V1, 7/18/05 9 can be said of policy and management decisions. The work by Sandy Bernard of Extension and Education made it clear to us that significant engagement of casual information users was a much bigger task than some anticipate because it requires development of highly tailored products.  The products in turn require focus groups and testing in a range of venues.  This level of focused effort is outside the realm of the SEACOOS development effort and can be viewed as encroaching on roles more appropriate to the private sector.   For all these reasons SEACOOS has chosen in its implementation plan to focus on a small set of users who are accustomed to working with raw data, whom we termed super-users.  We therefore differ with the evaluators’ recommendation on this point.  We have, however, embarked on a “Teams” approach in which we bring together multidisciplinary and multiagency representatives to help SEACOOS approach a few more well defined societal problems.   · The SEACOOS modeling program is a composite of three different model applications, with a pair of overlapping regions for intercomparison. The more likely RCOOS scenario will include the operational application of one or more different full regional models in addition to sub regional model applications that serve special user needs. Given the visibility of the SEACOOS effort in general and its investment in numerical circulation modeling in particular, investigators need provide significant guidance in the design and implementation of a truly operational SECOORA RCOOS. With regard to concerns about model configuration and domain, we agree with the need for some SEACOOS-wide efforts.  It is not sufficient to utilize only one model, however.  We need an ensemble of models to compare, evaluate, and validate the models under a range of conditions.  It is doubtful that one model will serve all purposes, at least in the near term, or that this is a wise approach in general. Along with our subregional models, as our starting point, we need to build SEACOOS-wide models, not competing among ourselves, rather making tools available for each of us in our subregions.  Moreover, there will always be a need for the higher resolution enabled by regional and subregional models.  In the same way that we are contributing to the design of an observing system we are also contributing to the design of a modeling system.  Our starting point consisted of the three subregional models already in existence, but this was never conceived as being our end state model design.   .   · In the next few years, SEACOOS investigators need to elucidate the issues that must be resolved in the implementation of a SECOORA RCOOS, by implementing and running a prototype near real-time application of a single, full- region prognostic circulation model, with prototype open boundary conditions and some data assimilation. The need to demonstrate a region-wide prognostic capability is essential and as