This is simply a rounding error for the factorial calculation (N!)
The rounding error precision for the factorial function is readily apparent after recent revisions, but has occurred in other previous revisions.
Currently, the calculation is 17 digits with an error of -1 to +1 in the 17 digit. The ncr and npr do not appear to be affected by this error.
When the product is obtained by multiplication, or exponent function(x^y) its precision is usually accurate to over 200 digits and up to 300 digits for certain bases and exponent values.
Calculation precision errors in Log, Trig, and other related functions are demonstrable. Subtle errors appear as apparent pseudo-random numbers occupying the last digits or decimal places in the resolved function outputs.
For Log and anti-Log functions, the errors have occurred for over 5 years.
This computer generated “calculator” is a brilliant peace of programming. At one time, this calculator would generate results with atomic accuracy at galactic distances (plus or minus the bugs and limitations of log and trig functions). Now, most functions (except for exponential) are rounded to 16 or 17 digits. This is more like galactic accuracy at atomic distances, but it keeps the FLOP load down when tens of thousands of users are seeking solutions.
~~D~~