This is a good trend. Will we see more work aimed at separating the data model and human interface (to make programming with attributed graphs safe and convenient) from back end issues (how to store data and process queries efficiently, where traditional databases are already very advanced.)