Waking up rather early this Monday morning, I continued to lie in bed, hoping but not particularly expecting to get back to sleep.
Lying in this way, in the dark, in the relative quiet of early morning, one is conscious of all kinds of noises. Noises arising from oneself, perhaps the blood rushing around in the ears. Perhaps a spot of wheezing if one is getting over a cold. Noises from the house around. Perhaps the water in the central heating pipes is starting to move around. And then there are noises from the world outside. Perhaps the wind and the rain. Perhaps the odd car leaving for work or the odd taxi delivering someone from something or other. Perhaps the odd train or aeroplane. Perhaps the traffic from the M25, from which the noise will carry to Epsom when conditions are right.
This morning, I thought I heard the hoot of a hunting owl. Not the ‘too-wit too-woo’ of children’s stories, of reference 1, rather just hoots. Usually, a pair of hoots a few seconds apart. Then perhaps another pair after quite a few seconds, or perhaps several minutes.
It took a while to be sure that this was an owl, rather some extraneous noise arising from me. Or just some random noise from anywhere, not the result of anything in particular.
The answer seemed lie in the repetition. When the hoots were indeed paired, one could use the first hoot to tune in. To suspend breathing temporarily. To lift the ear slightly from the pillow to reduce the noise coming from the ear itself. Then one was quiet and ready for the second hoot. No doubt about that one.
Which I believe fits in with what information theorists have to say about the difference between noise and data. For there to be useful data, there has to be redundancy in the signal, otherwise it might just as well be noise, white noise or otherwise. Perhaps I am thinking of the introductory material on randomness to be found at the beginning of reference 2: a signal is random if it cannot be compressed. With one example given being the decimal expansion of π. It might look pretty random, it might pass various tests of randomness, but it can be generated by a relatively simple computer program. The never-ending sequence of decimal digits can be massively compressed – and so is not random.
A variation is the use of check digits and such in data filed on computers. So, for example, when one has a long reference number like a VAT number, the last digit might be the units part of the product of the other digits, thus providing a degree of protection against clerical error, such an error then being likely to result in an ‘invalid’ number. Or a long file might be broken into records, with each record starting with a record length and ending with an end of record marker. If the record length does not take you to such a marker, something has gone wrong. The amount of this sort of thing can be geared to the amount of noise that one expects to find in the data.
An everyday example of this would be the habit, when telling someone of a date, to say something like 'Tuesday the 15th'. Checking that the 15th of the month in question is indeed a Tuesday is very much like checking the check digit on a reference number.
PS: I should add that I never got beyond the beginning of reference 2. Looked interesting, but too deep for me. Also that I have not bothered to check whether VAT numbers do indeed include check digits.
References
Reference 1: https://www.wildlifetrusts.org/wildlife-explorer/birds/birds-prey/tawny-owl.
Reference 2: An introduction to Kolmogorov Complexity and its Applications – Ming Li, Paul Vitányi – 2008.
No comments:
Post a Comment