We disagree with the emphasis and the roles defined for the general internist in the college's recent position paper [1]. The proposal omits the most important role for which general internists should be trained, that of internal medicine specialists. Such physicians know the breadth and depth of internal medicine but do not have the training or competencies to perform the complicated procedures of subspecialists. Recently, there has been a drift away from these masters of the practice of internal medicine to generalists at one extreme and subspecialized subspecialists at the other. To say that "imaging techniques and growth of subspecialties has seriously eroded the generalist role as a consultant" and "family practitioners and specialists often deal directly with subspecialists rather than general internists" may be true, but accepting "what is" as a dictate for the future of general internists is dangerous. Anyone who watches today's "generalists" or subspecialists flailing away at ordering tests and internal medicine decision making because they are incapable of practicing the art and science of internal medicine as a specialty must seriously consider the value (quality per cost) of such a care delivery system in the future. Emphasis on a generalist role and a diversity of roles will not address this problem.

As in the past, the role of the general internist as a specialist should be to sit directly between "generalists" and medicine subspecialists because the knowledge base of internal medicine is essential to the diagnostic and therapeutic decision making and the trusteeship or gateownership that is implicit in this role. This role is not limited to the inpatient consultant role.

Roles of "clinical information manager" and "generalists who possesses special skills for a particular environment" are not generic to general internists, should not be defining criteria, and should be encouraged for internists. However, the proposal that general internists "could focus on one area of subspecialty expertise" does not recognize that internists in the past and those of the future can manage most cognitive issues of subspecialty care and do uncomplicated procedures. It is not the knowledge base or cognitive skill requirements that have changed so dramatically during the past 20 years—it is the technology!

Finally, one can train a primary care or generalist or a general internal medicine specialist in 3 years, but to be trained as both will require 4 years of appropriate training.