simguru
Member
I was reading a book the other day on the evolution of the brain - from the original reptile version, through mammal to human. Shortly after that I heard that social/pack animals have smaller brains than their asocial counterparts (e.g. lions vs tigers). My conclusion from this was that social animals like humans are predisposed to be specialists, i.e. if the individuals of a group have non-overlapping skills then the group can have broader and deeper skills than an individual.
The upshot of this is that once you start into a particular methodology folks will specialize and become experts at it. If you combine multiple methodologies together into (say) a CAD design flow then it becomes more locked in as the experts become more specialized in the different components and the degrees of freedom are reduced, i.e. if there is a particular step that requires a specialist tool/person it will tend to have fairly constrained inputs and outputs. This problem gets worse if you put smarter people on a problem, e.g. the requirement for PhD level qualification or ten years experience for a job is an indication that the way the problem is being tackled has maybe become over constrained, and with no way to move sideways (to take the analogy of simulated annealing) you need to dig deeper, which locks you in more.
To run with the simulated annealing analogy: I'm wondering how hot does it have to get to break out of the current set of local optima and into a new paradigm for design?
The upshot of this is that once you start into a particular methodology folks will specialize and become experts at it. If you combine multiple methodologies together into (say) a CAD design flow then it becomes more locked in as the experts become more specialized in the different components and the degrees of freedom are reduced, i.e. if there is a particular step that requires a specialist tool/person it will tend to have fairly constrained inputs and outputs. This problem gets worse if you put smarter people on a problem, e.g. the requirement for PhD level qualification or ten years experience for a job is an indication that the way the problem is being tackled has maybe become over constrained, and with no way to move sideways (to take the analogy of simulated annealing) you need to dig deeper, which locks you in more.
To run with the simulated annealing analogy: I'm wondering how hot does it have to get to break out of the current set of local optima and into a new paradigm for design?