What are we aiming for?...

What are we aiming for? Does "intensions" mean "approved roadmap"? VA or OSEHRA roadmap? Fair enough, and important administrative matters, but what satisfies also "The beginning of a Learning Health System"?. My view of ranking of CDS systems, which I would need to modify slightly to include public health and epidemiological decision support, is as follows...

(1) “Workbench”: Simple support of standard EHR portal with patient record accessing EBM/CER measures, meta-analyses, systematic reviews, rapid PubMed summary access, etc. etc. that the human end user, not the system, reads to extract information and make decisions.
(2) As above but includes own data mining and data analytics (including biostatistics), or access to them and new results from them, and can solicit latest or specifically needed analyses. Up-to-date-to-last-minute EHR can act as query. Feeds are essentially data mining of structured data and relatively structured records such as prescription transactions. Human end-user still processes text sources.
(3) Probabilistic Semantics phase. As above but includes means of combining results from structured and what is, or was, unstructured data, e.g. mediating by a kind of Semantic Web approach in an automated way.
(4) Automatic or semi-automated assembly of decision nets in response to queries to resolve specific questions generated by the end-user. Delivers conclusions as probabilities.
(5) Decision nets evolve under action of metastatements or “rules” such as syllogisms, basic and higher order logic, definitions and grammatical rules, etc.
(6) System can pass basic medical qualification exam in Basic Sciences, including pathology, Patient-Doctor, etc., excepting 8 below. Extends to cover fuller set of possible answers, more realistic than an exam.
(7) As above but exam also includes open-ended numerical (calculation) answers to EBM measure, biostatistics, and quantitative epidemiology questions.
(8) Inclusion of workflow, compliance, interpretation of laws and guidelines at Federal, State and Institutional level.
(9) Satisfies a “physician-based Turing Test” - performs as well as average physician armed with systems capable of delivering 1,2 above.
(10) Performs significantly better than average physician.
(11) Performs significantly better than average physician and referral experts.
(12) Performs significantly better than a “jury” of 12 medical experts in a reputable hospital or medical center. Unstructured sources routinely include medical images, not just text.