[Next] [Previous] [Up] [Top] [Contents]

5. Pictures of Knowledge Development and Dissemination

5.4. Planning under uncertainty


Another fruitful analogy can be drawn with results from AI planning systems, that are intended to work acceptably under conditions of uncertainty. One is faced with a large space of possible plans and some anticipated constraints imposed upon one one's choice imposed by the environment. The task is to choose a good plan. The algorithm chosen depends somewhat upon the extent of the uncertainty about the constraints. In a situation where the constraints are known with complete certainty then it is most efficient to apply the most restrictive constraint first to reduce the search cost. However if there is great uncertainty then it is sensible to apply the loosest constraint first and retain the flexibility about what would be the best plan to implement until the last possible moment*1. In this way the delaying of constraints retains the maximum flexibility in order to be able to react to unpredictable changes in the environment.

With the advent of efficient search engines the search cost for readers is much reduced. Thus in fields where the journal will not be able to anticipate the selection criteria of its readers it is sensible not to prematurely select the content but to delay the selection process right up to the time the reader access the information. In this case it is better to `tag' papers with judgmental information in a form that search engines can utilise and allow readers to choose the selection criteria themselves rather than attempt to do the selection on behalf of a disparate set of readers prematurely. Previously the cost of implementing such a delayed search was prohibitive and this is why the journal system was appropriate, now with the shift in the cost structure this delayed selection is not only possible by fairly easy to implement.

Of course, if a journal is in the happy position of being able to anticipate its readers selection criteria with a high degree of faithfulness, then the (now minimal) search cost they save its readers may justify its existence. I think that for many journals this is not be the case, and that in the future the search cost saved will diminish as compared to the flexibility of review.


A Proposal for the Establishment of Review Boards - Bruce Edmonds - 16 MAR 99
[Next] [Previous] [Up] [Top] [Contents]

Generated with CERN WebMaker