From the beginning, I liked the impact measure (aka: people helped), but I must to admit that my personal case is pretty much inexact. A shadow grows between the datetime of the question and my answers. My impact measure take in account views made before that I answered.

My SO profile show ~405k of impact, but my linear estimation is less of ~140k.

Here are two examples:

Question: 1401 days from Q - 115k total views

Answer: 253 days from A - 20k estimated views (linear estimated)

Difference: 95k views

Question: 523 days - 240k views

Answer: 154 days - 70k views

Difference: 170k views

Total difference: ~405k - 95k - 170k = ~140k

So late answer should not be taken into account, or we need some kind of function to adjust this measure.

What do you think?

Personally I won't lie to my grandma.

All in all, I think the current way is fine. So you don't tell the whole truth so your granny, just an estimation - that's why they added a `~` in front of the value and clearly saying "estimated number".