The time has passed for reflection about the model's choice. Has not research become so international that a pan-European Grid organisation and infrastructure, like CERN or ITER, would have been a better match for the European Grid? Probably yes, but that is not the route that has been chosen.

Other questions that can be asked are about the size of the EGI organisation. Fifty persons with a lot of different tasks is much. From a co-ordinating-only organisation, one would expect something half the size. One of the reasons for the bigger size is that also a number of services are provided by the central EGI organisation itself.

One of these services is the provision of a standard middleware distribution. Although this looks like a good idea, there are some dangers. Currently there are several types of service Grids middleware, like gLIte, Globus, KnowArc, Unicore that are widely used. Despite efforts to convert to one set of software based on standards, the reality is that this has not happened during the past decade, despite efforts like OMMI Europe. In fact there are no more players than ten years ago. And, although a level of interoperability has been achieved it is unlikely that they will converge into one distribution. On the other hand, also in e-Infrastructures, continuity of the production infrastructure has to be coupled with the possibility to quickly integrate new paradigms and new types of middleware when they come available, such as Clour computing and Desktop Grids. In any case this will be a challenge for EGI and the NGIs.