One of greatest living scholars of quality, Professor Noriaki Kano of Tokyo University, and his American colleague, Tom Nolan, suggest that system improvements can come in three categories: 1. Reductions of defects experienced by customers; 2. Reductions of cost, while maintaining or improving experiences of customers; and 3. Creating and satisfying a new need, or satisfying an old need to an unprecedented degree.[1] Kano and Nolan suggest further that, at any specific time, improvements in one of these three categories are strategically most important. A modem computer manufacturer, for example, must be focusing on improvements of Type 3, or it will be shortly out of business. In computer industry, production costs are already low and computer purchasers simply will not tolerate noticeable defects. By contrast, most Americans, asked what improvements ought to occur in U.S. postal system, will immediately cite as key issue. Few will mention cost of stamps, and even fewer will request some new, delightful, unexpected service or product. The postal service needs Type 1 improvement most. For American medicine today, improvements that matter--the ones that contain strongest seeds of organizational survival--are Type 2 improvements: reduction in cost while maintaining or improving experiences of those who depend on these improvements today: payers, governments, patients, and families. This is one of reasons that report from SUPPORT project is so important, and, at this stage, so teasingly disappointing. Many, many opportunities exist in health care to reduce costs while maintaining or improving quality, but few are as direct and consequential as opportunity to stay our hand from delivering invasive, painful, undignified, and wasteful high-technology care to patients who neither want nor can benefit from that care. At best, such care is profligate; at worst, it is abusive. Despite opportunity, interventions of SUPPORT project came up dry. Aside from a few, marginally significant gains, patients offered voice, participation, and new communication channels experienced as much costly, painful, and unwanted care as did patients in usual system. What happened? Are potential gains from more appropriate use of technology at end of life out of our reach? Not at all. The SUPPORT study shows in this phase only that systems of care are complex, nonlinear, and difficult to manipulate in simple, cause-and-effect experimental modes. Here are some clues worth investigating further. In first place, one of crucial findings of study is that only half of doctors allegedly given information about patient preferences acknowledged that they had ever seen such information. Furthermore, only one-sixth of those who did recognize existence of SUPPORT information said that it affected their decisions regarding treatment. These results suggest that issues in communication may be crucial to success of future efforts to mold care at end of life--not efferent communication of desires from patients (which this study confirms is possible), but afferent communication of information to medical decisionmakers in receptive postures. The doctors in this study were poor receptors. How could we improve transfer and use of this crucial information? First, psychology of adult learning may help us to shape better effects next time. We know, for example, that adults learn better when they initiate questions, not when they receive answers never asked for. (When student is ready, says Zen aphorism, the teacher appears.) The SUPPORT intervention system appears to have lacked element of curiosity among doctors, who, like most adults, rejected or ignored answers to questions they never asked. Second, we know that leaders can help to create a context for learning, and yet there is little evidence of such a context in design of SUPPORT system. …