So I’m curious, what would be the ideal stack here? Or what would be the ideal approach? Because you were saying you’re trying the constitutional approach. Our relationship is a way to do this. It’s not bad. There is thinking about, you know, directly having graphs and graph representation speed model, much more state-based rather than state-free approaches. If I try to do this both as a fringe research and also add value to community, I feel like I would do 50-50 and neither would be the time out. I’m curious what experiments would be good here.