Computational techniques for supporting clinical decision-making have been discussed for over 50 years (Ledley and Lusted, Science, 1959) and research on clinical decision support (CDS) has been a recognisable subfield of medical informatics for almost as long. Since the Institute of Medicine report To Err is Human in 2000, there has been rapidly growing international awareness of CDS by governments because of their potential importance in improving quality and safety of patient care, and engagement by industry (e.g. suppliers of electronic medical record systems; scientific and medical publishers; specialist CDS companies). There is now wide acceptance that CDS is clinically valuable and key barriers to adoption are being taken down through government support (e.g. The Meaningful Use Programme in USA; Connecting for Health in the UK, and other national eHealth programmes).
However there are reasons to worry about how this field is evolving. In the scramble for commercial advantage there seems little time to absorb and apply the lessons from decades of basic and applied decision research, or adapt rigorous methods from computer science and software engineering for design and implementation. I will argue that many commercial CDS products are consequently simplistic, ad hoc in design (and potentially unsafe), and unaccountable to clinical scrutiny. They may also raise new barriers for translating new research findings into routine clinical practice and the growing interest of healthcare professionals may be choked off as these problems become evident.
Without a significant rethink the CDS field will not achieve its clinical potential and commercial markets will remain fragmented, small and shark-infested. A key requirement for mitigating such problems is the adoption of an open framework for formalising and sharing knowledge of best clinical practice  and delivering sophisticated services based on this knowledge.  In Clinical Decision Support: the road ahead (2005) Bob Greenes set out important ideas for institutional and intrastructure support  for developing a scalable repository of medical knowledge. OpenClinical.net provides a software platform for formalising and sharing best practice knowledge, deploying CDS services and monitoring their use and impact. Our goal is that OpenClinical should be publicly owned. However it, or something like it, will synergistically co-exist with a vigorous commercial sector by providing open access content and developing open standards (see http://www.openclinical.net).