"Better than doing it with the whole country, possibly.  But heritage optimization takes time.  Every time you try something and see what happens, you have to wait one human generation to see what happens, and if you need two generations, well, that adds up very quickly."

"Speed of discovery matters.  It's not enough to get there eventually.  How can you figure out what the ass you're doing with heritage optimization without spending twenty years every time you want to try something?  How can you learn about the sort of mistakes that only show up three generations later, in less than sixty years?"

"Better than doing it with the whole country, possibly.  But heritage optimization takes time.  Every time you try something and see what happens, you have to wait one human generation to see what happens, and if you need two generations, well, that adds up very quickly.  Gonna keep the whole country waiting while you play around in that one duchy?  Doing nothing also has risks."

"Speed of discovery matters.  It's not enough to get there eventually.  How can you figure out what the ass you're doing with heritage optimization without spending twenty years every time you want to try something?  How can you learn about the sort of mistakes that only show up three generations later, in less than sixty years?"