisions of an application, especially in the early days of software when you could be delighted by the new features of an upgrade and yet still see ways in which it could be improved.

It can make you feel involved in the evolution, and I suspect this is going to be much more the case with the Tablet PC.

Developing the interface of the future is going to be a long haul and Microsoft is well aware of the fact. Dresevic observed drily: "There is no shortage of ideas for the next version."

Confusion over pen input at the desktop
There has been some confusion over whether handwriting recognition will be available on desktops for use with digitising tablets and other input devices.

One press release claimed that the Tablet PC modules would be available on the latest Windows XP service pack, but it seems that this is not the case.

I was also told that XP could process the output of Seiko's interesting Inklink device, which translates into digital ink anything you write on a standard paper pad.

But when I tried copying and pasting from the Inklink manager as instructed, I got a message saying 'Format not recognised'.

Dresevic suggested that there were two issues over the use of third-party devices. "The active pens use a sampling rate of around 130 events per second. The rate on most pressure pads is way lower, which could cause degrading of recognition," he said.

The second issue involves licensing. "A lot of research and a lot of intellectual property has gone into the recognition engine," explained Dresevic.

"You can't expect to get that for nothing. If device makers want to use it, they will have to license the software."

He added that the confusion over the service pack arose because product activation allowed the installation software to distinguish between desktop and Tablet PCs and load only software authorised for each platform.