To be fair it's a lot easier to write on a screen like pen and paper than on a tablet because there is a huge disconnect with that. When I used Inkwell it recognized a lot more than what was shown in that video. As far as Microsoft's recognition goes, it's definitely better, but slow. I'd rather just type.
Lets call a spade a spade. The writing that is being used with Inkwell would
easily be good enough to be recognized with Vista's HW recog. I've played around with Inkwell in the Appel store by grabbing a sheet of paper, writing on it, and then using the WACOM tablet with the sheet on top, wrote in inkwell. The results were NOT even remotely impressive. To be blunt....Inkwell sucks....hard and rightly so. Apple hasn't done crap with it since day one.
Slow is a relative term. You do NOT want the recog to do it thing too fast or in the event of a mistake you are going to have to back out of the entire word and sentence instead of just a letter. Microsoft's solution, in Vista...XP's is all kinds of meh, is easily the most elegant.
I've spent plenty of time thinking about this. There are 3 levels of bandwidth when it comes to manual data input.
The slowest will always be handwriting. It takes time to write something out, and as such it is, what I term, bandwidth limited. However it is also the most flexible when it comes to taking notes, drawings, document markup, capturing signatures, etc.
The second is a keyboard. Nice, practical, cheep, and something in the range of high bandwidth when it comes to data input. however it consumes space, and it makes noise (Depending on the type of keyboard and the environment it is being used in.).
Finally you have voice to text recog. This is by far the highest in terms of overall "bandwidth". The catch is that the software, at least the last time I looked at it, hasn't gotten to the point where you can speak fast, which sadly puts it closer to the keyboard in terms of "bandwidth". Usually..it..needs..to..be..in..a..slow..rhythmic..and..even..tone to let it correctly ID your words, and even then most software packages are hit and miss usually in the greater then 80-85% range. With training it can usually get _over_ 90% though. The beauty of this is that it doesn't require any peripheral to use, beyond a microphone array. No pen to loose, no keyboard to take up space. However the drawbacks are obvious. There is no way, at least for people who aren't annoying pricks, that someone is going to be dictating to their tablet on the bus, or in the mall. And no way they are going to do that in a boardroom. That and I've seen CPU's spike when using voice recog. Battery life becomes an issues, for now, with such tech.