Programming semantic data on the Web is an emerging problem since the data representation follows quite flexible and extensible data models like the graph data model of RDF, where predicates denote relations between resources. While this flexible representation style is an essential feature of the Semantic Web, common procedural and object-oriented programming languages rely on a strict schema and typing system to access data. To resolve interoperability problems between both representation styles, there are several approaches that either extract schema and type information from RDF data or integrate semantic language constructs into a programming language. However, these approaches still try to "map" the flexible RDF data representation towards schema and typing systems. In this paper, we propose the other direction, i. e., to initiate a new programming paradigm that follows a rather schema-less data representation on the Semantic Web.We propose to consider properties of resources as an indicator for type and schema information.