New Paradigm

[Scientific] Mythology? The problem there is that most people don't want to challenge it and one has to do so, but can be unpopular in so doing. And I don't mean in the sense of a Kuhnian revolution. Even if you want to change a minimal portion of the dogma/paradigm, people just presume it is right and don't want to consider [alternatives] even in the conventional analytic framework.... Roger Dean, vice-chancellor, University of Canberra, newspaper interview

Many people have severe difficulty in handling a new paradigm. They prefer to think "inside the box" and can become quite unsettled, even hostile, when asked to question what they are doing. Part of this is the sense of investment they have in what they know - starting again as a tyro on an equal footing with everybody else with some new paradigm does not appeal.

If the current paradigms of programming, logic programming, neural networks et al worked well and met the demands of business, a new paradigm would be superfluous junk. Unfortunately, the current paradigms perform poorly at the cognitive level, for all the investment over many years by very many clever people.

Programming itself has no hope of being successful - how do you fashion a stream of instructions when the particular conditions have not yet been encountered? Programming was first introduced on a Jacquard loom to weave predetermined patterns - no problem there. Much further than that it cannot go, because the programmer cannot predict the future, whereas something that determines phasing based on the actual situation seems simple in comparison (you could make the argument that enough IF...THENs would do, but if the structure can change itself, no amount of conditionals can compete).

The other paradigms that have been introduced over the last thirty years were still about direction - the artificial neural network (ANN), with its direct single value in each connection in a static framework, was a step backward from an analog computer, not forward.

Some new "glamour" paradigms have been introduced - intelligent agents, evolutionary programming - but these are mostly marketing - put in a word that promises much and it will take a long time for people to realise the word's meaning has been debased in the process, in the same way that Expert Systems debased the meaning of "expert". If the agent is "intelligent", what does intelligent mean exactly, when there is no consensus on what it means in humans. If the programming is "evolutionary", do I have to suffer all the failures and the near extinctions - the only time that evolution works well - or will they be magically eliminated, and if so, on what basis. Evolution operates on a structure - DNA - so it is already an oxymoron to couple it with programming - a predetermined stream of instructions.

Still, a new paradigm requires that people think about their problems and think about the limitations they currently accept on what they can do. This is hard work, but necessary if there is a culture of excellence on what is to be achieved.

The other side of the coin is that some people are quite concerned about the ability of machines to make any inroads into the cognitive arena. If people always acted wisely and made the right decisions, this might be an acceptable stance. There are some areas of decisionmaking where humans are too slow (parts of air traffic control or fighting off smart weapons), or too busy (the boring details of legal or medical cases), there are some where their egos blind them (development projects), there are some where their mental models get overlaid or out of date (cyclic variations in financial markets). In these areas, we do have to let go of our monopoly on cognitive processing.

Technical Discussion