While there is a great cry about the importance of interoperability among disparate and distributed systems this really is no more than a stepping stone to the true objective, which is the development of permananent information infrastructures. If one things of message transport on the Internet, and by this I mean what is happening at the IP level there is no discussion of Interoperability – everything uses IP protocols. Now of course that was not always the case. The very name – "Internet" derives from "inter-networking", meaning between networks, as IP protocols were proposed to provide an interoperability bridge from one system to another. I doubt that many of the workers in that original DARPA project could forsee that this "inter-networking" protocol would replace all the protocols that it went between. In any event, the success of IP, has now given way to the message transport as infrastructure – as a part of the "background" of all of our lives – as the underlying foundation of telephone calls, chat messaging, e-mail and web pages. These components don't just "inter-operate", they form part of and interact with an integrated messaging infrastructure.
I do not mean to imply that the universal adoption of GML, WFS, WRS etc is some sort of panacea. I do mean to imply that by thinking of infrastructure we can focus on the real problem, while we deal with the issues of multiple parallel encodings and protocols. Think infrastructure not interoperability.
In a similar manner, the standards that underly the early steps of the GeoWeb (GML, WFS, WRS (CSW.ebRIM), FPS/WMS) are thought of by many in terms of interoperability – as glue between disparate systems that were designed and conceived largely in an era before widespread use of the web, before the clear separation of graphics and geography, and before the availability of rich data models such as GML and ebRIM. We think it is time to look at this in terms of building infrastructure.
Does this mean abandoning existing technology products? Does it mean that they will need to change? Certainly it helps to see developments of existing products – and gradual steps by Oracle and ESRI, while far from complete point in the right direction. More importantly, it will require the deployment of a robust infrastructure layer that provides for transactional synchronization of databases (this is the analogue of the messaging layer in "networks") through publication/subscription using open protocols (WFS/GML), and that provides for registries of the many artifacts that must be related, tracked and managed within the infrastructure. These components plug into the GIS database and Internet layers and hence build upon existing things rather than demanding replacement.
When we begin to think Infrastructure rather than Interoperability we begin to see a new layer emerging in the stack for managing geospatial information. This new layer is concerned with the management of the flow of information from producer to consumer. Looked at from this perspective we start to see the needed pieces more clearly. The need for example to manage multiple artifacts and relationships among them. Like the registry made infamous by Windows, we also need a registry in the geo-infrastructure layer. Such a registry enables artifact discovery and sharing and provides the foundations for managing information flow. We see also the need to deal with information publication and subscription. It is not enough to have agreed to interfaces – this is a necessary condition to be sure – but for infrastructure it is not in itself sufficient. We see also the need to monitor the flow of information within the infrastructure. Is information delivered in a timely manner? Have there been recent transaction failures? Focus on infrastructure also makes us realize that while user interfaces and visualization are terribly important, they are not an end in themselves. Existing GIS tools of ESRI, Intergraph, MapInfo etc. already provide powerful capabilities for visualization. Supported by the appropriate intermediate infrastrcture few if any changes to these products are needed in the near term. So thinking infrastructure gives a new viewpoint on the whole body of standards emerging from the OGC and elsewhere.