Notes on theories of information


---------------------
Robbie Fordyce
19 March 2024

---------------------

On reading notes from various theories of information.

---------------------
  1. Notes
  2. References
01. Notes

I went through a number of older computer science research papers where the authors are dealing with issues around trying to define information. There seems to be a general shift started in 1943 when Claude Shannon commences his work on rendering communication in a mathematical form with Warren Weaver. ‘Information’ is the term and there is no specific interest in an idea of ‘data’ in their work, and honestly much of it is mathematical and not especially relevant to HASS research. I guess there are obvious applications for cryptography (i.e. there are certain repeatable mathematical patterns in language that makes coded language decryptable), but nothing immediately relevant to HASS. This seems to be a point where information is first addressed as a form of thing with its own ontology.

Prior to this, according to Susan Artandi (1973) information apparently matters as a question of newness (i.e. learning something hitherto unknown) or as a matter of comprehension (i.e. did x make sense to the recipient). Shannon and Weaver’s mathematical model tries to make a generalisable idea of how we communicate vis a vis sender/noise/receiver. Artandi locates S+W as kicking off both a technologizing of communication through digital media, and a reciprocal need to comprehend information qua data due to digital information storage now being a thing. Artandi – consciously or not – seems to take a Piercean semiotics whereby semantics and syntactics are still important ways of locating the value of information relative to cognition and context. I think this is perhaps part of the debate we were having before about latency and objectivity. It is interesting here that at this point while there is a strong scientism in the discussion, I do not get the sense that the world itself is understood as informational in any sense prior to it being grasped by people or machines.

James Gleick (2011) writes in The Information, about a history of data that emerges out of Claude Shannon and Alan Turing during WWII which leads to the Turing tape and related developments, which I read as being something of a subtle trajectory between the idea of a universal computer (per Turing) and a world that is universally computable (i.e. the world can in total be understood as a thing of data to be dissected). The interesting thing here is that Gleick covers how Turing and Godel both pose pretty serious limitations to the computability of the world, and of mathematics itself.

Other related stuff that I found was a paper by Belkin and Robertson who come up with a typology of information based on its effects and concerns – I would argue that this account mistakes epistemology for ontology in some cases. Their typology is modestly hierarchical:

Genetic information which creates higher order structures, uncertainty (noise), perception (i.e. sense data), individual self-knowledge (i.e. concept construction), inter-human communication, social conceptual structures (I think we would recognise this as tacit institutions?), and formalized knowledge (which in their account seems to be a one-directional process that would be educational in a broad sense? In any case it would seem to be abstract information).

Many of the papers I found were pretty similar – mostly trying to give different cross-sections of what information/data does as if this answers what data is. Dretske is cited as noting that any definition of information needs to be at least somewhat related to the lay principle. It would make sense that any technical definition should not stray too from the conventional knowledge or else it would be breaching some sort of core value of ‘information’. I found a DJVU file of Dretske’s work, but it’s borderline unreadable because mathematical notation is frequent and in-line with the sections that attempt to define information. Perhaps a physical book is in order.

Robert Losee (1998) talks about information being variably valued in terms of perceiving or knowing. I think there’s something to this – perhaps because it signals a represented relationship to the world of ‘not data’ in the form of being perceived, and a crucial relationship to other data in the form of knowing.

I think based on this, the materiality of data doesn’t seem to be important, but the relationality of it is. The materiality is important for our ability to process data (i.e. we live in a material world, we encounter information in the , for the economics of data, for the problem of stewardship and responsibility, but it’s not necessary for the data itself to exist in a particular form (esp. w/r/t Indigenous Data Sovereignty). Data relates to both itself and to other things, and in this sense is relational, and I think there’s value in understanding data as an abstract thing that is ‘abstractable’ thing itself (hence metadata).

02. References
  • Artandi, S. (1973) Opinion Paper: Information Concepts and their Utility
  • Belkin, N. and Robertson, S. E. (1976) Information Science and the Phenomenon of Information
  • Gleick, J. (2011) The Information, A History, A Theory, A Flood.
  • Losee (1998) From Perceiving to Knowing. A Discipline Independent Definition of Information. <https://ils.unc.edu/~losee/b5/node8.html>