Well, who says that, and what do they say about it?

I've never seen reverse indexing like that in mathematics before. My first instinct would just be to count backwards, like skeeter.

My second instinct would be to interpret it like computer code **for k = n to 1 step 1**. If $k=1$, then sum the one term. if $k>1$, then no terms count and the sum is zero.