Wednesday, April 30, 2008

Agent theories redux

My sometime correspondent Roy Simpson asked me to comment on the following text from a European research roadmap project. 'What does it mean?' he asks.

"We need agent solutions for distributed, enterprise-wide environments with exacting development requirements. This might be achieved through transition approaches by which existing systems can be upgraded with a successively increased agent presence in a seamless fashion.

"Wrapping legacy systems within autonomous agents situated in a larger multi-agent system is one approach that is being tried, for example, in connecting new and old telecommunications switches together seamlessly, allowing legacy switches to be gradually replaced without major disruption to the overall system."

My answer

Well, where do I start. The term "agent" is not a theoretical term (where is the formal definition?). It's either informal anthropocentric chat from the less reputable side of AI research, or it's whatever the formal system under consideration wants it to mean. Or it's just marketing-speak.

It is a common practice in telecoms, as in computing, to wrap legacy systems with mediation interfaces to talk to more modern systems. The latest incarnation is to provide XML-based web services wrappers around legacy applications. We saw a similar thing with three-tier architectures where the first two tiers 'wrapped' legacy enterprise information systems for the new Internet age (and browser access).

The material you quote is basically gibberish, which could only find a home in a European funding application or as an output from sub-standard and mediocre euro-research.