| Ø There are two ways we can use the idea "in-form". Yes, this is the other notion of information. Shannon-type information does not “inform”, but is counter-intuitively defined as uncertainty (or probabilistic entropy) and measured, for example, in bits of information. It is based on probability distributions. |
Surely Shannon information is not uncertainty, but its opposite: the reduction of uncertainty. And it has that in common with meaningful or semantic information.
| Bateson (1973) and many others did define information as “a difference which makes a difference”. |
| Probability distributions contain only differences. If these first-order differences make a difference in a second dimension then a system of reference is assumed for which the first-order difference may make a difference. This system of reference may then discard some incoming information as noise and provide meaning to other information. Perhaps, it is useful to call this meaningful information (or observed information) as different from the expected information (or uncertainty) in the case of Shannon-type information. |
I find it useful to view Shannon information as "pure pattern". But that might be specific to my particular interest in it, which is its relationship to physical information. (I don't mean that in other contexts it might be wrong to view it that way, but it might not always be the most useful way to look at it.)
| The system of reference does not have to be “an observer” as is often presumed in the cybernetic tradition; it can also be discourse. Does this contribution make a difference for the discourse? |
Who or what but an observer can make that judgement? Only to a mind is anything ever meaningful. I read "a difference that makes a difference" as "a significant difference", and only a mind can judge significance.
| The two notions of information are to be kept apart because otherwise the discussion becomes confused. |
I certainly agree with that!
--
Robin Faichney
<http://www.robinfaichney.org/>
_______________________________________________ fis mailing list fis@listas.unizar.es https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis
