Thursday, December 1, 2011

Meaning and information

I don't remember meeting him, but, when I was a child, we often used to drive past a house belonging to my father's old colleague, 'Sticky' Glew. (My mother had been introduced to him, and had trouble deciding how to address him. She wisely opted for 'Mr Glew'.) It amused my father that his friend had written a dissertation on 'the meaning of meaning'.

"For thousands of years," writes Seth Lloyd (in his book, Programming the universe (Vintage, 2007)), "philosophers have tried to determine what 'meaning' means, with mixed success." (p. 24)

Lloyd's work (in quantum computing and quantum communication systems) is based around the technical concept of information. But strings of bits and so on mean nothing in themselves: they are meaningful only if they can be interpreted. "Meaning is defined only relative to a scheme of interpretation ..." (p. 25)

"Consider the string of bits ... : 1001001 1101110 0100000 1110100 1101000 1100101 0100000 1100010 1100101 1100111 1101001 1101110 1101110 1101001 1101110 1100111. Interpreted as a message encoded in ASCII, this string means 'In the beginning'. But taken on its own, with no specification of how it is to be interpreted, it means nothing other than itself." (p. 25) And, of course, 'In the beginning ...' is interpreted according to the conventions of the English language. As Lloyd points out, natural languages are rich in ambiguity, which is "a key aspect of poetry, fiction, flirting, and plain everyday conversation." (p. 27)

But sometimes, in order to understand basic concepts - like meaning - it is useful to strip away complexities and ambiguities and look at simple models, as Wittgenstein did in his account of language games. Imagine a simple language game in which a builder says "Block" and the assistant hands him a block, or "Slab" and the assistant hands him a slab. The meaning of each expression is to be found in the action the expression provokes.

Lloyd relates Wittgenstein's idea to computers and computing. "The meaning of a computer program written in a particular computer language," he writes, "is to be found in the actions the computer performs as it interprets that program. All the computer is doing is performing sequences of elementary logic operations, such as AND, NOT and COPY ... The computer program unambiguously instructs the computer to perform a particular sequence of those operations. The 'meaning' of a computer program is thus universal, in the sense that two computers following the same set of instructions will perform the same set of information-processing operations and obtain the same result." (p. 26-27)

Meaning, then, is interpreted information. We don't need a theory of meaning such as philosophers have attempted to build. Philosophy (most notably the philosophy of language and metaphysics) has drifted into unproductive areas reminiscent of scholasticism in which intellectual work is done without first ensuring that there is an important intellectual task to address.

Of course, the word 'meaning', like many English words, can be used in different ways, and a careful analysis of the context of use will reveal subtle - and not so subtle - differences. One sense of the word - rather different from most of the others is 'general significance' or 'point' or 'purpose', as in the sentence, 'My life seems to lack meaning.' I have not been talking here of this generalized kind of meaning, but of the ordinary - and primary - uses of the word in relation to information and communication. And, understood as interpreted information, it seems to me a perfectly clear and unmysterious concept. Trying - as some philosophers do - to make a big issue of what they call 'aboutness' or 'intentionality' (which doesn't mean what non-philosophers think it means) constitutes unnecessary mystification.*

The remarkable advances in computing and information and communication technologies during the last 70 years have thrown up many real problems of a fundamental nature, but they require scientific knowledge (not just knowledge of formal logic and philosophy) to address. In particular, the parallels between thermodynamics and information theory are clearly rich in new problems which would benefit from informed reflection. For example, all matter and energy is subject to the laws of thermodynamics, but all matter and energy - everything there is - is also subject to the laws of information. Information theory appears to be a more fundamental and all-encompassing theory than thermodynamics (which can now be seen as just a special case of information theory).

Though I doubt the value of much recent philosophical work in the areas of language and meaning, some very important foundational work has certainly been done by philosophers, philosophically-inclined mathematicians and logicians. The key advances were made during the 19th and early 20th centuries. The work of George Boole, Giuseppe Peano, David Hilbert, Bertrand Russell, A.N. Whitehead and many others provided the conceptual tools which made the development of the electronic computer and sophisticated communication technologies possible. Claude Shannon's development of what he called a 'mathematical theory of communication' and which has come to be known as information theory was based largely on the work of George Boole.

Information has come to be seen as physical - it is no longer seen as abstract, disembodied and unquantifiable. It is always tied to a physical representation: a mark on paper, a charge, a spin, a sequence of bases in DNA. Information processing occurs not just in computers and brains but throughout the physical world.

In 1961 Rolf Landauer came up with a principle with (it is said) startling implications: that one essential information processing operation (erasure) cannot occur without causing heat to dissipate into the environment (thus increasing the entropy of the universe). The processing of information is a thermodynamic process (just as thermodynamic processes are informational) and erasure is an irreversible operation. Negation can be reversed by a second negation. Addition can be reversed by subtraction. But erasure cannot be undone.

I will resist the temptation to discuss the implications of Landauer's principle (which I believe are not good!). I am a layman in these matters, but the role that information processing seems to play in every aspect of nature intrigues me. I am trying at least to understand the fundamental principles, and to relate the sort of thing one learns within the context of philosophical logic - e.g. that there are any number of alternative systems of logic - to the apparently more constrained context of real-world information processing.


* Much philosophical work in recent decades in areas such as ethics, metaphysics, the philosophy of language and even logic has been done by people with a commitment to a religious view of the world (or at least with an anti-physicalist orientation), and has been motivated (I believe) by an attempt to undermine physicalism and to save a space for the spiritual (broadly interpreted). Such a motivation does not invalidate the work, but I personally don't think a convincing case against physicalism has been made.