We explore some philosophical and scientific underpinnings of clinical research and evidence at the patient-clinician encounter scale. Insufficient evidence and a common failure to use replicable and sound research methods limit us. Both patients and health care may be, in part, complex nonlinear chaotic systems, and predicting their outcomes is a challenge. When trustworthy (credible) evidence is lacking, making correct clinical choices is often a low-probability exercise. Thus, human (clinician) error and consequent injury to patients appear inevitable. Individual clinician decision-makers operate under the philosophical influence of Adam Smith's "invisible hand" with resulting optimism that they will eventually make the right choices and cause health benefits. The presumption of an effective "invisible hand" operating in health-care delivery has supported a model in which individual clinicians struggle to practice medicine, as they see fit based on their own intuitions and preferences (and biases) despite the obvious complexity, errors, noise, and lack of evidence pervading the system. Not surprisingly, the "invisible hand" does not appear to produce the desired community health benefits. Obtaining a benefit at the patient-clinician encounter scale requires human (clinician) behavior modification. We believe that serious rethinking and restructuring of the clinical research and care delivery systems is necessary to assure the profession and the public that we continue to do more good than harm. We need to evaluate whether, and how, detailed decision-support tools may enable reproducible clinician behavior and beneficial use of evidence.