Saturday, June 25, 2005

Information as Probability

Information has been the inverse of probability since the 1940s: that's how the telephone people saw it. That's how they needed to see it. Specifically, Shannon's definition is well known:
  • H = -∑pi logepi
  • Thus there's more information in the series "1, 15, 138 ..." than in the series "2, 4, 6, 8 ..." If you go to the movie and everything happens exactly as you would expect, if the blond looks like you would expect, dresses as you would expect, if the fiancé says exactly what you would expect, right down to the expression on his face, then the movie contains no information, it isn't very good, even fans of dullness won't like it.
    Information is the spice: the only spice ever needed.
    But how about considering information as related to probability in the following sense: cultures hear and understand only what they expect to hear and understand.

    I've seen people listen to sermons; and hear only what they wanted to hear: read books; and understand only what they expected ...

    I'm still fussing with the Mi.org Entrance, still don't have it quite right, and that must take priority; but when I can I must develop this wrinkle. Meantime, for example: if you take a people whose idea of god is a patsy for favors, you can then steep them in a literature of martyrdoms and non-communications, non-cooperations, you'll still have a people who believe in god as a lucky charm. Indeed, we do: Noah, Abraham, Moses ... Jesus ... make no impression.

    No comments: