In the early part of his book "Notes on the Synthesis of Form", Alexander describes the concept of "fit". He makes the point that, as humans, we find it very hard to detect "fit", but are well-adapted to detecting "lack of fit". Think of trying to decide if the edge of a piece of wood is straight: the natural thing to do is to put it against a straight surface and see if any gaps show in between. As we try to improve things by reducing the misfit factors, we soon discover that these factors tend to be interrelated: fixing one may exacerbate another, so there is a continuous process of compromise between a number of less than ideal solutions. Intuitively, the less interrelated the misfit items are, the easier it will be to reach a state of acceptable fit. This means that our only hope of reducing misfit factors is if these factors occur in clusters such that the connections between clusters are relatively loose. In application development, this works best if the infrastructure software components also define boundaries between misfit factors.
maybe that's why I'm so critical....
btw. I've started reading J. Paul Morrison's book. I like it, I recommend it.
Posts
