What the study showed

In 1979, eight specialists compared treatment recommendations for ten meningitis cases. They did not know whether each recommendation came from MYCIN or one of nine human prescribers.

MYCIN received an acceptability rating of 65 percent. The five faculty specialists in the comparison received ratings between 42.5 and 62.5 percent.

This was a small study of one carefully chosen task. It did not show that MYCIN was better than doctors in general.

It showed something narrower: a rulebook could produce recommendations that specialists rated as acceptable inside the task it had been built for.