19 October 2010

Information theory meets writing

Somehow the Royal Society's programme of talks fails to inspire me and so my recent visit there was only my second. The last was two years ago to hear a discussion on Telling stories with numbers, telling stories with words.

The carrot this time was Information Theory laced with Writing, which last combined at a Gurteen event early last year.

Professor David MacKay's talk was well structured, content-rich and well delivered, revealing the speaker to be very comfortable with his subject area.

He opened with some basics of Information Theory which more-or-less comes to effective versus efficient. To make information efficient we reduce redundancy, e.g. "Mry hd a litl lmb", but to make it effective we add redundancy (check digits and the like) to improve its readability over poor quality communications.

The talk then went on to describe some of the obvious weaknesses with keyboards where we use our very flexible fingers to press keys that are either on or off.

This led us in to a demonstration of Dasher, which Professor MacKay described as a walk through a library containing every conceivable book.


Dasher guides the writer through text showing, at each stage, the letter (or punctuation mark) that is most likely to follow what has gone before.

Here we can see that "writ" has been selected and the most likely options are "e", "ing" and "ten". In this case the finger has scrolled down to select "ten".

The idea is simple and has its obvious attractions, particularly for those of us used to writing with one finger while standing up on the train or tube.

And it's a free iPhone App so it's easy to try for yourself.

I did so and found it harder than I expected (not that I tried it for that long, to be honest). Part of this is that I am not very comfortable with the alphabet and I found letters whizzing past me before I could find them. A touch of Dyslexia perhaps.

But my biggest problem with Dasher is it does not address the problem that we talked about at the start of the talk, the inherent redundancy in the English language. A (possibly) better input system does nothing to solve the issues of inconsistent spelling and words inflated in length well beyond their phonetic requirement.

The problem it does help to solve is to help people who do not have the flexible fingers that I mentioned earlier and the talk gave several examples of where new technologies have enabled even the very immobile to communicate.

The talk did not quite do what I expected but I did learn a lot and I am grateful for that.

No comments:

Post a Comment

All comments are welcome. Comments are moderated just to keep out the spammers and all valid comments are published, even those that I disagree with!