I note that there has been some relatively recent work bridging Shannon's 
theory (and probabilistic theories more generally) and the Channel Theory of 
Jon Barwise and Jeremy Seligman. 


Allwein, Gerard. 2004. A qualitative framework for Shannon information 
theories. In Proceedings of the 2004 workshop on New security paradigms , 
23-31. Nova Scotia, Canada: ACM. doi:10.1145/1065907.1066030. 
    • Allwein, G.T., Moskowitz, I.S., Chang, L.W.: A new framework for Shannon 
information theory. Technical Report A801024, Naval Research Laboratory (2004) 

Seligman, Jeremy. 2009. Channels: From Logic to Probability. In Formal Theories 
of Information , 193-233. http://dx.doi.org/10.1007/978-3-642-00659-3_8. 

and some intriguing work approaching quantum mechanics via Chu spaces, which 
are of course related to channel theory: 


Abramsky, Samson. 2011. “Big toy models.” Synthese . 

These may be of interest to some. 

Jacob Lee 

----- Original Message -----
From: "Loet Leydesdorff" <l...@leydesdorff.net> 
To: "joe brenner" <joe.bren...@bluewin.ch>, fis@listas.unizar.es 
Sent: Sunday, May 8, 2011 12:16:01 PM 
Subject: Re: [Fis] replies to several 

Dear Joe: 

1. If I follow Loet, I must accept that Information Theory is essentially a 
mathematical theory that requires abstractions for extension to complex 
contexts. But Bob says that the mathematically derived “meaning” for antibodies 
is a pale representation of meaning in the human context and only reflects how 
wanly quantitative models in general prefigure more complicated human 
situations. CONCLUSION: something else that is non-mathematical and 
non-abstract beyond IT as so defined is required to capture meaning. 

Yes, I would agree. Shannon-type information is yet meaningless. Information 
can only be provided with meaning by the substantive specification of a system 
of reference. For this reason, one needs not only a formal theory of the 
exchange, but also substantive theories. For example, a theory about the 
exchange of molecules in biology, and of atoms in chemistry, or of transactions 
in economy. These theories of specific communications cannot be expected to be 
unified because the substances (of “what is communicated and why”) are 
different. The formal theory of communication serves us, among other things, 
for moving from one substantive theory to another and for developing metaphors 
that can thus heuristically be transported, because of the abstraction 
involved. Additionally, these confrontations can lead to further developments 
of the algorithms that are relevant for studying the dynamics. 

The dynamics in the communication of meaning is different from the 
communication of information! Information can also circulate as noise (without 
meaning). I doubt it that meaning can be communicated without communication of 
information. Meaning is generated when information can be related by “an 
observing system” or more precisely in a discourse. It seems to me that 
semioticians focus exclusively on the communication of meaning without relating 
it to the communication of information. The latter, for example, has to confirm 
to the entropy law, while the former does not. The possibility of generating 
negative information has first been discussed by Brillouin as negentropy (- 
Delta H). 

Meaning circulation generates redundancies because the historical case is one 
of possible cases from the perspective of hindsight and thus the maximum 
entropy (of possible states) can be continuously enlarged. This is further 
reinforced when meanings are codified in terms of models. Models enable us to 
consider more possible case in the future. Such systems – e.g., scientific 
discourses – can be considered as strongly anticipatory. They act against the 
axis of time. 


3. Two aspects of the exchange between Koichiro and Loet merit attention: 1) 
Loet said that his point of replacing “why” with “what” did not seem necessary 
to him. In my mind, however, when Koichiro refers to “what is communicated by 
what”, he is insisting on not losing the qualitative components of the 
information involved. 

This seems confused to me. What is the qualitative aspect of “information”? (As 
a sideline: you did not answer any of my questions!) The qualitative aspect can 
only be the system of reference attributed to the information which provides 
the information with meaning. This system of reference is qualitative and 
therefore a qualitative and substantive theory of communication is then needed. 
This theory is different from the formal theory of communication. 

“By what” refers to the carriers of the information. Emphasis on the “why” 
instead provides focus on the theory about the dynamical system under study. 
For example: Why are molecules communicated in the autopoiesis of life? How is 
this different from the communication of atoms? And why? 

2) Loet seems to think that the role of time is covered by the following: “ 
Meaning is communicated incursively, whereas information is communicated 
recursively, that is, with reference to a previous state (t-1). Meaning is 
provided to the events from the perspective of hindsight, and with reference to 
other possible meanings (at t +1). ” This suggests a background framework and a 
world (or model of a world) limited to a state-transition concept of time, 
where, in addition, only Markovian processes occur. Koichiro envisages times 
that are closely related to or perhaps dependent on the actual communication 
processes in progress. CONCLUSION: Is there anyone in the group besides me who 
could say that both of these perspectives are necessary for a satisfactory IT? 

“satisfactory” is not an analytical category, but an emotional one. Do you wish 
to vote on these issues? 

Your “seems to think” is not such a nice formulation. Let’s keep the discussion 
rational! My example, indeed, was a (first-order) Markovian process, but it 
does not follow from the example, that “I seem to believe that only Markovian 
processes occur”. I have a preference to understanding first the simplest 
processes before moving towards the more complex non-Markovian ones because 
otherwise one may lose analytical clarity and empirical relevance. J It seems 
to me that the Markovian processes against the order of time are far from 
solved in the computation of anticipatory systems. We are only at the 

Best wishes, 


fis mailing list 
fis mailing list

Reply via email to