Is Decidable Principle Type Inference a Good Idea?

I don't think higher kind of types pass the bar for this. But if there were a sufficiently big benefit, then I would be open to it. The drawbacks as I see them alongside the benefits, I don't think they justify their costs or the drawbacks. That's a controversial take in the world of like statically typed functional programming thing - but that is just why my view on that.