The algorithm used in age grading is based on the open world record and an age related decay rate, presumably based on the average decay across all distances. This is apparent in the marathon records. If you take Ed Whitlock's marathon records and age grade them you get:
M 70 (73): 2:54:48 - 98.51%
M 75 (76): 3:04:54 - 97.60%
M 80 (80): 3:15:54 - 99.99%
M 85 (85): 3:56:38 - 95.32%
They all look pretty similar and the slight drop off at M 85 is probably because Ed was already feeling the effects of the cancer that was to take his life some six months later. However, Ed set the M 85 record in 2016 whereas the calculator (Howard Grubb) used factors from 2015. If we consider the world record prior to Ed's we have:
M 85 (86): 4:34:55 - 84.94%
Effectively, the Grubb calculator was saying that the M 85 record was very soft and that a talented athlete should have no difficulty breaking it (and Ed did by 40 minutes). So you could argue that the calculator is a good predictor of human potential but the real challenge, at the higher ages, is to find an individual who has natural ability, is still very healthy and is still motivated to go for it.
So for an individual I feel it is a very good way of seeing how you compare to yourself 5 or 10 years ago as it is based on a natural decal with, presumably, a large amount of data generating the decay function. However, using it for bragging rights - my 76.37 beats your 75.83 - is stupid.