Most references to contingency that I can find in his blog are in relation to Meillassoux, how the latter espouses it whereas he, Harman, remains more skeptical. Meillassoux advances that only absolute contingency should be necessary. My understanding thus far is that for Harman contingency remains at the core of correlational processes. I am not entirely sure about the validity of this last sentence. I may have to come back to edit this.
Realism 