General human intelligence is needed in order to process Linked Open Data (LOD). On the Semantic Web (SW), content is intended to be machine-processable as well. But the extent to which a machine is able to navigate, access, and process the SW has not been extensively researched. We present LOD Observer, a web observatory that studies the Web from a machine processor's point of view. We do this by reformulating the five star model of LOD publishing in quantifiable terms. Secondly, we built an infrastructure that allows the model's criteria to be quantified over existing datasets. Thirdly, we analyze a significant snapshot of the LOD cloud using this infrastructure and discuss the main problems a machine processor encounters.