This is possible already with hyperdb, but because of the way hyper-readings stores information as a graph which is then rendered by hyper-reader only some versions will be open correctly. We need to establish a version tagging system so that we know which points in the history of the db represent 'saved' moments.