5. Pictures of Knowledge Development and Dissemination
The archetype foundationalist subject is mathematics - there the criterion for validation of a piece of mathematics is clear: a piece of mathematics is valid and may be relied on if and only if the proof is correct, which is checkable (at least in theory). However, even in mathematics there is not clear agreement about what should be published because publication is largely dependent upon the importance of the results, and what is considered important is a subjective matter.
The evolutionary picture is one where academics are continually producing variations on older work and a selection process is acting upon these to give preference to the better ideas [3]. Those that are selected are more likely to be used and varied by future academics, so the population evolves in response to the selection pressures it is subjected to. These selection pressures are largely determined by the academics themselves (what lasts is what will appeal to academics over the years for good and bad reasons), but is ultimately grounded in the needs of the society that the academics inhabit.
In the field of evolutionary computation (where artificial evolutionary processes are designed and studied), it is clear that there is some sort of trade-off between brittleness and cost. A greater and sharper selection pressure (as is applied in the breeding of show dogs) implies a quicker and cheaper convergence to a solution, but also means that one is more likely to get stuck in a sub-optimal solution. The reason for this seems to be that in hard problem spaces a good solution is sometimes reached by a variation on a very poor solution, so if one uses a `crisp' selection criteria (i.e. only those above a certain threshold) one will select out this poor solution and never reach the good solution.
"xbc there is a necessarily a trade-off between concentrating the search in promising parts of the search space which increases the chance of finding local optima versus a wider ranging search which may therefore be unsuccessful but may find a more remote but better global optimum." [13] p.207In the field of the evolutionary population dynamics, there is a key theorem, `Fisher's fundermantal theorem' [8] (updated and clarified in [15]), broadly it states that (under quite a broad range of assumptions) the rate of increase in the fitness of a population is proportional to the variation in that population. The proliferation of review boards and the increase in the use of paper archives will promote that variation.
Paper archives do go somewhat towards increasing the variation and have the effect of softening the selection process, but the proposal I describe goes further. Of course, the selection pressures are indispensable but perhaps a more graded selection process may promote the quality of knowledge by helping (in a small way) the evolutionary process.
Generated with CERN WebMaker