We had an inquiry last week from a reporter about handwriting recognition. This reporter found it interesting that literally no software provider had broadly addressed handwriting recognition it in the consumer space since the days of the Palm handheld PDAs.
This is a very interesting question and there are many reasons why engineers and companies have not been successful in tackling this problem. There are both technical and practical business reasons for the absence of a solution that when combined make this problem almost impossible to solve.
First, let’s discuss the technical difficulties and solutions.
Tackling this problem on a PDA was easier for a few reasons. First, a PDA has the benefit of dynamic information that provided the time and position of the stylus. In a paper or scanned image of the document this information must be calculated for each character. Secondly, the PDA required block (non-cursive) letters and in some cases was even stricter in exactly how you drew your characters.
There exists very good handwriting recognition software today, including technology available from Nuance, which can recognize block letter handwritten characters on scanned documents. Very good means that with neatly drawn block letters you can attain a recognition rate in the mid 80% range. Machine print can be recognized in the 99% range.
Now let’s discuss the business reasons for not making this current technology available.
Even though Nuance has hand print recognition technology we don’t provide it in our consumer products. It’s only available to other developers who wish to incorporate it into commercial applications. The reasons are as follows:
Consumers are using less and less handprint in their communications. When was the last time you received a handwritten correspondence that you wanted to convert into an editable digital document? Almost everyone sends email these days so the demand to convert handwritten text has greatly diminished.
Second, while an 80% recognition rate sounds impressive enough, in practical use recognition can be far less accurate. Even in the best cases it’s simply not accurate enough to please most consumers. Therefore the value of the technology is greatly diminished.
And finally, consumers simply don’t like to be told how they must write a character. This was the principal reason it failed on PDAs. Good penmanship or the lack thereof, is not something people like to concentrate on. It’s human nature. And trying to recognize cursive handwriting is simply impossible.
How many of you handwrite communications on a daily basis? We’re interested in hearing your thoughts about handwriting recognition and how you perceive its value. Feel free to comment on this post.
Comments on this entry are closed.