[Python-ideas] PEP 504: Using the system RNG by default

Stephen J. Turnbull stephen at xemacs.org
Wed Sep 16 03:30:21 CEST 2015

Sorry for the self-followup; premature send.

Stephen J. Turnbull writes:

 > In fact "secure" randomness is *higher*-quality for any purpose,
 > including science.

It does need to be acknowedged that scientists need replicability for
unscientific reasons: (1) some "scientists" lie (cf. the STAP cell
controversy), and (2) as a regression test for their simulation
software.  But an exact replication of an "honest" simulation is
scientifically useless!

More information about the Python-ideas mailing list