Which is why ggvis has a different architecture which will make working with large datasets easier. Not to mention the pipe as a unifying interface across ggvis and dplyr which makes it easier to do efficient data manipulation within the visualisation.