You mean pull-oriented models are easier to reason about, right?