The current SOA software stack is an answer to the lack of information about remote service
providers and lack of control over them. The future of SOA is in our view dependent on the ability to limit the complexity, either by simplifying the languages and protocols, e.g. by falling back to simple HTTP instead SOAP. Or one has to be able to recognize patterns from which the WS-* code can be
largely generated. 

We have developed a classification framework for patterns from the Ishikawa diagrams, which relate contextual properties (observed variables) with emerging properties, such as the presence of an interoperability problem. We propose that patterns should be classified by such a structured scheme rather than by simple keywords. The classification framework allows to formulate causal queries like "Which patterns are applicable when negotiating (=process) a business model (=product) between multiple business partners (=producers). 

More details are in the paper: