Tech talk: The writing's on the wall for Tablets

Microsoft's refusal to enable the Tablet PC to "learn" will limit its value as a business tool.

Microsoft's refusal to enable the Tablet PC to "learn" will limit its value as a business tool.

As Samuel Johnson once said, in less enlightened times, "Sir, a woman's preaching is like a dog's walking on his hind legs. It is not done well but you are surprised to find it done at all."

So it is with handwriting recognition. I have seen many demonstrations of Microsoft's Tablet PC converting handwriting to text and, without fail, it fails - even on short demos. It attempts its trick boldly and does impress but in the business world it's still a dog.

To succeed, handwriting recognition has to be 100% reliable. At the moment, it is good for personal notes but for insurance firms using it for form filling it is too unreliable. When using a Tablet PC, it is better to stick with the handwritten word preserved in Digital Ink and let a human try to decipher it later.

What we are asking of the hardware cannot be delivered because the analytics are not there. Part of what the brain does is to examine content and context of the legible part of the text to decipher any unknown words.

It is a bit like the discovery of the Rosetta Stone. Once the archaeologists saw the Greek translation of the hieroglyphs on the stone, these known words could be applied to other examples of hieroglyphics and new words could be guessed. Now a large part of what was once an enigmatic code can be read by most Egyptologists.

Tablets lack this contextual intelligence and are doomed to fail in the task of translating written script but, with careful use, they are probably more usable now than at any time during the past decade of tablet development.

Although the manufacturers seem sure that Microsoft's backing will push tablets into the business arena, not all are convinced. Notably, IBM has no plans to re-enter the tablet market following its failure to make an impression with its own attempt in the late 1990s. Dell, too, is holding back - but the company usually ignores new markets until there is sufficient volume to make entry worthwhile.

Where Microsoft may have erred with handwriting recognition is in deciding to not allow the translation engine to learn or be taught new tricks.

One of the potential users the company suggests is the "information worker" - someone who attends meetings and makes notes or takes minutes. Microsoft paints a picture of these people scribbling down the proceedings, converting it to text and being able to publish this immediately. Even assuming recognition would be 100% accurate, as a journalist I have to ask, "Does everyone in the world, apart from me, take notes in long hand?" I think not. These will be in shorthand or some form of speed-writing, using abbreviations.

If a tablet could learn, it could be taught shorthand, or whatever, and this would greatly improve the information worker's life.

It's such a good idea I think I'll patent it.

Read more on Data centre hardware