The very notion of productivity improvement involves measurement against a context. The success of computers and other quantitative approaches during the past half century has led to an ideational context wherein transmitters of information often assume that the content of their message is like code--ascertainable to the recipient by means of a look-up table. Complexity science suggests that contexts themselves need to be continually examined lest a boundary or frame be inappropriately assumed. When, for example, efficiency improvements are the result of externality creation, a loss of resilience is often the price and a shift in frame would reveal the "improvement" to be either an increase in risk or a resetting of a boundary. Improvements are not always "code" for "better," but may, in the light of a complex systems approach, be a cue for danger ahead.