Update 18 March 2013 – Paper now online.

Humans are approximate creatures, we aren’t precise, and if this blog is anything to go by, we aren’t concise either! So then why do we persist in pursuing work which is ever more precise using tools which are sold on their precision. Eye-tracking is just one example of this – an individual gaze plot maybe precise, but start to add participants and you get gaze-spaghetti; nothing precise there. So lets step away from precision – in order to get some focus (no pun intended). This idea has led us to think about a new kind of metric we’re calling attention approximation, and our paper at the CHI13 DigitalTV Workshop is it’s first outing.

Here’s the abstract…

The move towards the provision of television content over two or more screens represents an enormous opportunity and a considerable challenge. A scientific understanding of how people switch attention between screens during television viewing is key to the development of this technology. We describe how ‘attention approximation’, a technique we have used to model visual attention and design screen reader presentation of Web content, can be used to investigate viewing behaviour, and ultimately drive the provision of content across multiple screens.

So why is this so important for my focus area of accessibility… well it’s all about ‘experiential transcoding’ – transcoding web resources not based on the underlying DOM sequence and hierarchy, but on the experience the designer/developer/author wanted the user to have. This is based on the design best practice in the web-development field. A web page is designed the way it is for a reason, to draw attention of various places. Transcoding based on an understanding of where people look means cognitive (and speed) benefits for blind and situationally impaired users. And I submit that attention approximation is the only way to drive how this transcoding should proceed.