Just been doing some extensive background reading into information theory, enough to know that Dr. Gitt's article has almost nothing to do with Shannon, despite several references.
For those who either have not read the article or do not understand it, I shall provide a summary here.
Gitt begins by discussing Shannon's Information Theory, and in the process demonstrates his complete failure to understand both its content and its intent. He complains that:
(I)nformation according to Shannon does not relate to the qualitative nature of the data, but confines itself to one particular aspect that is of special significance for its technological transmission and storage. Shannon completely ignores whether a text is meaningful, comprehensible, correct, incorrect or meaningless.
Why would Shannon have worried about that? Information theory deals with the transmission and optimal compression of data, not its actual meaning. (For those looking to learn more about information theory, there is the Wikipedia article
, or, for those of a less statistical bent, this site
, which rather amusingly has managed to spell "information" wrong in the domain name, and again in a different way on the home page.) However, Gitt ignores this, even stopping so low as to attempt to ridicule Shannon's method, but only humiliating himself in the process.
The section that follows is about information density, and it is entirely unclear in what way it is relevant to Gitt's point. However, it is notable for one section:
Shannon’s information theory makes it possible to determine the smallest number of letters that must be combined to form a word in order to allow unambiguous identification of all amino acids. With 20 amino acids, the average information content is 4.32 bits/amino acid. If words are made up of two letters (doublets), with 4 bits/word, these contain too little information. Quartets would have 8 bits/word and would be too complex. According to information theory, words of three letters (triplets) having 6 bits/word are sufficient and are therefore the most economical method of coding.
Another failure to understand information theory, which is irrelevant to the calculation of optimum "letters" to code for amino acids. Simple mathematics does that.
He then examines various ways of storing information, again with no obvious purpose. How are either of these sections relevant to the paper? So far, it appears to be a mixture of misunderstanding and irrelevancy. However, in his next section, Gitt brings in a new factor: illogicality.
Gitt defines a new system for interpreting information, based around five levels:Statistics
The first is an attempt to merge his misunderstanding of information theory into the system, while also relegating it to the lowest level. He fails to realise that the statistical analysis of information has absolutely nothing to do the information itself.
At this point, I should point out that throughout the article Gould proposes a number of "theorems", which are mostly rewordings of what he has already said, although in some cases they are completely unrelated. More accurate descriptions of them would be variously "truism", "irrelevancy" and "non-sequitor". It does not appear that Dr. Gitt is aware of what a theorem is.
The next level (syntax) basically states that all information needs a way of interpreting it, called a "code". Here are the "theorems" that Dr. Gitt draws from his logic:
Theorem 4: A code is an absolutely necessary condition for the representation of information.
Theorem 5: The assignment of the symbol set is based on convention and constitutes a mental process.
Theorem 6: Once the code has been freely defined by convention, this definition must be strictly observed thereafter.
Theorem 7: The code used must be known both to the transmitter and receiver if the information is to be understood.
Theorem 8: Only those structures that are based on a code can represent information (because of Theorem 4). This is a necessary, but still inadequate, condition for the existence of information.
Let's examine these further:
4) Truism. If you have information, of course you need a way of interpreting it.
5) Non-sequitor. That "mental process" thing came out of nowhere, with not a scrap of evidence for it. He doesn't even define what he means by a "mental process" (as in, what level of mental activity is required). His only support for the "consciously established conventions" part is language, both completely ignoring other forms of expression of information and a little hazy in itself, being that no one sat down to define languages. They come together gradually; one could even say they evolve.
6) Not true. You can use any number of codes for any number of sequences of information. If I am talking to someone, we can switch between English and French, for example, providing that we both know both.
8) Equivalent to 4.
His next level of information is "semantics", i.e. information must carry meaning. Again, a truism. How else would you define information? However, in addition to stating the blindingly obvious, Gitt makes a rather interesting diversion from ordinary logic:
The Dortmund information scientist Werner Strombach emphasises the non-material nature of information when he defines it as ‘an appearance of order at the level of reflective consciousness.’ Semantic information, therefore, defies a mechanistic approach. Accordingly, a computer is only ‘a syntactical device’ (Zemanek) which knows no semantic categories. Consequently, we must distinguish between data and knowledge, between algorithmically conditioned branches in a programme and deliberate decisions, between comparative extraction and association, between determination of values and understanding of meanings, between formal processes in a decision tree and individual selection, between consequences of operations in a computer and creative thought processes, between accumulation of data and learning processes. A computer can do the former; this is where its strengths, its application areas, but also its limits lie. Meanings always represent mental concepts; we can therefore further state:
Theorem 10: Each item of information needs, if it is traced back to the beginning of the transmission chain, a mental source (transmitter).
The problem here is that Gitt makes no distinction between recognising information and creating it. One can reword from "Information is the appearance of order at the level of reflective consciousness" to "Information is that which appears ordered to a reflective consciousness" with no change of meaning. If we take the second statement (equivalent to the first), Gitt's logic seems farcical.
As an aside, we should point out that "A computer can [only] do the former" should in fact read "A computer can at present
only do the former".
Following on from this, theorem 10 is clearly a non-sequitor. It should be mentioned that this does not technically prove that it is incorrect, only that it has no basis in logic. However, it is easy to prove incorrect with a simple counter-example.
The absorption lines in the light spectra of distant stars and galaxies contain information about how fast they are moving (the so-called "red shift"). This is undeniably information, and undeniably does not come from a mental source.
Gitt then makes some irrelevant statements about languages - it is not enough to focus on one type of information. This kind of logic requires abstraction. Otherwise all he is doing is proving a point about languages.
The next two stages are pragmatics and apobetics - i.e. the result of the conveyance of information, and its intent. However, it should be stressed that these terms ONLY APPLY TO LANGUAGE. The red shift example above is one counter-example, E=mc^2 (to pick an example at random) is another. Neither of these have result or purpose, yet both are clearly information.
Gitt then brings in more irrelevancy regarding information in living beings. He finishes by proposing some "laws" about information and tying the whole thing into humanity.
Here are those laws again (so you don't have to scroll up):
1.No information can exist without a code.
2.No code can exist without a free and deliberate convention.
3.No information can exist without the five hierarchical levels: statistics, syntax, semantics, pragmatics and apobetics.
4.No information can exist in purely statistical processes.
5.No information can exist without a transmitter.
6.No information chain can exist without a mental origin.
7.No information can exist without an initial mental source; that is, information is, by its nature, a mental and not a material quantity.
8.No information can exist without a will.
2) He doesn't specify what he means by "free" and "deliberate" is highly questionable.
3) Plain wrong. The first, fourth and fifth are not required.
4) What? Another failed attempt to tie in with Shannon.
5) Truism, ignoring Gitt's personal definition of "transmitter". This is addressed above.
6) Wrong again. See above.
7) Equivalent to 6.
8) What? A complete non-sequitor, unrelated to anything that has gone before. He should also define his use of "will".
Hope that clears things up.