ABSTRACT
The designers of adaptive environments should seek to minimize the risk of over-investment in adaptive features, and also minimize the risk of under-investment in adaptive features, which may be a more common shortcoming. This requires measurement of costs and benefits – at the time of design, not 30 years later. The arguments in favour of adaptive architecture will be stronger if its adaptive performance can be measured. This is difficult because of uncertainty about future activities, but assuming that activities will change radically in unforeseeable ways can lead to excessive investment in over-complex adaptive designs. An example shows a systematic approach to forecasting a range of possible future activities and using this data to evaluate alternative adaptive strategies. The concept of adaptive architecture has an elusive charm, but it would be of far greater practical value if it could be subject to the same tests of efficiency as other design attributes.
