Re: [Fis] Fw: Responses
Dear colleagues, This discussion and reading the beautiful book of Bob Logan entitled What is information? (shortly forthcoming) made me go back to reading MacKay (1969) once more. I cannot find the distinction that makes a difference as it is quoted by Floridi (2005) -- and thereafter repeated by many -- so that I think that the honour goes to Bateson (1973) for a difference which makes a difference. MacKay, however, makes the point, for example, on p. 136 that the sentence S is a source of information is incomplete. It must always be completed (even if sometimes implicitly) in the form 'S is a source of information to receiver R'. Two sentences later he calls this significant information that must be capable of embodying and abiding by an agreed code or symbolic calculus. Elsewhere, he distinguishes this substantive concept of information from amounts of information that can be measured using (Shannon's) information theory. It seems to me that any discourse (physics, biology, psychology, sociology, etc.) can be further informed specifically in terms that are defined within and relevant to the specific discourse. This accords with the intuitive sense of information as meaningful information: meaningful for a discourse. Shannon's definition of information is counter-intuitive, but it provides us with a calculus that has major advantages. Katherine Hayles suggested that the two concepts can be compared with the discussion of whether a glass is half-full or half-empty. A Chinese colleague (Wu Yishan) once told me that in Chinese one has two words: sjin sji and tsin bao which correspond respectively to Shannon's and Bateson's definitions of information. A substantive definition of information (e.g., as a distinction that makes a difference for a receiver) requires the specification of the concept in a theory about the receiving system. This definition is therefore a priori system-specific; for example, for some of us this system is physics; for others it is biological discourse. At this level, one can again abstract from the substance and use Shannon's IT as entropy statistics. Sometimes, this allows us to explore the use of algorithms developed in one field (e.g., biology) in another (e.g., sociology). Concepts such as autopoiesis or auto-catalysis have carried these functions. For example, in the context of Ascendency Theory, Bob Ulanowicz showed how one can use the mutual information in three dimensions as an indicator of systemness. I use that as a systems indicator when operationalizing the triple helix of university-industry-government relations. Such translations of metaphors are always in need of further elaboration because the theoretical context changes and thus the specification of what the information means. However, the advantage to be able to measure in bits (nats or dits) frees us from the philosophical confusion about what information is. In my opinion, information can only be defined within a discourse. The mathematical definition of Shannon has specific functions which enable us to combine with different discourses (among which, specifically physics since S = k(B)*H). H, however, is dimensionless and defined as the expected information content of a message *before* it is received. It is yet to be provided with meaning. One could consider this meaninglesness as the specific difference of a mathematical concept of information. (Perhaps, it is easier to use uncertainty for this mathematical concept.) Best wishes, Loet -Original Message- From: fis-boun...@listas.unizar.es [mailto:fis-boun...@listas.unizar.es] On Behalf Of Robert E. Ulanowicz Sent: Tuesday, January 21, 2014 8:45 PM To: Christophe Cc: fis@listas.unizar.es Subject: Re: [Fis] Fw: Responses The reason of being of information, whatever its content or quantity, is to be used by an agent (biological or artificial). Dear Christophe, In making this restriction you are limiting the domain of information to communication and excluding all information that inheres in structure per-se. John Collier has called the latter manifestation enformation, and the calculus of IT is quite effective in quantifying its extent. Perhaps John would like to comment? Cheers, Bob U. ___ fis mailing list fis@listas.unizar.es https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis ___ fis mailing list fis@listas.unizar.es https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis
Re: [Fis] Fw: Responses
At 09:45 PM 2014-01-21, Robert E. Ulanowicz wrote: The reason of being of information, whatever its content or quantity, is to be used by an agent (biological or artificial). Dear Christophe, In making this restriction you are limiting the domain of information to communication and excluding all information that inheres in structure per-se. John Collier has called the latter manifestation enformation, and the calculus of IT is quite effective in quantifying its extent. Perhaps John would like to comment? I developed this concept in order to reply to Jeff Wicken's complaint that Brooks and Wiley did not distinguish properly between the complement of entropy and structural information, but I used it in print to discuss, in the context of cognitive science and especially John Perry's use of information (see Barwise and Perry Situations and Attitudes and his What is information?, as well as Dretske's book on information and perception) what the world must be like in order to make sense of information coming from the world into our brains. The article can be found at Intrinsic Information (1990) In P. P. Hanson (ed) Information, Language and Cognition: Vancouver Studies in Cognitive Science, Vol. 1 (originally University of British Columbia Press, now Oxford University Press, 1990): 390-409. Details about information are there, but the gist of it is that can be measured, is unique, and depends on time scale to distinguish it from informational entropy in information systems. The uniqueness hypothesis was developed very carefully in my former student, Scott Muller's PhD thesis, published as Asymmetry: The Foundation of Information (The Frontiers Collection) by Springer in 2007. I am rather busy now at a conference, or else I would say more here. John Professor John Collier colli...@ukzn.ac.za Philosophy and Ethics, University of KwaZulu-Natal, Durban 4041 South Africa T: +27 (31) 260 3248 / 260 2292 F: +27 (31) 260 3031 Http://web.ncf.ca/collier ___ fis mailing list fis@listas.unizar.es https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis
Re: [Fis] Fw: Responses
Dear Bob U, If your are talking about resident information, as available for usage, I take it as being part of information that can be used by the agent. Let me go through John's paper (thanks John). Best Christophe Date: Tue, 21 Jan 2014 14:45:15 -0500 Subject: Re: [Fis] Fw: Responses From: u...@umces.edu To: christophe.men...@hotmail.fr CC: lo...@physics.utoronto.ca; fis@listas.unizar.es The reason of being of information, whatever its content or quantity, is to be used by an agent (biological or artificial). Dear Christophe, In making this restriction you are limiting the domain of information to communication and excluding all information that inheres in structure per-se. John Collier has called the latter manifestation enformation, and the calculus of IT is quite effective in quantifying its extent. Perhaps John would like to comment? Cheers, Bob U. ___ fis mailing list fis@listas.unizar.es https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis
Re: [Fis] Fw: Responses
The reason of being of information, whatever its content or quantity, is to be used by an agent (biological or artificial). Dear Christophe, In making this restriction you are limiting the domain of information to communication and excluding all information that inheres in structure per-se. John Collier has called the latter manifestation enformation, and the calculus of IT is quite effective in quantifying its extent. Perhaps John would like to comment? Cheers, Bob U. ___ fis mailing list fis@listas.unizar.es https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis
Re: [Fis] Fw: Responses
Hi Christophe - I enjoyed your response - full of meaningful information - :-) Your point is well taken. I agree what might be meaningful information for one agent might be meaningless for another. I can add another example to your list of examples which I encountered some time ago. An author whose name I forget pointed out that a book written in Urdu is information for a literate Urdu speaker but perhaps not for those that cannot read Urdu. According to the definitions of Doug MacKay in 1969 and Gregory Bateson in 1973 'information is a distinction that makes a difference' and 'information is a difference that makes a difference' respectively. Meaningless information does not cut it by their definitions as it does not make a difference. Of course one could define information as a distinction or a difference that has the potential to make a difference for some agent. It seems to me that defining what is information is no easy task. My conclusion from our discussion is that depending on how you define information context can be an important part of what is information. The notion of information is extremely nuanced with multiple meanings and we seem to have only one word for it as pointed out by Shannon himself. In the abstract to his paper, The Lattice Theory of Information Shannon (1953) wrote, The word information has been given many different meanings by various writers in the general field of information theory. It is likely that at least a number of these will prove sufficiently useful in certain applications to deserve further study and permanent recognition. It is hardly to be expected that a single concept of information would satisfactorily account for the numerous possible applications of this general field. The present note outlines a new approach to information theory, which is aimed specifically at the analysis of certain communication problems in which there exist a number of information sources simultaneously in operation. MacKay made a distinction between 'selective information' as defined by Shannon's formula and 'structural information', which indicates how 'selective information' is to be interpreted. Structural information must involve semantics and meaning if it is to succeed in its role of interpreting selective or Shannon information. Structural information is concerned with the effect and impact of the information on the mind of the receiver and hence is reflexive. Structural information has a relationship to pragmatics as well as semantics where pragmatics tries to bridge the explanatory gap between the literal meaning of a sentence and the meaning that the speaker or writer intended. Shannon information has no particular relation to either semantics or pragmatics. It is only concerned with the text of a message and not the intentions of the sender or the possible interpretations of the receiver (Logan 2014). The above material is from my book What is Information? to be published simultaneously as a printed book and an e-book by DEMO press in Toronto. I would be happy to share this with you as a PDF, Christophe or any other reader that finds the above information meaningful and interesting. Thank you Christophe for providing me with the opportunity to muse so more about the meaning of information especially 'meaningless information'. I quite enjoy going down the rabbit hole. all the best Bob __ Robert K. Logan Prof. Emeritus - Physics - U. of Toronto Chief Scientist - sLab at OCAD http://utoronto.academia.edu/RobertKLogan www.physics.utoronto.ca/Members/logan On 2014-01-16, at 5:52 AM, Christophe wrote: Dear Bob, Thanks for your answer. So for you, information is always meaningful. Such a statement is surprising when many examples can display cases of meaningless information. The well known Chinese Room Argument: a sentence written in Chinese is meaningless to a non Chinese speaking reader. A vervet monkey alarm is meaningful information for other vervet monkeys but is meaningless for a passing by dog. Cyphered information is meaningful or meaningless depending if the receiver has the cyphering key or not. We can agree that the meaning of information does not exist by itself but is a result of an interpretation by a system. The interpretations of given information can deliver different meanings and no meaning is a possible outcome. Looking at downgrading to signals the intrerpretation of information is surprising. Now, regarding information theory, I still understand it in a scientific background as it is part of mathematics and computing. Things can be different indeed if you look at applications of IT (to linguistics, psychology, …). But the key point may be that disregarding the possibility for meaningless information shuts a road in analyzing the possibilities for computers to understand us. A lot is still to be done in this area and my take is that using the
Re: [Fis] FW: Responses
Dear Christophe, I tried to qualify my use of meaning, but perhaps I wasn't clear enough. In my example I wanted to say that I(A;B) is a quantity that can be considered a proto-meaning of B to A. Another way of saying the same thing is that I(A;B) quantifies A in the context of B. I should have added that I don't confine my notion of information to the scenario of communication. I feel that it's discovery in that context was an historical accident. Rather, like Stan, I consider information more generally as constraint, and the information of communication becomes a subset of the more general attribute. Hence, anything that is held together by constraints is amenable in one form or another to the Bayesian forms of Shannon capacity. The real advantage in doing so is that the complement of the Bayesian information is made explicit. Vis-a-vis constraint this complement becomes flexibility. Such flexibility is an apophasis that is missing from most scientific endeavors, but is essential to our understanding of evolution. You are probably correct that my terminology is not orthodox and possibly confusing to some. But I see such inconvenience as a small price to pay for opening up a new window on quantitative evolutionary theory. I really want folks to think outside the box of communication theory. What Shannon started many (such as Terry Deacon) have prematurely cast aside. My message is that we need to re-evaluate Shannon-type measures in their Bayesian contexts. The have the potential of becoming very powerful quantitative tools. (E.g., http://www.cbl.umces.edu/~ulan/pubs/EyesOpen.pdf.) Peace, Bob Bob, You seem to implicitly consider all information as being meaningful. I'm afraid such a position is source of confusion for a scientific approach to information. As we know, Shannon defined a quantity of information to measure the capacity of the channel carrying it. He did not address the meaning that the information may carry (as you write in your paper Shannon information considers the amount of information, nominally in bits, but is devoid of semantics ). Today DP Telecom activities use Shannon type information without referring to its meaning. Considering information as being meaningful looks to me as potentially misleading and source of misunderstandings in our discussions. Information can be meaningful or meaningless. The meaning comes from the system that manages the information. Some research activities explicitly consider information as meaningful.data (see http://www.mdpi.org/entropy/papers/e5020125.pdf). I'm afraid such a position creates some vocabulary problems if we want to keep a scientific background when trying to understand what is meaning. The meaning of information is not something that exists by itself. The meaning of information is related to the system that manages the information (creates it or uses it). And the subject has to be addressed explicitly as systems can be animals, humans or machines. Focusing on meaning generation by a system can bring some clarification (see http://www.mdpi.org/entropy/papers/e5020193.pdf). Hope this helps Christophe From: lo...@physics.utoronto.ca Date: Sat, 11 Jan 2014 10:49:28 -0500 To: gordana.dodig-crnko...@mdh.se CC: fis@listas.unizar.es Subject: Re: [Fis] Fw: Responses Dear Friends - I have been lurking as far as the discussion of Shannon information is concerned. I must confess I am a bit confused by the use of the term information associated with what you folks call Shannon information. To my way of thinking Shannon produced a theory of signals and not information. Signals sometimes contain information and sometimes they do not which is exactly what Shannon said of his notion of information. Here is what troubles me: A set of random numbers according to Shannon has more information than a structured set of numbers like the set of even numbers. For me a random set of numbers contains no information. Now I am sure you will agree with me that DNA contains information and certainly more information than a soup of random organic chemicals which seems to contradict Shannon's definition of information. I would appreciate what any of you would make of this argument of mine. Here is another thought about information that I would love to have some comments on. The information in this email post to you folks will appear on multiple computers, it might be converted into ink on paper, it might be read aloud. The information is not tied to any particular physical medium. But my DNA cannot be emailed, printed out or in anyway separated from the physical medium in which it is instantiated. As far as I know it has been transferred in part to my 4 kids and 4 grandkids so far and there it stops for time being at least. The information in my DNA cannot be separated from the medium in which it is instantiated. This information is not symbolic. DNA is not a symbol of RNA
[Fis] Fw: Responses
Dear Hans and All, This is a very useful form of responses, which enables further directed comments. I start with Lars', which is perhaps as Hans says crucial: Lars -- How does QBism differ from Copenhagen? This is a crucial question. It differs not at all in the formalism, and only subtly in the interpretation. Many users of quantum mechanics regard the wavefunction as a real property of an electron. They talk about the wavefunction in the same way you might say the speed of the car. They must then deal with perennial problems such as action-at-a-distance and the collapse of the wavefunction. QBists regard the wavefunction the way Bruno de Finetti regarded probability, when he wrote, in caps, PROBABILITY DOES NOT EXIST. I think he meant that the probability of a coin falling heads is not a measurable property of a coin. All it is is a personal belief of how much an agent should bet. And that belief changes instantly and locally when you make a measurement, or hear that someone else has made one. Joseph (New) -- One does not learn much about the way things are by reference to simple, binary phenomena (coins) or more complicated versions in game theory (profit-loss). All these have little to do with processes, such as information, which embody complex oscillations between presence and absence, non-meaning and meaning, and so on. Some people call the the Copenhagen wavefunction ontic, the QBist one epistemic. Joseph (New) -- This is an extremely important statement by Hans whose consequences, IMHO, should be discussed as they relate to information. We all agree that one cannot surf on ontic Copenhagen waves, while we 'know what we know'. But ontic positions, when some of the deficiencies of the original Copenhagen interpretation are corrected, have a lot to say about /how/ we know, what the properties of what we know are, and how the two are interrelated. Rather than saddle QBism with an ignorantist position, I would like to see it expand to include this relation. Gordana -- I am out of my depth in a discussion of phenomena/noumena/Dinge-an-sich. But when I agree that the Higgs exists out there in the world, I am sure it's not an object like a marble, but a symbol for a collection of experiences that many people have had, and have discussed, and codified, so that if they perform another experiment where it might play a role, they can be prepared with betting odds for what they might experience next. Joseph (New) - - That it is a collection of experiences does not exclude that it is an object, or better process, of a kind other than a 'marble'. As such, in discussing it, we can go beyond binary game metaphors. Joseph -- the electron is a point means that no experiment to date has found evidence for a finite size. In the theory (quantum electrodynamics) there is no room for any parameter with dimensions of length, although there are mass, charge, spin, and magnetic moment. When you introduce a finite size into the theory, it makes wrong predictions. (This is not true for protons, for example.) Joseph (New) -- Same as above. The fact that an electron does not have a 'finite size (diameter)' does not mean that is does not exist objectively. It is fuzzy with a 'size' greater than the Planck length and less than that of a hadron. The gravitational field lives in 3D was not supposed to deny that Einstein's elegant formulation treats time as a fourth dimension. But a quantum field is an altogether different and much more complicated beast which lives in infinite dimensions, and has no analog whatever in our everyday human world. Joseph (New) -- In my opinion, 1) current theories of gravity add more to Einstein's original number of four dimensions to the gravitational field; 2) current quantum theories do not saddle the quantum field with a mathematical infinity. There are no such infinities in nature. That there is no analog of the quantum field at the macroscopic level does not mean that there are no isomorphisms between levels. One aspect is that of the couple duality - self-duality as I mentioned earlier. Having a proper view of physics among the many possible is critical to placing information theory on a sound basis. I have proposed Logic in Reality as one way of giving meaning to the statement that energy and information processes are non-separably related and how they are related. Are there others? Thank you and best regards, Joseph___ fis mailing list fis@listas.unizar.es https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis
Re: [Fis] Fw: Responses
Having a proper view of physics among the many possible is critical to placing information theory on a sound basis. I have proposed Logic in Reality as one way of giving meaning to the statement that energy and information processes are non-separably related and how they are related. Are there others? Dear Joseph, It seems to me that there is at least one alternative: Shannon's mathematical theory of information. Information is then defined as content-free. Thermodynamic entropy (physics) is the special case that H is multiplied by the Boltzmann constant and thus one obtains the dimensionality of S. (S = kB * H). Information theory, however, can also be used in other contexts such as economics (Theil, 1972). It does not have a realistic interpretation such as in your argument. From such a more differentiated perspective, concepts (e.g., electrons) do not exist, but are meaningful and codified within theories. There can be sufficient evidence (in physical theorizing) to assume (for the time being) that the external referents (electrons) exist. The logic is not in reality, but in the argument, and one cannot jump to the (ontic) conclusion of existence. Thus, perhaps the sentence we all agree . (with you?) is a bit premature. Best, Loet ___ fis mailing list fis@listas.unizar.es https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis
