Note on Friedman's ‘fundamental theorem of biomedical informatics’

Charles Friedman recently proposed a fundamental theorem of biomedical informatics (henceforth called the fundamental theorem) which states that ‘a person working in partnership with an information resource is better than that same person unassisted’.¹ The person could be a clinician, researcher, student, patient or administrator interacting with the resource to perform some specific task at hand. Without loss of generality we can be agnostic about the various roles of the person and assume that the person is interacting with the information resource for decision making. Using the framework of decision removes from consideration pedestrian uses of a computational resource such as watching a movie or checking news. Hastie and Dawes² formulate decision making as a response to a situation consisting of three parts: (1) availability of a set of actions to perform; (2) the decision maker has some prior notion regarding possible outcomes for each action; and (3) consequences for the outcomes. An example of a decision in a healthcare setting would be whether to manage a condition such as shoulder pain conservatively or by surgery.

Cognitive scientists and researchers in artificial intelligence have used games such as tic tac toe, checkers and chess to study decision making. A chess contest described by Garry Kasparov,³ the former world chess champion, has important implications for the fundamental theorem under consideration. After his loss to the IBM chess supercomputer Deep Blue, Kasparov became interested in partnership play with computers instead of the traditional human versus machine contest in what came to be known as ‘advanced chess’. He reports that the online chess site http://Playchess.com organized an advanced chess competition in which players could participate in teams by partnering with other people and/or computers. Attracted by the prize money many grandmasters working in partnership with one or more computers joined the contest. The team of humans partnering with machines convincingly defeated even the strongest chess-playing computers such as Hydra, a chess supercomputer. Human strategic strengths combined with the tactical analytical powers of the machine became almost invincible. This was the predictable part. What was surprising was that the winner of the contest turned out to be a pair of amateur chess players who were using three computers simultaneously and not a strong grandmaster with a dedicated chess supercomputer. Kasparov makes the following observation:

‘Their skill at manipulating and “coaching” their computers to look very deeply into positions effectively counteracted the superior chess understanding of their grandmaster opponents and the greater computational power of other participants. Weak human + machine + better process was superior to a strong computer alone and, more remarkably, superior to a strong human + machine + inferior process.’

It is clear from the foregoing that the process of interaction with a computer is important for improved human decision making partnering with a computer. Recently, Hunter⁴ suggested a modification to the fundamental theorem by incorporating a scientific method to state that a person working in partnership with an information resource and using a scientific method is better than that same person unassisted. In Hunter's setting there is an implicit assumption of the user being a clinician or researcher. I argue that the enhancement is needed assuming the person in the theorem is a decision maker and suggest the following.

A person working in partnership with an information resource and using a correct process is better than that same person unassisted.