Christian, you should really be the first to vote to close this.  The title of this bug report is about whether it would be good to reduce the _number_ of bytes Random initialization consumes from os.urandom(), not whether to stop using os.urandom() entirely.

But if I grasp your position, anyone who thinks Random initialization should continue using os.urandom() at all is, ipso facto, so insufferably ignorant they shouldn't even be allowed to express an opinion ;-)

So, to me, your position on _this_ bug report is clear:  no, merely reducing the number of os.urandom() bytes consumed is of no value whatsoever, so this report should be rejected.

If you want random initialization to stop using os.urandom() entirely, that argument belongs in one of the other bug reports - trying to hijack this _different_ report just adds to the noise.

I haven't seen an actual reason advanced to believe that reducing the number of bytes would help anything either, so I'd also like to see this closed.

Note that, today, Guido suggested that Python's os.urandom() should grow a flag to specify the desired behavior (block; raise an exception; just keep going and maybe get lower-quality bytes).  In that case, the default random initialization would surely pass the "just keep going" flag.  In which case, I've still seen no reason to expect that reducing the number of bytes requested would help anything.