Friday, 3 August 2012

Competency, Information and Entropy

One of the most pertinent questions about a person's competency profile is "how much information does it convey?".  I will argue in this post that the answer to that question is directly connected to the extent to which  others examining a competency profile are able to predict its content based on a sample. This in turn is related to the extent that employers can predict the behaviour of individuals in professional situations (based on a competency profile), and therefore form a judgement as to whether to give them a job or not.

A competency profile is a bit like a code. If a standard like the ISCO-88 classification of jobs in the EU is used, then the competency profile is a finite code. But information is conveyed through the various forms of expression using the code. It is, in a very simple way, rather like calculating the entropy of the English language, which Shannon famously did experiments on in the 1950s.

If we see a competency profile like the English language, then we might hope to identify some degree of pattern. Of course, unlike the letters of an English sentence, competencies aren't sequential... the whole meaning of a competency profile is achieved through scanning the entirety of what's there. Nevertheless, the examination of sample should give an indication of others in the portfolio. It is perhaps here that calculations of entropy and meaning bear some relation to gestalt psychology.

In "Meditations on a Hobby Horse", Ernst Gombrich asked the question as to how a child imagines the horse from the stick with a cloth head on it that they play with. What does this mean in terms of probability or entropy? The imagined horse is, in a sense, improbable - that is why Gombrich asks the question. Is it because it is improbable that it is meaningful? The more improbable a communication is, the more information can be said to have transferred. But what's in the improbability? For we cannot just look at the stick with a head, we have to look at the child. An unimaginative child may only see a stick - and in such a case, we might identify less improbability in the communication; less entropy (although we have to be careful with this word!)

The same applies to a competency profile. We see a competency of 'hang gliding' on someone looking to become a psychoanalyst. There are two pieces of information there to begin with: the competency and the context within which the individual wishes that competency to be assessed. An unimaginative response (one which carries less information) would be "you're a hang glider - you're not a psychoanalyst!". An imaginative response, which conveys more (less probable) information might be "you're adventurous!".

Equally, a person applying to be a psychoanalyst might claim a competency of "Psychoanalysis 101" - a module on a degree they studied. In both cases - in the unimaginative and the imaginative interpretation, this does not convey much information except to say "this is consistent with the application". That raises a problem which relates to the necessity for a competency profile to present "consistent" information (which in fact amounts to "redundant" information) - and which actually serves to convey very little information at all; and a competency profile which conveys rich information through presenting improbable communications.

When I dislike competency as a concept, it is because it tends towards redundancy and the transfer of little information. If I am hopeful that there might be something in it, it is because of the possibility of meaningful presentation of rich (but necessarily improbable) information.

No comments: