The use of web services as the basis for grid middleware has allowed scientists to wrap legacy applications as services in order to provide their capabilities to grid users. However, this "wrapper" model can be poor at satisfying the requirement for software and data to be efficiently positioned on processing resources around the grid as part of larger coordinated computations. We argue that some aspects of the earlier more exposed grid model, where the client coordinates the use of resources, support these issues better and allow more efficient grid solutions. We propose a model to take advantage of both systems.