The research code has been reviewed several times, both by me as
well as third parties, one of whom was Russell Wallace. No problems
in have ever been found. (Nor have any in the other %age chance
routines where players have experienced long strings of failures).
I doubt there are problems in the code, but you never know. Are
these statistical apparitions normal? If someone gets 17 failures
in a row over N turns in a game with ~200 players, each on average
executing Y of the same command on average per turn, over the course
of ~150 turns, does that indicate there is likely a problem with
the code? Or would it be unusual if someone _hadn't_ seen a string
of 17 failures?
Beats me, the last stats class I took was 11 years ago.