I just read on gizmodo.com that researchers at the University College London have developed new software that can “perfectly” replicate anyone’s handwriting. I suppose to a layperson, that sounds “perfectly” feasible. After all, computers can do a lot of things that we never would have dreamed of. But, I question how can a computer algorithm recreate the intricate workings – our brain, nervous and muscular-skeletal system – that produce handwriting?
First, I wonder just how natural such handwriting can look. Supposedly, there are line width variations that represent the up- or down-strokes and ligatures (connecting strokes) are added and blended into the image. It seems to me that this program would have to contain a profound amount of data in order to provide the natural variation of these human features.
Then, what about the incidentals – the quirks and nuances of natural writing? Can software provide for all the ways that a person dots his ‘i’, for instance? I rather doubt it. The sampling for this quest would have to be huge to take into account all of these variables.
My point is that as good as computers have gotten at replacing some of our human endeavors, they cannot replace us (yet) when it comes to creating authentic handwriting. I must admit, however, that the result below is a very decent rendition of the original handwriting. I am reasonably certain that it would fool almost everyone depending on the writing being analyzed. But there is still a quality about it that appears stilted and not spontaneous. As a handwriting expert, I would definitely be cautious about my comparison of these two words. Fact is, that there would probably be more writing to compare in a real case, making it more likely that the patterns of the computer output would be easier to spot.
So, for now, my role in identifying and authenticating handwriting is safe; maybe even more valuable than before.
What do you think? Can you tell the difference?