Machine readability

The general aim of separation of presentation and content is machine readability, that is, making it possible for machines to detect meaning or intent. (The machine readability is then a means to desired ends, as discussed below.) For example, a human being reading a document has little trouble to grab from context that an oblique rendering in one place would be emphasized text, but in another place is a title of a book. However, as robots and crawlers have more difficulty with this task, separation of presentation and content generally aids in their distinguishing of such things which are presented in the same way, but have a different meaning—or have the same meaning, but are presented in a different way.

Machine readability allows affordably serving the information to a wider variety of users (in a presentation that they can understand), where users may be humans or machines. This requires the ability to recast abstractions in new instances quickly and cheaply (that is, without time-consuming reworking), which generally requires automation rather than person-hours of labor. For example:

• The ability to deliver the same information in different media, and to change the medium quickly and cheaply; and within one medium, to change instances easily
  • To serve the same message to different users:
    • as printed display (for technophobes or for users with contextual desire for print)
      • as printed display with the typesetting recast into various graphic designs without time being spent on any manual reworking of the content (a good example is given at CSS Zen Garden)^[2]
    • as online visual display (for most users in most instances)
      • as online visual display in various graphic designs
    • as online audio (for blind users or for sighted users with contextual desire for audio)
    • as braille (for blind users)
    • as input to an API (for users that are machines)
      • The other machines can then take the information and do further transformations or actions. These may be ones that people cannot do (or cannot do quickly and cheaply), but machines can do (and can do [more] quickly and [more] cheaply). For example:
        • Take a book and translate it into another natural language
        • Take audio of speech and translate it into another natural language
        • Take audio of speech and transcribe it for reading (for deaf users or for hearing users who want transcription of voicemail into e-mail or IM)
        • Take the data contained in an entire library and search through it for ABC-XYZ, then turn every instance of ABC blue, and serve every instance of XYZ to a machine that will categorize it