What would a symmetric prior look like? Is there a simple description, like “symmetric about the number of switches expected per N trials”?
Oh, I see, there’s no simple description because the definition of “switchy” vs “sticky” prior hypotheses are actually quite complicated—and if they were much simpler, you wouldn’t see this phenomenon at all.
What would a symmetric prior look like? Is there a simple description, like “symmetric about the number of switches expected per N trials”?
Oh, I see, there’s no simple description because the definition of “switchy” vs “sticky” prior hypotheses are actually quite complicated—and if they were much simpler, you wouldn’t see this phenomenon at all.