The lecture is divided into four parts. In the first part, I offer a brief and simple introduction to four well-known senses in which different scientific fields speak of complexity, namely state complexity, Kolmogorov complexity, computational complexity, and programming complexity. I then suggest an intuitive way in which they can all be linked in a conceptual, unified view. Against this background, in the second part, I outline a new concept of complexity, which I shall call coordination complexity. This completes the unified view. I then argue, in the third part, that the semantic web helps us dealing with problems with increasingly high degree of coordination complexity, which require the mobilisation of whole systems to be tackled. In the last and concluding part, I highlight one of the consequences of the resolution of problems with high degree of coordination complexity: the predictability and manipulability of autonomous choices.