The database and information retrieval communities have long been recognized as being irreconcilable. Today, however, we witness a surprising convergence of the techniques used by both communities in decentralized, large-scale environments. The newly emerging field of reputation based trust management, borrowing techniques from both communities, best demonstrates this claim. We argue that incomplete knowledge and increasing autonomy of the participating entities are the driving forces behind this convergence, pushing the adoption of probabilistic techniques typically borrowed from an information retrieval context. We argue that using a common probabilistic framework would be an important step in furthering this convergence and enabling a common treatment and analysis of distributed complex systems. We will provide a first sketch of such a framework and illustrate it with examples from our previous work on information retrieval, structured search and trust assessment.