At the end of all this, then, one has, starting from the lowest level:

<!--[if !supportLists]-->a)      <!--[endif]-->information as what is
processed by a computer;

<!--[if !supportLists]-->b)      <!--[endif]-->information as a scalar
quantity of uncertainty removed, the entropy/negentropy picture;

<!--[if !supportLists]-->c)      <!--[endif]-->semantic information as
well-formed, meaningful data (Floridi);

<!--[if !supportLists]-->d)      <!--[endif]-->information as a process
operator that makes a difference to and for other processes, including above
all those of receivers and senders.



Dear Joseph and colleagues, 
 
I agree with the distinction of four operations, but it seems to me that
this can be expressed more parsimoneously using information theory. Given
Bateson's (1972) formulation that information can be considered as "a
difference which makes a difference",  one should distinguish between the
first type of differences and the second. Let's say difference(1) and
difference(2). (I'll need difference(3) and difference(4) below.)
 
A difference(1) can only make a difference(2) for a system (or more
generally the expectation of a system). This difference(2) is analytically
preceded by difference(1), that is, pure differences. Shannon-type
information is contained in probability distributions. In the binary case,
this is only one difference (Y/N, F/T, open/closed); in the non-binary case
probability distributions provide us with sets of differences(1). These
differences(1) can only make a difference(2) for a system which contains
other (orthogonal) differences. In this case one needs one-more (orthogonal)
dimension of the probability distribution that positions the incoming
(Shannon-type) information at specific moments in time. Thus, difference(2)
presumes at least a dimensionality of two in the probabilistic entropy.
 
When the system develops, difference(3) can be defined with reference to the
time axis (recursion). This is Brillouin's (1962) Delta H. The difference(1)
that made a difference(2) for the system makes a difference(3) over time.
When the system operates as a self-organizing, autonomous or autopoietic
system it is additionally able to provide the information with a meaning
from the perspective of hindsight, that is, against the axis of time. This
"incursion" can make a difference(4). 
 
In other words, one needs at least a vector (one dimension of the entropy)
for containing an uncertainty. One needs (at least) two dimensions of the
probabilistic entropy for positioning the information in a network (matrix)
at specific moments of time. Three dimensions are needed when the time axis
is additionally included; four when the direction in the time axis can be
considered as another degree of freedom.
 
The two approaches seem very akin to me, but I claim that mine is more
strict and parsimoneous because I only need numbers of dimensions of the
probabilistic entropy and not concepts like differance. The next-order
probability distributions can be considered as the probability of
probability distributions, etc.
 
Best wishes, 
 
 
Loet
  _____  

Loet Leydesdorff 
Amsterdam School of Communications Research (ASCoR), 
Kloveniersburgwal 48, 1012 CX Amsterdam. 
Tel.: +31-20- 525 6598; fax: +31-20- 525 3681 
 <mailto:[email protected]> [email protected] ;
<http://www.leydesdorff.net/> http://www.leydesdorff.net/ 

_______________________________________________
fis mailing list
[email protected]
https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis

Reply via email to