Key Rate Duration - Reason for using Par Rate curve

Dear all,

I am struggeling to understand two details regarding the key rate duration:

Effective and key rate duration reflect effects of changes of par rate curve - This has led to some confusion for me and, as reading through the forum, also to regular confusion for a lot of other candidates. For me it would seem to be more reasonable to reflect changes in spot rate. In a flat yield environment using both curves lead to the same effective duration, but for non-constant curves using par rates results in ugly things such as negative key rates.

What is the reason behind using par rates?

What curve so you use for modified duration?