Information is Provisional

My paper arising from DTMD 2013 has been published in the Difference that Makes a Difference Special Issue of Kybernetes.

Here's the abstract:

Title – Information is Provisional.

Purpose – The purpose of this paper is to contribute to a transdisciplinary understanding of the nature of information.

Design/methodology/approach – The work draws from previous work done on information in several disciplines and contexts, and proposes a new framework for describing and understanding information. It applies the framework to several different situations involving information, explores the insights revealed by the use of the framework and discusses some of the implications.

Findings – Information is usefully described as a situated event that extracts or generates meaning, with the distinction between extracting or generating meaning corresponding to the distinction between semantic and environmental information. A diagrammatic convention based on communications theory and making use of hierarchies of levels is found to provide a powerful means of conveying many of the aspects of the nature of information, and of understanding the role of information in a wide range of applications. An additional specific finding of the work is that information is inherently provisional.

Originality/value – The diagrammatic framework is a new way of presenting, describing and understanding information, and the suggestion that information is inherently provisional is believed to be new.

David Chapman, (2014) "Information is provisional", Kybernetes, Vol. 43 Iss: 6, pp.895 - 910 DOI: 10.1108/K-08-2013-0178

(There is also an un-edited preprint of it on the Open University's ORO repository, available from my entries on ORO.)

Information in school reports: the layered communication model

Further to my post last week, let's model the communication from teacher to parent, considering just a single subject for simplicity.

The teacher know lots about John, but inputs into the software just the name ("John", which is the address of the message), and the selection from one of fours levels, choosing in this case "m" (which is the message payload).

The software delivers a full report to the parents:

John has extended his knowledge of a variety of computer programs and he can log into the network without support.  He has explored a variety of features included in software for composing music and is aware that questions can be turned into search criteria when using data handling programs.  He has found information relating to his topic work from given websites on the worldwide web and explains patterns that govern a computer simulation

This is how I see the communication using the layered model and trapeziums to represent the coding and decoding (see my chapter in Ramage and Chapman or read the posts labelled trapezium in this blog).

The binary coding at the bottom isn't really relevant to this discussion (and could even introduce confusion between information bits and binary digits) but I think helps as a reminder of the way layers work.  So the dashed lines show the virtual communication, and we can see the virtual communication of an m which is actually communicated by a binary representation. (Further confusion here because of course the Report King software will actually be working with binary representations and the m will presumably be ascii coded somewhere - but this isn't relevant to the model.)

The interesting bit, of course, is the layer above. The pretence is that the whole paragraph that the parent reads has been communicated virtually from the teacher. The reality is very different. The teacher - hopefully - does know lots about John and could communicate a full paragraph (and much more), but actually they have encoded their entire knowledge of John into the choice of 'm' (that John is an "average to more able achiever").  It was Report King that generated the paragraph.

A key point is the mismatch between the encoding and the decoding. The binary decoding just reverses the binary encoding, but the teacher encoding is different from the Report King decoding. Not entirely different of course, because the school teaches to the National Curriculum which the Report King uses the National Curriculum to do the decoding - that's why the Report King system works at all.

Apart from the message about information inflation, which is where this discussion started in last week's post, this model draws attention to the role of the encoding and decoding in communication. For example, decoding can never be perfect (even the binary decoding might get it wrong), and since they happen at different times and/or different places in some senses they are always completely independent? That's for further exploration at another time.

Information inflation: school reports

It is said there's more information around today than ever before. It is taken as a given fact, though people argue over whether it is a good thing or not (information riches or information glut?). But is it necessarily true?

A few years back, before the financial downturn, people in the UK would say ‘there’s more money around’ to account for the fact that people seemed to be consuming and acquiring goods more than ever before.  You could have used the same expression: ‘there’s more money around’, about the Weimar Republic, in the 1920s & 30,  but it would have meant something entirely different

Stamps from the Weimar Republic, showing inflation (from left to right: 1923, 1926 & 1928)

So, is the apparent increase in information real or hyperinflation?

With money you can make some meaningful comparisons - I was once told that the price of a Mars Bar is a good benchmark - but how can you do that for information?

I've been pondering this for a while (and had a paper on the topic rejected!). Here's some evidence for information inflation.

School reports, when I was a child, consisted of one or two hand-written sentences on each subject.  The reports that my children have brought home are much more substantial: a printed booklet with paragraphs of, typically, between 50 and 200 words on each subject. At face value, I’m getting a lot more information about my children than my parents got about me.

School reports today are frequently put together with the aid of specialised software that reduces the workload on the teacher.  An extreme example of this is The Report King that can be used in England.  Using this software, all that a teacher needs to input is the name of the pupils and their grades for each subject, selected from a pre-defined set such as: h for higher achiever, m for average to more able achiever, l for average to less able achiever and sen for students with special educational needs.  The software then writes the whole report, drawing on the statements contained in the National Curriculum for England. For example,  entering m for ICT for one pupil (John), generated:

John has extended his knowledge of a variety of computer programs and he can log into the network without support.  He has explored a variety of features included in software for composing music and is aware that questions can be turned into search criteria when using data handling programs.  He has found information relating to his topic work from given websites on the worldwide web and explains patterns that govern a computer simulation

Entering the same grade for another pupil would generate a similar but different paragraph, because the software makes use of different wording and draws on different parts of the curriculum to ensure two different pupils don’t get the same report.

In terms of the Shannon model of communication, the message from the teacher is entirely specified by the name of the pupil and the grade.  Since the grade was a selection of one from four, the information content (assuming each of the four grades was equally probable) about the pupil’s performance is 2 bits.  That whole paragraph (73 words) is a symbol for the message m.

Comparing this to the hand-written sentence of times past, it is easy to see that a written sentence is likely to contain a lot more than two bits of information. Even the most harried teacher is likely to be selecting their sentence from a lot more than four possibilities.  For really effective communication, though, the face-to-face meeting at the parents evening is still, as it always was, better than the report, whether hand-crafted or computer generated.

There’s an argument that there is more information in the computer-generated report, in this case information about the National Curriculum, but it’s not about John and it’s not from the teacher.

In conclusion, just as to say 'there's more money around' due to hyperinflation is a misuse of the word 'money', so too - maybe - to say there is more information around today is a misuse of the word information.