not sure what you mean Eric

@eric.d.scott: Data can be consistent across any arbitrary set of graphs… but yes it’s common for applications to make assumptions that some graph sets have certain properties that might not hold in a less controlled context

I'm thinking of the case where you might want to represent the contents of different natural language discourses, say speeches by various politicians. One might say for example that the birthplace of Barack Obama was Kenya...

I personally try to avoid using the Default Graph as much as possible, and put all my data in meaningfully named graphs with suitable metadata about the source and purpose of the data.