Cal writes:
Percy writes:
I wasn't saying anything about Gitt information.
Oh, sorry. I guess I let the thread's title lead me astray.
Well, now I'm confused. Are you saying that when you wrote this in
Message 13:
Cal in Message 13 writes:
My problem is in the circular way "information" is defined:
What is a code? That which contains information.
What is information? That which is represented by a code.
What is DNA? Coded information.
That you were talking about a problem with Gitt information, not Shannon information? It was written in response to a paragraph from WK about Shannon information.
If I axed you to relay a message for me, and it contained sum miss pelled words, would you have a reilable way of knowing whether those apparunt errors were kritical to the meaning of the "aksuall message"?
Yes, of course, but that's because of redundancy. The correct letters are only one of the many cues for the correct information.
Meaning isn't merely irrelevant to Shannon's purpose. It is assumed that among the set of all possible messages exists one which is the most meaningful,...
No, this is not assumed. As Shannon correctly states, meaning is irrelevant to the engineering problem.
...but it is not only unnecessary to choose between them on that basis, it is absolutely imperative that no attempt be made to do so; the message must be considered to have been at its maximum fidelity at the signal end.
My guess is that you're making the mistake of thinking that communicating information in the presence of noise is somehow associated with meaning. If an encoding of information contains insufficient redundancy such that after transmission through a noisy medium it is ambiguous which message from the set was originally sent, selecting a message from the set based upon some criteria is not an exercise where meaning has any role.
--Percy