question

Doubles only store up to 16 decimal digits accurately although in practice the last digit often suffers rounding inaccuracy. The local variable window is showing sixteen digits, so you're seeing the affect of the inaccuracy and some rounding on that last digit which accumulates over many calculations and starts affecting the 15th and earlier digits. Most applications don't need that level of precision on large numbers, and those that do probably use a multiple precision arithmetic library to overcome the machine level limitations.