Showing posts with label Randomness. Show all posts
Showing posts with label Randomness. Show all posts

Thursday 28 June 2012

Thatlessness, Redundancy, Probability and Glitchiness.

A lawyer, a specialist in Shipping and Transport, wrote to me recently, pointing out there are  precedents in Marine Law being cited and discussed in relation to the case of the Costa Concordia tragedy. The bulletin was littered with redundant that’s, which obscured the legalese and what turned out to be an extraordinarily subtle argument.

Here, I have taken the liberty of editing the main argument by reducing six that’s to three.

It was argued by the First Claimant in the ‘Saint Jacques II’ that, as a matter of logic, the more times the First Claimant had navigated contrary to the separation scheme, the less there was any or real prospect of inferring that he had actual knowledge when he set the course on the day in question that a collision would probably result.

So here we see, ‘as a matter of logic’, a probability argument deployed in Maritime Law.  In this case, the more an incorrect thing is done without error the greater the expectation that similar illegal manoeuvres will be free of accident.

In practically the same post came intelligence of the development of randomised software algorithms whereby inexactness, if introduced into an ‘adder’ electronic chip, can yield computationally faster, more effective operations that are cheaper, because they are more energy-efficient.

What does the fate of the Costa Concordia have to do with Probability Theory applied to an electronic chip?

As the electronics guru makes clear, inexactness using probabilistic computation provides an opportunity to take ‘a more relaxed approach to what is correct.’

The results of processors that are ‘glitchy’ and reproduce errors might not be of the same quality as those that are the products of exact computing, free of randomness, but ‘background imperfections’ will not be important if the deviation from accuracy is of a low order.

However, let it be said, these considerations as to the definition of correctitude surely have no place in any examination of navigational contrariness when reliance is placed on probabilistic outcomes.