The Semantic Web changes the way we deal with data, because assumptions about the nature of the data that we deal with differ substantially from the ones in established database approaches. Semantic Web data is (i) provided by different people in an ad-hoc manner, (ii) distributed, (iii) semi-structured, (iv) (more or less) typed, (v) supposed to be used serendipitously. In fact, these are highly relevant assumptions and challenges, because they are frequently encountered in all kind of data-centric challenges – also in cases where Semantic Web standards are not in use. However, they are only partially accounted for in existing programming approaches for Semantic Web data including (i) semantic search, (ii) graph programming, and (iii) traditional database programming approaches. The main hypothesis of this talk is that we have not yet developed the right kind of programming paradigms to deal with the proper nature of Semantic Web data, because none of the mentioned approaches fully considers its characteristics. Thus, I want to outline empirical investigations of Semantic Web data and recent developments towards Semantic Web programming that target the reduction of the impedance mismatch between data engineering and programming approaches.