I’ve been banging on about experiential transcoding for quite awhile now - well since the EPSRC ‘SASWAT’ project and the Google ‘ACup’ project. The rationale is simple, sighted designers build pages to attract a sighted users attention to ‘important’ parts of the website. Even if they aren’t good designers, people will generally look at similar parts of the page as each other - or so eye tracking studies tell us.