Joint Cognitive Systems

Automation Surprises 133

assum ed were standard were, in fact, quite over-simplified. Their model of practice

missed many of the important constraints and demands, e.g., they assumed low

tempo , low workload, low a ttentional demand situations, and they assu med that

anoma ly response consis ted of mapping symptoms into a few d iagnos tic

classifications—starkly inaccurate, given the actua l char acter of anomaly response

as we saw in Chapter 8.

But notice how people interact during anomaly response as in the upda te case

(pp. 91-92). Exchanges are brief, coded, to the po int. They do not break from the

flow of the events and responses to engage in long-winded explanations (except

when te mpo drops low and then these exchanges take the form of tutori ...